Terminology

Quantreno speaks the language of a real trading desk — the same words appear here, across the app, and in the assistant's answers. In the app, the small (i) beside a term opens the same one-line explanation you see below.

The stance ladder

A stance is the desk’s graded, dated view on a theme it covers. These are the grades it can carry, top to bottom, and what each one claims. Every grade is scored against the measurement basis the stance names — that is what makes it checkable.

Constructive: The desk's favorable view on a theme: it expects the measurement basis it published — a constituent basket, an instrument, or a market's probability — to rise over the stance's horizon. A dated view the record later scores, never an instruction to trade.
Neutral: The desk covers the theme but takes no directional view: it asserts neither a rise nor a fall over the horizon. A neutral stance stays on the tape and counts in coverage, but it is scored on neither headline figure — there is no direction for it to be right or wrong about.
Cautious: The desk's unfavorable view on a theme: it expects the measurement basis it published to fall over the stance's horizon. A dated view the record later scores, never an instruction to trade.

Stance is the desk’s word for its own view. Conviction stays yours: it is what you record on a thesis you adopt, and the desk never sets it.

Stance vocabulary

Stance: The desk's dated, graded view on a theme it covers — research with a view you can check, never an instruction to trade.; A stance is what the desk actually believes about a theme, written down with a date on it: a grade, what the view is measured against, and by when it should have played out. It rides the brief that publishes it, so the brief history doubles as the record of what the desk has believed over time. It is the same for every reader, it is never about your situation, and nothing trades because of it.
Stance ladder: The desk's opinion scale — constructive, neutral, or cautious on a theme, each meaning one thing.; The fixed set of grades a stance can carry, each with a published meaning. Publishing the scale is the point: a grade whose definition you cannot look up is a mood, not a view. The grade's direction is also what the record scores the call against later.
Stance record: How the desk's published calls actually did — hit rate and average return together, over the calls that closed.; Every stance the desk publishes is scored the same declared way, whether or not anyone traded it. The two headline figures always appear together over one shared count of closed calls that asserted a direction — a hit rate on its own would flatter three small wins and one large loss. Below the stated minimum number of closed calls, the record says so instead of quoting a rate.
Focus list: The desk's ranked shortlist — the covered themes it's most confident in right now, with changes shown.; A ranked subset of the themes the desk covers, not a re-sort of the theme index: it answers what the desk most believes, while the gallery keeps answering what you can run right now. Every addition and removal publishes a dated note saying why, so the list cannot quietly change shape.
Coverage: The set of themes the desk keeps a current view on — narrower than every theme the desk lists.; The desk does not hold a live opinion on every theme it lists. Coverage is the maintained set — a live theme with a live way to trade it — and a theme outside it says so plainly, with the date the desk last looked. Entering and leaving coverage each publish a dated note.

The ladder, explained

Seven short reads, top-down: the policy, the strategies under it, and how a thesis becomes a holding.

The mandate

The policy governing your capital — written down, visible, and enforced.

Themes

The story itself — style-free, researched on one clock, above every method.

Styles

The method behind a program — one style catalog, six families.

Programs

A theme's runnable implementation in one style — the preset a book is minted from.

Books

A slice of your portfolio running one method, drawing on the mandate's pool.

The thesis

The written reason behind a discretionary trade — recorded before the trade, judged after.

From recommendation to position

The four rungs from recommendation to holding.

The portfolio

Everything you hold, viewed as one.

Venues & instruments

Where you trade, when it trades, and what you trade.

Venue: Where your orders go — a connected venue: an exchange you join directly, or a brokerage carrying your account.; The desk's word for a connected market-access endpoint. The relationships differ, and the desk says so: Kalshi is a CFTC-regulated exchange traded by direct membership; Alpaca is a brokerage that carries your account and routes your orders onward to market centers. For a brokerage account, the executing marketplace lies one hop beyond the connection.
Product line: One account a firm carries under one sign-in: spot, futures or stocks at Coinbase, each its own kind of account.; A firm that carries several accounts under one sign-in carries each as its own product line: its own kind of account, at its own regulated firm, with its own cash and positions. The desk states each line's state on the venue's card, and reads a line only once the phase that serves it has landed; until then the card says the desk does not read the line yet.
Not opened: A product line you have not opened at the venue — stated as such, never shown as an empty account.; The venue answers that a line is not opened, and the desk records that state and never an empty account: an account that holds nothing is a different fact from a line that does not exist. The line opens at the venue, and the desk's next read of the grant states it.
Perpetual future: A futures contract held to its index by a funding payment, on margin. A contract with a listed expiry states it.; A futures contract held to its index by a funding payment, on margin: the holder posts a requirement and not the contract's value, and a periodic funding payment between the longs and the shorts holds the contract's price near its underlying's. A contract with a listed expiry states it. The desk places it at Kalshi and at Coinbase; at Kraken the user places it by hand, and the desk reads the fills. A Coinbase perpetual-style contract states its expiry in its name, and the accounts page states that date, with the class from the definition's own fields, never from its symbol.
Index future: A Coinbase futures contract on a US equity index, on the index schedule: Sunday evening to Friday afternoon.; One of Coinbase Derivatives' contracts on a US equity index. Its definition states the class — a non-crypto contract of the stocks asset type — and the desk reads the class from those fields, never from the symbol. It trades the index schedule group's hours: Sunday 19:00 to Friday 16:00 Chicago with no stop each day, and no trading on a business day the exchange lists as a holiday. The accounts page states index future on the position's row, beside the perpetual futures.
Digital token: A crypto asset's identity: its network with the contract's address, or the network alone for the native coin.; The desk identifies a crypto asset by a digital token: the network with the token's contract address in the network's normal form, or the network alone for the coin that pays the network's fees. One asset can be several tokens — Ether on Ethereum is its home, and Ether on Base is a second token of the same asset — and never two assets. A venue's own code for the asset is never the identity: the desk resolves Coinbase's currency to its home token from Coinbase's public currency reference, and where that is not served the contract stands with no underlying link, by name, until a later pass sets it.
Trading pair: One market at a venue: a base asset priced in a quote asset, as Ether in dollars. The holding is the asset.; One market at a venue: a base asset priced in a quote asset, as Ether in dollars. An order and a fill are on a pair, and the holding is the asset: a buy under ETH-USD and a sell under ETH-USDC move one holding of Ether. A pair the venue has restricted — limit orders alone, cancels alone, or delisted — says so on its holding.
Quote asset: What a fill is priced and paid in: the dollar, or a dollar stablecoin.; What a fill is priced and paid in: the dollar, or a dollar stablecoin. Each fill states the quote asset it settled in. A fill that settled in neither — one under ETH-BTC — is shown as the venue states it and named as outside the desk's record.
Dollar stablecoin: A crypto asset designed to hold one dollar's value — valued in dollars, with its issuer's risk, never counted as cash.; A crypto asset designed to hold the value of one dollar: a holding with its issuer's risk. The desk values it in dollars at the venue's stated value and never counts it as cash — the dollar balance and the stablecoin balance are two lines of the spot account. Where the venue counts it toward the futures buying power, that figure is shown as the venue's.
Contract hours: A contract's hours as the desk derives them, checked against the exchange's own record of the contract at each read.; The hours on a Coinbase futures position's row come from the desk's schedule for the contract's group — the exchange's list of dates, its expected dates and its weekly hours — and each read checks them against Coinbase's own record of the contract: the session's state, its open and its close. Where the record and the schedule agree, the row reads open, or closed with the next open; where they differ, the row reads unavailable as of the read's instant and claims no reopening, until an entry to the record resolves it. An early close whose hour the exchange has not published is stated as such, and the record's close does not refute it.
Venue session: When a venue itself trades — its own sessions, holidays, early closes, and maintenance windows.; When a venue itself trades — its own sessions, holidays, early closes, and maintenance windows. The desk's clocks follow the venue's calendar, never one market's: prediction markets run through every ET day, weekends included; US equities keep the exchange's trading days, half-days honest. Every desk claim about hours names whose calendar it follows.
Venue state: What the venue line claims — open, closed (reason, next open), unavailable (as-of, no reopen), or unobserved (schedule).; What the venue line claims, one pill per venue in your working set: open — the venue's current session in its own words; closed (scheduled) — the reason and the next open stated; unavailable (observed) — an outage or halt, as-of stamped, claiming no reopen; or unobserved — the venue serves no status read, so the pill stands on its schedule alone, and it claims no halt and denies none. Observation overrides the schedule: a venue scheduled open but observed unreachable states unavailable, and only the schedule ever claims a next open.
Trade date: The day a venue books an execution to.; The day a venue books an execution to. An overnight-session order executes on the next trade date; the record states it on the entry when it differs from the act's instant on the venue's own clock. A cross-venue day is labeled or absent — never silently assumed.
Your clock: Every act records at a universal instant and every stamp shows in your own zone; venue times state their zone alongside.; The desk records every act at a universal instant and shows every stamp in your own zone — whoever sits down, wherever, reads the same record translated; nothing is stored per zone, and nothing re-files when you travel. A venue's own times state the venue's zone with your clock alongside (“overnight opens Sun 8:00 PM ET — 5:00 PM your time”).
Instrument: The thing you trade — a prediction-market contract, a stock, or an option.; The specific security or contract traded, such as one share class or one prediction-market strike. An expression kind names its trading category.
Expression kind: The supported trading category of an expression, such as prediction market, equity, options or crypto.; Names the expression category. Instrument identifies the specific tradable object; the category alone is not a security identity.
Market: A single tradable contract or symbol and its live price.; The app's narrow sense: one tradable contract or symbol and its live prices — a Kalshi ticker or an equity symbol. (In wider finance usage a market is the venue or order book those instruments trade in; here it names the instrument's own listing.)

Policy & strategy

The policy that governs your capital, and what runs under it.

Desk: Your home surface — the queue first, then what moved, what's due, your portfolio, and the morning's research.; The home surface (/desk), composed fresh every trading morning before the open: the queue first — the lines that wait for your confirm, by risk, time left and arrival, with the orders at the venue below them — then the overnight recap, the day-ahead calendar on your clock, your own portfolio beside it, and the current brief on every covered theme, movers first; each zone quiet when nothing needs you; one capture stamp. Signed-out readers see the same shape without the portfolio. The standing state — mandate headroom, venue accounts, each book's open trades, latest fills — lives on the Portfolio page (/portfolio).
MandatePolicy tier · yours: The policy governing your capital — what's committed, risk limits, benchmark. Paper and Live each hold their own.; The client-policy layer above your books: committed capital, the enforced risk budget (exposure cap, daily loss limit, drawdown pause, same-instrument concentration warning), and the default benchmark. Paper and Live each hold their own mandate. Edited only on the Mandate page — the AI reads it but can never change it. Read the explainer →
Benchmark: What performance is compared against — the mandate sets the default; a book can override it.; The comparison index for performance. The mandate carries the policy default (SPY unless you change it — a broad-market yardstick that fits directional equity books better than market-neutral or event ones); an individual book's configuration can override it. Read the explainer →
ThemeProduct tier · the desk's: One story about how the world is moving — the idea itself, above any method. The desk researches it and briefs it.; The style-free narrative at the top of the ladder: the story in plain words, its subject (who's in the story), and the analyst's current brief — one research clock per theme. A theme holds no positions and can't spend anything; its programs beneath are the runnable ways to play it. Read the explainer →
StyleStyle tier · the desk's: The trading approach a program follows — e.g. market-neutral or tail risk.; The trading approach a program (and the books minted from it) follows — one of six shipped today: Event-Driven, Relative Value, Momentum, Mean Reversion, Macro Thematic, Tail Risk. The style catalog groups the recognized styles into six families. Read the explainer →
Catalog programProduct tier · the desk's: A theme's runnable implementation in one style, which a book or a program copies. A preset, not personal advice.; One way the desk runs a theme: a single style, its exact rules per market, and the sizes it's built for. A theme can carry several catalog programs — same story, different machinery — and each keeps its own "Can run in" coverage and availability chip. Opening a book from a catalog program creates your own book. A catalog program of the event-driven style opens no book: Add a program… puts your own program in your forecast book. Read the explainer →
VersionProduct tier · the desk's: The exact parameter set a program was published with — your book keeps its minted one until you accept an update.; Every publish that changes a program's methodology freezes a numbered version — the exact parameter set as of that moment, readable backstage. A book pins the version it was minted from and shows it in its provenance line ("v3"); the desk moving the program forward changes nothing on any minted book. Books minted before versioning existed read "minted before versions" — the app never guesses a pin it cannot prove.
BookFunded tier · yours: A slice of your portfolio running one style — one thesis, authored limits, drawing on the mandate's pool.; A standing strategy running one style: one thesis, authored limits (per-trade, family, an optional deployment cap), and a risk target — drawing on your mandate's pool. The main thing you create and track — usually minted as your own copy of a catalog program. An event program mints no book: it joins your forecast book as a program. Read the explainer →
Forecast bookFunded tier · yours: Your one book of event forecasts: the event programs you add sit inside it, and it holds their positions.; Each mandate has one forecast book, created with the mandate. Every event program you add sits inside it, and no event program opens a book of its own. The book holds the positions, the orders, the trades and the record; each program holds its settings, its theses and its research. One position stands for each contract and side, and each lot names the program and the method version behind it. One solve sizes every program's lines together, and the book's page states the forecast strategy's version and its release date. With no program yet, the book trades nothing. Read the explainer →
ProgramFunded tier · yours: An event program you added to the forecast book: its settings, theses and research. The book holds its positions.; Your own instance of a catalog program of the event-driven style, added to the forecast book with Add a program…. It holds its settings, its caps, its theses, its research and its standing, and no capital or position of its own. Pause stops its new lines and cancels no order; its lots stay in the book and count against every cap. Close its lots… sells that program's lots alone, and another program's lots on the same contract stay. Remove from the forecast book… waits until no lot, working order or order in unknown venue state names the program; a removed program does not come back, and its records stay. Two programs can come from one catalog program. Read the explainer →
Kind of edgeFunded tier · yours: Where a book's return comes from: a forecast, a risk premium, an arbitrage or insurance.; Where a book's return comes from. A forecast book prices a view of an event. A risk-premium book earns a premium for a risk that its basket holds. An arbitrage book locks a spread that its scan finds. An insurance book buys protection. A book runs one kind of edge, set when the book is created: a catalog program that runs two kinds opens two books, and an event program joins the forecast book as a program.
ConstructionFunded tier · yours: Who picks a book's holdings: the model (systematic) or you (discretionary). Fixed when the book is created.; Who picks a book's holdings. A systematic book follows the model's ranking, and an allocation sets its capital. A discretionary book has no model that ranks its holdings: you confirm each one under its thesis, such as the protection an insurance book buys. A book keeps its construction as it was created.
BriefBelief tier · the desk's: What the desk is watching in a theme right now — research, not a recommendation; nothing trades without your confirm.; The analyst's published research on one theme: the current view — catalysts, live markets in scope, why now — always stamped as-of, verified on the desk's 15-minute clock, and marked Stale past five trading sessions or when the market moves off its basis. Desk-level and identical for every user; it becomes actionable only when you mint a book and confirm its recommendations. Read the explainer →
VerificationBelief tier · the desk's: Every standing view is re-checked against the market every 15 minutes while its venues trade — never a silent pass.; The desk's currency contract on its own research: each published view freezes the market basis it priced, and a sweep re-verifies that basis every 15 minutes, any hour the view's venues trade — weekends included where the venue trades. A quiet market re-dates the verified stamp without a research run; a market past the view's stated rule flags Stale on every surface at once, with the cause ('repriced since …', 'at least 5 trading sessions old'); a source that does not answer reads Unknown, stated. Verified-at and re-researched-at are separate stamps — verification is a check against the standing view, never new research.
Age limit: A research view 5 or more trading sessions old reads Stale, even when its prices still verify.; The age limit of a research view: 5 trading sessions on the fastest calendar of the venues it trades, counted from the market basis it priced. A day the venue is closed ages nothing; a venue that trades every day, as Kalshi does, ages the view every day. At 5 sessions old, the view reads Stale with the cause 'at least 5 trading sessions old', and the desk researches it again.
Analyst: The desk's research seat — it studies themes and publishes briefs; it never trades.; The desk's research seat: it refreshes each curated theme's brief around the clock, whenever that brief goes stale — research and market context, stamped as-of. It reads nothing about you, holds no client positions, and never places or sizes a trade. Read the explainer →
Overnight recap: What moved since the prior close, with the desk's so-what — a stamped quiet note when nothing did.; The desk landing's recap zone: 3–5 authored lines on what moved since the prior regular close, each grounded in the desk's own measurements, with why it matters this morning. A genuinely quiet overnight publishes a stamped quiet recap — silence you can tell apart from staleness.
Day-ahead calendar: The week's scheduled prints and events, shown on your clock — plus your own thesis horizons.; The desk landing's calendar zone: the coming week's scheduled releases and events across covered themes. A release with a stated clock time renders in your timezone; a date-only event renders as its calendar day. Signed in, your own thesis horizons appear in the same list.
The queue: The desk's work across your books: the lines that wait for your confirm, and the orders at the venue.; The desk landing's lead zone: the lines that wait for your confirm across your books, kept current through the session — re-derived at every check while your venues trade, republished on material change — from recorded facts and frozen figures. The queue orders the lines when you read it: the lines that reduce risk first, then the work due today — a line due now, then the least time left — then each line by its arrival. No record ranks one book's line against another's: each book keeps its own trade list, and each line names its book and its place in that list. The capital decisions the desk owes show on one line above the queue, and the orders at the venue show under Working with their fills. Every figure on a line is the stored dossier's own value, and the queue only recommends: acting still runs through the same review and close paths.
Line: One trade that waits for your confirm in the queue — its book, its place in that book's list, its due time.; One trade that waits for your confirm: a published row of the queue. A line names its book; its place in that book's trade list, where the book's rule sets one; its due time — a rebalance at the end of its book's window, an event line at its market's first close, a carry entry at once; and its arrival, the instant from which the same trade has stood in the queue; a new size keeps it. A line stands only while what it cites still holds: a newer answer, a paused program or allocation, or a closed ticket takes it out of the queue.
Trade list: One book's own list of trades, in the order of each line's place. The book's page shows it.; A book's own list of trades at one generation: the lanes of its current answer, and the lines that stand on their own cause — a crossed exit level, a lapsed thesis, a closing sale. The book's page shows it in the order of each line's place, and a line with no place follows. Review opens each line from the queue.
Internal cross: Your lines that buy and sell one contract, matched inside your account at the mark; only the rest goes to the venue.; Where lines in the queue buy and sell one Kalshi contract in one account, the queue shows them as one netting row. Its one confirm decides every line together: the matched quantity crosses inside at the mark — the exact midpoint of the venue's quote, inside every line's limit — and only the net goes to the venue as one order. The cross pays no fee and crosses no spread, and each book's or program's lots and cash move by its own line, while the account's cash and position do not move. A side that fails its own check refuses the whole row, and each line stays confirmable alone. A line that meets another book's or program's order that already works at the venue crosses it at that order's own limit, where the order rests at the best price on its side: the desk takes the crossed part off that order at the venue first, then books the cross. Where the order is not at the best price, the confirm refuses and names it.
Close ticket: The record of a closing sale that no book's answer holds — its terms wait for your confirm in the queue.; Some closing sales are no lane of a book's answer: Close its lots… on a program, the closing sale that a pause freezes, and the rest of a sale that an internal cross took only a part of after a stop. A close ticket holds such a sale's legs and terms, with no price: the review prices them when you open it, and a card that lapses is priced again from the same terms. Its line stands in the queue while the ticket is open. Your confirm turns it confirmed, your dismissal dismissed, and its market's resolution resolved; a program's resume ends the tickets of its pause.
Owed decision: A capital decision the desk owes about a book's allocation — counted on one line above the queue.; A capital decision the desk owes about a book's allocation: to open it, to review it, or to act on a stop level or a withdrawn approval. The queue counts them on one line, with a link to the Mandate page, where you decide each one. An owed decision is never a line of the queue: it decides capital, not a trade.
Expected edge: What a priced entry is expected to clear after costs at the staged size — always shown with its sample size (n).; What a priced entry in the queue is expected to clear: the desk's stated probability against the venue's price, net of the frozen fees, at the staged size. It always renders with the record's sample size (n) so the figure carries its own evidence. The figure freezes with the publication — the queue renders it verbatim and never re-prices.
Held out: A candidate whose record is under its record minimum, such as 5 settled questions. You can still review and place it.; The record minimum's mechanism in the queue: a candidate whose track record is under its record minimum — fewer than 5 settled questions, 5 scored windows or 5 resolved funding days — is held out instead of being silently dropped or sized on thin evidence. It still reads as a line you can review and place, marked Held out, with the reason and its n stated on the line — so you decide knowing you act ahead of the record, not on it.
Record minimum: The count a track record needs before its figures count, such as 5 settled questions.; The number of resolved results a method's or a carry pair's track record needs before its figures govern a size: 5 settled questions for an event method, 5 scored windows for an equity method, and 5 resolved funding days for a carry pair. Under the record minimum, a candidate is held out on its count and the record says “not enough resolved observations yet”. At the record minimum and past it, the record's own figures govern.
Minimum margin cover: A carry instruction keeps each venue's funds for margin at least 1.5 times its maintenance margin.; The margin cover is a venue's funds for margin divided by its maintenance margin requirement. A carry instruction's minimum margin cover is 1.5 times. Under it, the confirm refuses the size and the desk holds the acts that open; at or over it, they work. Coinbase states its own margin ratio the other way up, the requirement over the funds, with liquidation at 100 %, so a cover of 1.5 times reads as 66.7 % on Coinbase's screen.
Retired (line): This standing decision ended — closed out, lapsed, archived, or no longer expressible. Nothing places from it.; A standing decision's recorded death in the queue: the cited thesis was closed out, its horizon lapsed, the book was archived, or the view is no longer expressible. The line leaves the current queue at the death — it renders on the Re-underwriting tab with its cause and stamp while the desk's next check re-underwrites the seat, and the record keeps the decision permanently. Retired is a fact about the DECISION, not the evidence: a Stale or Unknown backing view is the view grain's own word, shown beside it, never merged with it.
The desk's automation: Under your mandate it maintains theses, stages answers, and restates your queue, unattended. It never places an order.; The unattended chain behind the desk landing: whenever a market the seat expresses into is open — each venue's own calendar, weekends included where a venue trades — the desk refreshes its briefs, restates its stances, maintains the theses they support — adopting the one its graded stances best support onto your minted books, and retiring, as a dated close-out, one whose stance no longer holds and nothing expresses — stages its answers within your mandate's limits, and republishes the queue only when its content actually changes. Every act it takes is stamped and actor-lined on the surface that shows it: “by the desk's automation under your mandate.” The boundary is fixed: it never places an order — review opens a recommendation at current figures with a stated window, and nothing reaches any venue without your confirm on the review sheet. A stop control halts it per environment, and the published queue stays as it was.
ThesisBelief tier · yours: Why you hold a trade in a discretionary book or a program: your reasoning, and the conditions that would kill it.; Your recorded rationale on a discretionary book or a program: the statement, the catalyst, the kill criteria that would invalidate it, your conviction, and an optional time horizon. Each open in a discretionary book needs a current thesis, and so does each program's line in the forecast book. A systematic book that trades under an allocation needs none: the allocation governs it, and the theses it held before stay on the record, read only. Adopt one from a desk brief, accept the AI's draft, or write your own — the recommendations a book opens carry the thesis they express. The AI drafts; only you revise (the old version keeps its trades) or grade one. Adopting and closing out are yours too — or the desk's automation's under your mandate, always actor-lined, and it only ever closes out a thesis nothing expresses. A thesis past its horizon lapses on its own — it moves to the record and your positions stay put. Read the explainer →
Stance trackingBelief tier · yours: While on, the desk's automation keeps the book's standing view pinned to the desk's stance. You place every order.; The consent dial for delegated re-underwriting (ADR-0099), layered: the mandate carries each environment's default, born off, and any book can set its own value either way — the book page states the governing value and its source. While effectively on, the desk's automation revises that book's standing view to the desk's words as new briefs land, adopts the stance where no view stands, and retires, as a dated close-out, an older vintage nothing expresses — every act stamped and actor-lined. Turning it on takes an explicit confirm that states the reach and the terms. Your own pen turns it off: a Revise or an own-words adopt sets that book's tracking off in the same act, and the mandate dial is untouched. The order boundary never moves: nothing reaches a venue without your confirm on the review sheet.
Horizon: When the idea should have played out — the thesis's clock; past it, the thesis lapses on its own.; The thesis's time stop: the date by which its view should have paid. Every trade expressing the thesis shares this one clock. At review the desk suggests one from its own records — the journaled catalyst clock for this trade (an earnings date, an authored event date) or the book's preset time stop — always naming the source, always yours to change or clear. When it passes, the thesis lapses on its own: it moves to the record, your positions are untouched, and the desk alerts you only if open risk still rides on it.
Lapsed: A thesis whose stated horizon has passed — it moved to the record by itself; your positions did not change.; A thesis whose stated horizon has passed. It moves to the record by itself — no ceremony, nothing to dismiss — and your positions do not change. If open positions still express it, the desk raises its one thesis alert: revise the thesis to re-underwrite it, or close the positions through the ordinary reviewed flow. (Your subscription plan can also lapse — that's billing's word for the same idea in different chrome; this entry is about theses.) Read the explainer →
Grade: Your after-the-fact verdict on an ended thesis — realized, invalidated, or timed out — optional, dated, never required.; Your after-the-fact verdict on an ended thesis — realized (it played out), invalidated (the kill criteria hit), or timed out (it never converged) — with an optional note. Optional, dated, and revisable, whenever or never: nothing waits on a grade, and an ungraded thesis renders honestly. It's your own annotation on the record, never the desk's judgment. Read the explainer →
Run: One press of a book's method — it ends in a staged answer, an honest 'nothing qualified,' or a named failure.; One execution of a book’s method against current markets, recorded from start to finish. Every run ends visibly: a staged answer ready to review, an honest "nothing qualified" naming the rule that bound, or a named failure — never silence. Start one from the book page’s Run control or by asking in chat; nothing a run stages places without your review.
Expression: The market a run chooses when it carries a program into an act — recorded where it's real: legs, orders, positions.; Desk vocabulary, mostly backstage: the market/instrument choice a run (or the agent) makes when pressing a program — never pre-bound on the program or the book. In the app it appears as derived facts: "Can run in" coverage chips, "expressed in" on a book, and run outcomes naming what they scanned. Read the explainer →
Market scope: Which slice of the market a scanning book looks at — the venue's own categories and tags, never words in a title.; A scanning book's declared market scope: the venue's own classification — categories, series tags, explicitly named series, and exclusions — either a named list of categories or everything except a declared few. Every run screens the whole scope and says so, stamped with the market snapshot it read ("as of the ⟨time⟩ sync"). Nothing is filtered by words in a market's title. The scope is set when the book is created, and you change it any time by asking in the book's chat.
Category: The venue's own grouping of markets — Economics, Politics, Companies, and so on.; The venue's top-level classification of its markets. A scope names categories to include, or declares the few it excludes and takes everything else. The venue's grouping is the truth the desk reads — it is never guessed from wording.
Series tag: The venue's own label on a market series — scopes narrow by these (e.g. Inflation, Fed), never by title words.; A label the venue puts on a series inside a category — Economics carries tags like Inflation, Fed, Jobs & Economy. A scope's tags narrow its categories to the series that carry them; a series the desk wants that the tags miss stays in by NAME, listed explicitly in the scope.
Close horizon: How far out a scan looks — one style caps it (trade near catalysts), another sets a minimum (time for a long shot).; The close-time bound beside a scope, set by the style's method: an event-driven deployment caps how far out markets may close (catalyst trades live near their dates — 60 days unless you set it), while a tail-risk book sets a minimum time to close (a long shot needs time left to pay). One bound each, never both.
RecommendationTrade tier · yours: The desk's presented, priced ask — a basket or a single order — for you to confirm.; The desk presents a recommendation when you review a ranked move or a run's research stages one: figures computed at that moment, stated with their as-of time and a minutes-grade window. Confirm inside the window, pass on it, or let it expire — every outcome lands on the record. Past the window a confirm re-opens fresh figures; nothing places without your confirm. Read the explainer →
Rebalance: The scheduled re-forming of a momentum book's basket — entries and exits stated together for your review.; On a momentum book's clock — counted in trading sessions, not calendar days — the coded signal re-ranks the theme's names and the desk compares the standing positions to the fresh target. Every difference lands in one recommendation: entries, size changes, and exits together, each stating its cause. One confirm authorizes the whole list; nothing trades unattended, and an off-clock refresh never moves the clock.
AnswerTrade tier · the desk's: The desk's recorded answer on an actionable view each cycle — staged, primed, or declined with the reason.; Each decision cycle, every active book of a convertible style whose theme's view is constructive gets exactly one recorded answer: staged — order terms ready to review, citing the view they came from; primed — a standing entry at a stated condition; or declined — the reason and what re-checks it. A staged answer stores terms, not prices: review opens a recommendation at current figures with a stated window. In the forecast book each program's answer states its standing, and the book's own answer holds the staged lines, one for each market. The note's action zone renders them book by book, each with its as-of, and an actionable view holding no current answer shows a warning row — never silence, never reassurance. Answers state the desk's own verdicts; nothing places without your confirm.
PrimedTrade tier · yours: A standing entry at a stated condition — nothing rests at the venue; the cross stages the entry for your review.; The desk's plan for an entry it judges right at a level rather than at the market: the condition and plan stand on the book page and in the action zone, nothing rests at the venue, and when the venue trades through the level the desk stages the entry for ordinary review — review-gated like every other. A regrade off constructive or a view expiry cancels the standing entry in the same motion, journaled and visible on the book page. Deliberately not called "armed": the shipped Armed exit executes without a further approval — the opposite promise.
Nearest miss: How close a scan's best candidate came to its bar — the fact a declined answer re-checks.; When a scan finds no qualifying candidate on a constructive theme, its decline records how close the best candidate came to the rule's own bar — say a spread at 0.8σ of a 1.5σ bar — and what would re-check it. The scan's bars, caps, and mathematics are unchanged; only its silence is.
PoolFunded tier · yours: Your mandate's committed capital. You split it into your books' amounts, inside the cash limit; the rest is the reserve.; Declared once on the mandate (seeded at signup) and edited there. Each book's allocation takes its amount from the pool, and your books' amounts together stay inside the cash limit. The pool outside the limit is the reserve: no book uses it, and it is cash for margin calls. The desk sizes each book inside its allocation, and at placement the mandate's limits bound every order. Read the explainer →
AllocationFunded tier · yours: A book's capital: a dated amount and gross limit that you confirm. The desk sizes the book's positions inside it.; It also carries the book's review mark and stop level, its expected ranges and its review date. An insurance book's allocation is a premium budget for a year instead. It licenses a systematic book, and a discretionary book holds one too. The desk offers an amount from your mandate's policy, and shows the method's record, pooled across users' books. On Live you confirm each amount and gross limit; on Paper the desk grants it. Your books' amounts together stay inside the mandate's cash limit. Every act that changes an allocation is dated and stays on the record.
Seed: The first, small allocation of a new book: the mandate's seed share of the pool, 5% by default.; You set the seed share on the Mandate page. On Live the desk offers the seed for your confirm; on Paper it grants it. Each review can step the amount up while the book's results stay inside its expected ranges. A book that already trades when it takes its first allocation starts at the capital it commits, and at least at the seed, bounded by its capacity.
Paper allocation: The allocation a paper book receives at the desk's next check after a method it runs is promoted.; It takes no act from you. A paper book created while its method is live receives one at the next check too. A new book receives the seed, or the largest amount under it that fits the paper cash limit and gross limit. The paper books that already trade receive theirs together: each its offer, scaled by one factor where the offers pass a limit. An insurance book receives its premium budget. Each grant is dated and stays on the record as the desk's act.
Deployment cap: An optional ceiling you author on what one book may deploy of the pool — absent, the pool is the bound.; The authored maximum a book may commit of the mandate pool, enforced at placement with the figures when it binds. Optional by design: most books run uncapped and draw on the shared pool under the per-trade and family caps — you cap a book you want walled. In the forecast book each program holds its own, and each line meets its own program's cap. Closes and cancels are never blocked. Read the explainer →
Program capFunded tier · yours: The most one program's lines may hold, as a share of the forecast book's amount once the book holds an allocation.; The most that one program's lines may hold in the forecast book, as a share of the book's amount, so its dollars move with the amount. It counts the program's own lots at cost and its own working orders, never another program's. Until the forecast book holds its first allocation, a program's cap is its deployment cap, a share of the pool, and the screen says Deployment cap. Placement holds each line to its own program's cap, and a refusal names the program.
Cash limit: The limit on your books' amounts together, as a share of the pool: 60% by default.; Set on the mandate page. Each book's amount counts under it, and a book with no allocation counts what its positions and working orders use. An insurance book counts its premium budget left for the year with what its open contracts cost. An act that raises the sum past the limit is refused, and a limit set below the sum is refused with the least share it can take. The pool outside the limit is the reserve: cash for margin calls.
Gross limit: The most full value a book may hold; the mandate's gross limit holds your books' gross limits together.; A book's gross limit counts every position at its full value, margined legs included, and it equals the amount for a book that pays in full. The mandate's gross limit is 100% of the pool by default, and you may set it up to 400%. Both legs of a hedged pair count in full. An act that raises the books' gross limits past it is refused, and a limit set below them is refused with the least share it can take.
Stop level: The loss at which a book's new opens pause at once. Closes and exits still run.; For a risk-premium or arbitrage book, and a discretionary book that buys no protection, it is the mandate's one share of the amount, stated in dollars, read from the book's peak since its amount was last set or since its last resume. When the book reaches it, the desk asks for your decision: Reduce, End or Resume. An insurance book has none: its claim bounds it.
Review mark: The loss at which the desk reviews a book, before its stop level.; The forecast book's mark comes from its sizing: the loss that its stakes reach with a 1-in-10 chance over its review period. No other book's design states a drawdown range yet, so its mark sits at its stop level, and the stop comes first. The mark reads the drawdown inside the current review period, from a peak that starts at the book's value when the period begins. When the book passes it, the desk states its finding and its offer for your decision, and the book keeps trading. An insurance book has none.
Premium budget: What an insurance book may spend on protection in a year: 3% of the pool by default.; An insurance book holds a premium budget in the place of an amount, a gross limit, a review mark and a stop level. Its claim is its budget left for the year with what its open contracts cost, and the claim counts under the cash limit and the gross limit. A payout goes back to the pool and does not refill the budget.
Review date: The date an allocation is next reviewed: the date of its last decision plus the mandate's review period.; Each decision sets it. The review period is 30 days by default, and you set it on the Mandate page. At the review the desk offers Step up, Hold or Reduce for your decision. An insurance book's review comes once a year, at the end of its budget's year: you renew, change or end its premium budget.
Step up · Hold · Reduce · End allocation · Resume · Change: The acts that change an allocation. Each one is dated and stays on the record.; Change sets a new amount and gross limit at any time, and End allocation ends the allocation at any time, with the book's exits on one confirm. At a review you Step up, Hold or Reduce the amount: Step up raises it by the mandate's step, ×2 by default. Resume lifts the pause at the stop level. The review's acts and Resume arrive with the dated review and the stop level's watch. Each act is yours: the desk states its figures and its offer, and nothing changes an allocation without your confirm.
Expected ranges: The turnover, gross, volatility and drawdown that a book is built to show. The review shows the results beside them.; The desk computes them from the book's design, states them on the Allocate decision and freezes them with each act. At each review the period's turnover and gross stand beside their ranges as results; the review measures no volatility and narrows no range yet. The forecast book's levels come from its sizing, and an equity book's review mark stands at its stop level until the desk measures its drawdown range.
Style driftComing later: A book's turnover, gross or volatility outside its expected ranges. The review names it.; At each review the desk measures the book's turnover, gross and volatility from its own record and compares them with its expected ranges. A figure outside its range is style drift. The review names it beside its finding and its offer, and you decide.
Capacity: The largest amount a book can use in its markets, measured from their daily volume. The desk's offer stays inside it.; Each position stays at a quarter of its market's median daily traded value, so it can exit within five days at a tenth of a stressed day. The desk measures it over every market the book trades, and a market with no volume on record gives no measure. Capacity refuses nothing: you may set an amount above it, and the excess sits idle.
Overlay: Your standing instruction on a book: an exclusion or a de-risk, dated and with your reason.; You set it, and the desk never does. Every rebalance names it. An overlay belongs to the book, not to its allocation, so it stands across an end and a new allocation. A lifted overlay stays on the record.
Exclusion: A name that you keep out of a book, with your reason.; An overlay on a book that a catalog program's model forms: a momentum, relative-value or arbitrage book. The book's rebalance leaves the name out and names the exclusion until you lift it. A held excluded name comes to the book's trade list as a close, before any trim. A name leaves such a book's universe in no other way.
De-risk: A cut that you set on a book's target, by a stated fraction, until a date or until you lift it.; An overlay on a book that a catalog program's model forms. The book's next move, its rebalance or a trim, brings the whole book to the cut, its held positions included, and names the overlay. No order adds past the cut. One stands at a time: a new de-risk lifts the standing one, which stays on the record. When the stance on the book's theme is regraded cautious, the desk asks De-risk? for your decision.
Family cap: The most one event family may stake together, as a percent of the pool — sibling strikes share it as one exposure.; The percent you edit in a book's configuration bounds a whole event family's combined worst-case stake against the mandate pool — the solve admits a family's legs together, never separately, so the cap binds the basket as one. The pool cutover re-based each converted book's cap to keep its old budget's dollars.
Staking policy: How the desk sizes from records, stated once: the Kelly multiplier, the exploration stake, and the governor curve.; Lives on the mandate beside the pool — one policy per environment, never book configuration. The Kelly multiplier scales record-sized stakes (50 is half-Kelly); the exploration stake is what an unproven method may place; the governor curve scales sizes down as drawdown deepens. The desk applies it at every solve; you edit it on the mandate page. Read the explainer →
Exploration stake: What an unproven method may stake per question while its record is young — basis points of the forecast book's amount.; Set on the mandate's staking policy as basis points of the forecast book's amount (200 = 2%), so the stake scales with the book's allocation, never with the pool. One stake per question and method version in the forecast book: a ladder's strikes settle on one print, so the trial goes to the one offered strike with the largest edge per dollar, and a question with no fee-net edge at any offer waits. Two programs that ask one question with one young method version fund one trial between them. A question the book already stands on opens no new strike: one standing strike may add exposure inside the one stake, the others hold, and a held sibling's capital counts against the caps in full. Bounded per trade and per family like every stake. It is what lets a new method earn a record without meaningful capital: prove skill and the record sizes it up; let the record decay and deployment falls away on its own.
Drawdown governor: Before the pause, sizes scale down on a stated line as drawdown deepens — the sizing line names it when it binds.; A straight line from full size at the governor's start to zero at the drawdown pause, applied to every solved stake. The pause itself is unchanged: at its threshold, opens still refuse exactly as shipped. Both ends read and edit on the mandate — the governor start and the pause.
Target book: The desk's solved answer for the whole book — what it would hold at current views, sizes included.; Every run solves the whole forecast book at once against the mandate pool: every program's current views, held positions and working orders included. One target stands for each contract, owned by the program whose method's view the solve elected. The queue list is the standing difference between what you hold and this target — a move leaves the list the moment the target no longer wants it, and an empty list is the honest at-target.
Sizing regime: Why a trade is the size it is: an exploration stake while the record is young, or a record-sized computation.; Every solved entry states its regime in plain words on the sizing line: an exploration stake — small because the method's record is young — or a record-sized stake, the method's scored record shrinking a fractional-Kelly computation. The line also names the constraint that bound: the per-trade cap, the family or theme cap, an authored deployment cap, the pool, the mandate's exposure headroom, or the drawdown governor.
Event family: Sibling strikes of one event ladder, treated as one exposure — one view, one family cap, one confirm.; Strikes of one event ladder move together, so the desk never stages them as independent full-size entries. While the record is young, the family takes one exploration trial on the strike with the largest edge per dollar. A record-sized view over a ladder stages as a family basket: a per-strike shape under one family cap, confirmed once, each leg stating its cause. The family cap bounds each family's share of the pool.

Execution

How a confirmed recommendation becomes a venue order.

Order ticketTrade tier · yours: The memorandum your confirm mints — the fixed order terms the desk then places.; Confirm is the assent: it fixes the order terms and mints the order ticket in the same act that reserves capital and writes the orders. The ticket is the record of what you agreed to; what the venue then did — filled, partial, failed — renders beside it from the order record and never rewrites your decision. Read the explainer →
OrderOrder tier · yours: An instruction sent to a venue to buy or sell.; One instruction to buy or sell an instrument, with its quantity and price terms — journaled verbatim along with the venue's response. Cancellation requests and reviews are journaled too, as their own entries. Read the explainer →
Maximum order value: The largest order the desk places for you, on every order in every lane. Above it, placement refuses the order.; Set on the mandate in dollars. It starts at a tenth of the pool when the pool is first set, and a later change of the pool leaves it; a mandate with no pool holds none. Each order an instruction sends is valued before anything is sent: an equity buy at its notional, or at its shares times its limit; an equity sale at its shares times the current ask, or the latest trade where the quote is absent, never at its limit; an event contract at its contracts times its limit; a perpetual future at its notional. One order above the limit refuses the whole ticket, closes included, and the refusal names the order and its value.
Price collar: How far from the current quote an order may fill. The desk sends each equity order as a DAY limit at this bound.; Set on the mandate as a percentage, 5 % by default. A buy goes as a limit at the quote's ask plus the collar, and a sale, a short or a close at the quote's bid less it, rounded inside the collar to the venue's price increment. Where the quote is absent or crossed, the latest trade stands in for it, and an order with neither is refused. The order is a DAY order: a part that hasn't filled by the close expires at the venue. A limit price you set yourself places at your price when it sits inside the collar, and is blocked at placement when it sits beyond it: a buy above the ask plus the collar, or a sale below the bid less it. On Kalshi the collar admits at least one cent, and a close sells down the displayed bids no further than the top bid less the collar; the rest rests at that price. Where a venue takes a fraction of a share only at market, a close sends its whole shares as a DAY limit inside the collar and the fraction at market after them.
Limit order: The worst price you'll accept — executes at that price or better, or waits.; An order with a price bound: a buy executes at your limit price or lower, a sell at your limit or higher. If the market isn't there, it rests at the venue until the price is touched, you cancel it, or its stated lifetime ends. Even a limit that looks immediately executable may not fill — eligibility is never a guarantee.
Resting order: A live order waiting at the venue for its price.; An order the venue has accepted that hasn't fully executed — it waits (“rests”) until the market reaches its price, it's cancelled, or its lifetime ends (a day order expires with the session). Partial executions book as they land; the remainder keeps resting.
Armed exit: Your exit plan resting as live venue orders — the first level to trigger executes without another approval.; By default your exit plan is watched: the desk alerts you at your levels and closing stays your own reviewed act. Arming the exit places that same plan at the venue as a resting order pair — take profit and stop loss, good-til-canceled, executing in regular hours — so whichever level trades first fills WITHOUT a further approval, and the venue cancels the other leg. You arm each position explicitly, after a confirmation that spells exactly this out; disarming (which only cancels the resting orders — your watched plan stays) and level moves are each one confirmation away. If a resting leg is ever lost without executing — cancelled or expired at the venue — the desk tells you plainly and you can re-arm. Available on Alpaca equity positions with whole-share quantities; everywhere else, including Kalshi, the plan stays watched.
Extended sessions: Trading outside regular hours — pre-market, after-hours, and overnight, as one eligibility.; US venues run sessions beyond the 9:30 AM–4:00 PM ET regular session: pre-market, after-hours, and an overnight session. The venue treats these as one extended-hours eligibility — you opt a limit order into all of them together, not one at a time, and only a limit order qualifies (a market order always waits for the regular open). Overnight execution also depends on the security being overnight-tradable; where it isn't, the order simply trades in the sessions it can.
FillFill tier · the venue's: One completed piece of an order, at a specific price.; One completed piece of an order at a specific price. A single order can fill in several pieces. Read the explainer →
Simulated fill: A fill the desk's own engine made against live Kalshi market data — real prices, simulated money.; A simulated fill is how the desk-run Kalshi paper account executes: the engine prices each fill from the live Kalshi order book at the decision instant — a marketable order crosses displayed depth at displayed prices, never more than the book showed, and a partial leaves the remainder resting. Fees charge Kalshi's real schedule at the venue's own arithmetic. Each fill row states its provenance and its as-of. Two things are deliberately NOT modeled: your order's market impact and its queue priority — the same limits any simulator carries, stated rather than hidden. Read the explainer →
Blotter: Your running log of every order and fill.; The running journal of orders and fills — the Activity page renders it across all books.
Capital reservation: Capital held for an approved order until the venue confirms it or the order resolves.; When you confirm an order, its worst-case cost — including estimated fees — is held against the pool's headroom until fills confirm it or the order resolves. It usually lives for seconds. If the venue's answer goes missing, the hold stays on the book — labeled and aged — until the order sweep resolves it: capacity is over-held, never under-held.

Holdings

What you own once orders fill.

PositionHolding tier · yours: Something you hold now (or a closed one), with its profit/loss.; One holding, open or closed, with its profit and loss — attributed to the book that holds it. In the forecast book one position stands for each contract and side, and each of its lots names its program. Read the explainer →
Exit plan: The optional take profit and stop loss on a trade — watched levels; nothing sells automatically unless you arm them.; A per-trade pair of watched price levels — a take profit on the winning side and a stop loss on the losing side — authored on the recommendation at review, carried onto the position when it books, and amendable there while it stays open. They are the desk's alarm levels, not venue orders: when the market crosses one, the desk raises an alert and the close remains your action. Nothing auto-closes while the plan stays watched — arming it at the venue (see Armed exit) is the one exception, and you confirm it per position.
Take profit: The price where you'd take the win — the desk alerts you there; selling is your call unless the plan is armed.; The profit-side level of the exit plan: above entry for a long, below for a short. Stored at exact per-unit price grain; crossing it raises an alert — it places no order unless the plan is armed at the venue (see Armed exit).
Stop loss: The price where you'd cut the loss — the desk alerts you there; closing is your call unless the plan is armed.; The loss-side level of the exit plan: below entry for a long, above for a short. A watched level, not a venue stop order (Kalshi has no stop orders at all) — the desk watches and alerts, and the close is always your action, unless you arm the plan at the venue (see Armed exit).
Alert: A standing notice the desk raises when something it watches for you happens — it informs, it never trades.; One fired notice about something the desk watches for you: the one thesis alert (open positions still riding a rationale that no longer governs — its horizon passed, you closed it out, or a pair's hedge stands alone; it clears when you revise the thesis or the positions flatten), a threshold alert when a position's price crosses its set take profit or stop loss, or a drawdown that reached your pause threshold at the daily close. Alerts stand on the desk beneath your decision queue, stating what happened and the moment it was observed, with a link to the surface where you'd act; the block also states what the watcher last checked and its cadence — it never claims real-time. An alert settles itself the moment its subject is handled — close the position and the notice stops ringing on its own — and once you dismiss one, that exact cause never nags again. Alerts never place or close anything; the decision is always yours, on the ordinary review surfaces.
Threshold alert: The notice raised when a position's price crosses its take profit or stop loss — a scheduled check, never an order.; The alert a watched level fires: every 15 minutes through each venue's checked span — prediction markets around the clock, equities in the regular session — the desk reads market quotes for each open position that has levels set and compares the price it observed to your take profit and stop loss. A crossing raises one notice stating the level, the price observed, and exactly when it was read — figures from that observation, never recomputed. The same cause never fires twice: crossing back and forth re-raises nothing, and editing the level arms a fresh watch at the new price. It is not a venue stop order — nothing sells; the close stays your action on the ordinary surfaces.
PortfolioHolding tier · yours: All the positions you hold, grouped by book.; Your positions, viewed together. The Portfolio page (/portfolio) shows one group per book holding open positions, its trades at your mode's live marks, and the forecast book's group lists each position with its lots; its By-account tab shows positions as each connected venue reports them. Read the explainer →
Lot: One purchase batch, tracked for tax cost basis.; One purchase batch, tracked for cost basis. In the forecast book each lot names the program whose line opened it and the method version that sized that line, and a close of one program's lots sells them oldest first.
Reconciliation gap: The named difference between what a venue reports for the account and what your books account for.; When a venue's account-level holding differs from what your books record for an instrument, the difference renders beside the figures it qualifies — both quantities, the direction, and when it was last checked. A gap is named, never silently adjusted away.
Spot holding: A crypto asset you own outright in a spot account, paid in full — a balance of the asset, and no contract.; A quantity of a crypto asset that you own outright in a spot account, paid in full: a balance of the asset, and no contract. The accounts page shows each holding as the venue states it — the quantity exact to the venue's precision, the venue's value in dollars, and its cost basis labelled as the venue's — and the desk prices nothing itself. A holding is the asset, whatever pair it was bought under.
Available to trade · available to withdraw: The part of a balance an order can use, and the part that can leave the venue — a hold can bar one and not the other.; The part of a balance that an order can use, and the part that can leave the venue. A hold can bar the second and not the first, so a balance under a hold states both, each where the venue states it and as not stated where it does not.
Transfer in kind: An asset that arrives from, or leaves to, an outside address — an external flow at its instant's value, never profit.; An asset that arrives from, or leaves to, an outside address. It raises or lowers the holding with no fill and enters the account's journal as a transfer at its value at its instant: an external flow, and never profit. Where the value cannot be evidenced, the day states no result.

Attribution & performance

How orders and positions tie back to a book, and how it's measured.

Book attribution: Which book an order or position belongs to — in the forecast book, each lot also names its program.; The attribution of an order or position to a book — what makes per-book P&L possible. In the forecast book an order, its reservation and each lot also name the program whose line placed them, and a lot names the method version that sized that line, so each program's lots, trades and P&L add up on their own.
TradeTrade tier · yours: One decision from entry to flat — its legs, orders, and profit or loss as one record.; The decision unit: everything from the entry that opened it to the moment it went flat, recorded as one thing — a pair's two legs are one trade with one package result, an ad-hoc order is a one-leg trade, and adding to an open trade keeps one trade while re-entering after flat starts a new one. Round-trip profit or loss and the win rate count trades, so a completed pair counts once, not twice.
Performance: How a book is doing — P&L, exposure, drawdown.; How a book — or the whole mandate — is doing over time: profit and loss, exposure, drawdown.
Book cost basis: Cost and realized figures from the activity this app executed — venue fees included, not the venue's tax accounting.; Cost basis and app-recorded realized P&L come from the orders and fills this app placed and booked, so a book's figures always trace to its own activity. Venue fees are included on both legs: a purchase costs its fee more, a sale returns its fee less. Your venue's statements remain the authority for taxes. When a venue later reports exact fill details that replace an estimate, the affected figures restate and show a dated “updated from final fill details” status.
Total P&L: Realized plus unrealized plus attributed dividends — venue fees are already inside those figures, never added again.; A book's whole result: app-recorded realized profit or loss, plus the live mark-to-market on what it still holds, plus dividend money attributed to it. Venue fees are already inside these figures — an open position's fee sits in its cost basis, a closed one's in realized — so nothing here double-counts them.
Dividends (app-recorded): Dividend money attributed to this book from the venue's records — a short leg owes dividends, so it can be negative.; Dividends the account received (or owed, on short legs) are divided among the books whose positions held the paying stock on the venue's entitlement date, pro-rata by quantity. Money the records cannot tie to a book is named at the account level, never guessed into one. The sum shown is included in Total P&L.
Venue fees to date: The venue fees this book's orders have paid — already inside the P&L figures, shown so the cost is visible.; The venue's own trading fees, summed from this book's recorded fills. They are already inside the P&L figures: an open position's fee sits in its cost basis and unrealized P&L, and a closed position's fees are inside realized. This line discloses the cost — it is never added on top.
Day P&L: Today's change vs the prior daily close — deposits and withdrawals never count as profit.; Today's profit or loss measured against the prior daily close, shown per basis — live and paper never sum. Intraday it's a labeled estimate from current account values; deposits and withdrawals are excluded, and a venue with no baseline or no current answer is named rather than silently folded in.
Daily close: Each venue day's closing value; stock accounts file the venue's post-close read — it can differ from the official close.; The once-a-day recording of each connected account's value at its venue's own close — stock accounts after the market's close (half-days honest), prediction-market accounts after their ET-midnight day roll, weekends included where the venue trades — whether or not you open the app; the dated series that day P&L and drawdown measure from. For stock accounts the desk files the venue's own stated account value from a read minutes after the close, so the figure can differ slightly from the official closing prints when prices move just after the bell (largest around post-close earnings); where the venue serves activity records, the desk first proves the holdings sat unchanged through that window. A Coinbase futures account files Coinbase's own figure for the account from the desk's first read within 30 minutes after the exchange's close at 16:00 Chicago, so its marks and its funding can move inside that span; a Coinbase spot account is valued at the close of the one-hour candle that ends at its boundary, with the stablecoin at Coinbase's stated value. A day whose value can't be captured honestly files as a named gap, never a guess.
Armed: A risk control that is measurable and live — until then it shows as arming, never as silently protective.; A risk control is armed once the number it watches actually exists. The drawdown pause arms per basis — live and paper each on their own — at your first daily close, where the peak seeds; until then it shows as arming with its condition stated, and it never pretends to protect. Once armed, a breach pauses new trades on that basis only; closing positions is never blocked.
Win rate: Closed trades that ended green — always shown with how many it's measured over.; Of a book's completed trades, how many closed with a gain — always stated with the sample, like "4 of 6 trades closed green," because a rate without its count implies more than a small record can promise. A trade is the decision unit, so a completed pair counts once. A break-even trade counts in the sample but not as a win; a trade with no recorded entry evidence stays out of the sample entirely; and until the record holds at least one counted trade the book shows a quiet state rather than 0%.
Transaction: Any account entry — a trade, dividend, fee, or transfer.; Any account entry — a trade, a dividend, a fee, or a transfer. The app journals trades, settlements, dividends, and account-level fees today; transfers and interest render when the activity view lands.

Reading the numbers

The finance terms shown on baskets and books.

Net beta: Dollar beta = the sum of each leg's value × estimated benchmark beta; near 0 = hedge target met.; The basket's estimated sensitivity to the benchmark in dollar terms: sum each leg's signed value multiplied by its estimated benchmark beta. Near zero means low estimated first-order market exposure over the estimation window — it reduces market risk, it doesn't eliminate it.
R²: Regression R²: the share of historical basket-return variation explained by the benchmark.; How much of the basket's historical movement the benchmark regression explains — a model-fit diagnostic, read alongside beta. A neutral basket usually shows a low value, but low R² alone doesn't prove the hedge is good.
Gross / Net: Gross = long value + |short value|; net = long value − |short value|.; Gross = total capital at work (long value plus the size of the short value). Net = the directional remainder (long value minus the size of the short value).
Gross exposure: Open position value at risk: equity opening value plus event-contract stake (its maximum loss), using absolute amounts.; The mandate gate sums the absolute app-recorded opening value of every open position in the current environment, Paper or Live. For equities that is the long cost or short sale value at entry; for a fully funded event contract it is the stake paid, which is its maximum loss before fees. It is recorded exposure, not a live mark, factor exposure, or venue margin requirement.
Same-instrument concentration: How many open positions share one symbol — a warning, not a factor-correlation model.; Counts concurrent open positions carrying the same symbol and warns above the mandate's configured guideline. It does not know that different instruments may depend on the same company, index, event, sector, or macro factor; genuine factor correlation is a future risk-model upgrade.
Signal: A named, coded measurement the method ranks the basket's names by — written down and versioned.; A signal is a named, coded measurement the method ranks names by — momentum and reversal form the default blend, with earnings drift optional per book. Each signal carries an explicit version, and its record accrues per version: a formula or parameter change mints the next version instead of silently changing the figures. Distinct from a style: some styles run agent research in chat rather than a coded signal, and their copy says so.
Score: One name's standing on a signal across the basket's universe — positive leans the basket toward it, negative away.; A signal score is one name's standing on a signal relative to the rest of the basket's universe at that run — standardized, so different signals compare fairly. Positive tilts the basket toward the name, negative away. Scores are written down when the run computes them and rendered verbatim afterward; a “—” means no usable score existed for that name.
Tilt: Leaning a neutral basket toward the names the signal favors — without breaking its neutrality.; Tilting leans a market-neutral basket toward the names the blended signal favors and away from the ones it disfavors, while the construction keeps the basket's overall neutrality — the tilt changes who is long and who is short inside the hedge, not the hedge itself. The recommendation's signal summary names each blended signal and the tilt in plain words.
Method: A repeatable procedure that produces forecasts or rank signals under a versioned contract.; The registered procedure and version own its evaluator, applicability and record. A Style describes the trading approach; it is not a method version.
Method statement: A method's recorded response for an issue time: its forecast or a named reason it could not provide one.; Probability, distribution and categorical-weight responses are forecasts. A named gap stays visible but adds no forecast sample.
Statement policy: A method version's rule for when its statement is owed: the releases it needs, its not-before, its bound.; The policy sits on the version's register row. A statement is due once the venue's question is received, the named releases are held and the not-before has passed; it is owed by the bound, never later than a minute before the question closes. A version with a policy states on the statement clock's five-minute tick, not on a clock hour.
Due and bound: Due: the instant a statement's conditions hold. Bound: the instant an unissued statement becomes a named gap.; Every statement a version owes has both instants, derived from its policy and the world — nothing fires on a schedule. Past the bound the record carries the cause: the venue's question never arrived, a release was late, the reference could not be read, or the statement simply did not land in time.
Tracked: What the desk follows at the venue: its listings arrive as they happen, set and recorded by a named actor.; A venue series, event or market, or a publisher source, that the desk follows by an explicit claim of a named actor — the register, a mint, a program, a design session or a person — recorded with that actor and its instant. Nothing is tracked by rule: categories and tags bound exploration without creating a claim, and a recurring family the venue lists that nobody tracks reads not tracked.
Listing received: The instant the desk received the venue's question; against the venue's own listing instant, the listing lag.; Every tracked question carries the venue's listing instant and the desk's receipt instant; their difference is the listing lag, stated on the plane page's feed line in the viewer's clock. A statement's not-before may be the listing itself.
Feed freshness: Per tracked family: the lag behind the venue's clock and the objective it is held to — past it, the family reads late.; The objective is the serving window, fifteen minutes. A family past it reads late and a failed read names its error; the state renders live at the viewer's clock and crosses the objective without a navigation.
Expected release: A source's expected arrival for the current period; one with a publication instant reads late once it passes unreceived.; The registry's rule per source. A source with a publication instant — a daily UTC window, or a station window in local time — carries a receipt objective after it, thirty minutes: a period unheld past the objective is late on the source's standing and in a dependent statement's gap cause, and one late station among held ones is that station's lateness, never the source's. A daily read with no publisher instant is owed at its clock and has no lateness verdict.
Forecast: A predictive probability or distribution about a defined target, issued from the evidence available at that time.; Includes probability, distribution and categorical-weight arms. Named gaps and raw equity rank signals retain their own counts and never become probability forecasts.
Market view: One listed contract's projection of a method's forecast, or the named reason that projection is unavailable.; The recorded projection retains its comparison price and evidence. It is distinct from a Brief's current research view or an observed price mark.
Method record: The history of what a signal said versus what the market then did, naming its owner: this book or a method version.; The Method record compares each forecast the book's method issued with what the market then did at the signal's stated window — whether or not anything was placed. Every figure carries its sample size and interval, misses are named and counted rather than filled, and under the methodology's record minimum (5 settled questions or 5 scored windows) the record says “not enough resolved observations yet” instead of showing a number. Each record names its owner. This book's method record reads the book's own signals. A method version's method record pools every user's books that run it and freezes at its promotion: an Allocate decision cites it with that date, and never as the book's own.
Source series: A published statistical series the desk holds, cited with its period, its publication date, and the desk's read time.; A source series is one keyed statistical line within a published dataset the desk holds — the CPI index, a weekly gas price. When a stated view cites one, the Sources item names the series and three facts in order: the period the figure describes, when the publisher issued it, and when the desk read it — three different clocks, never blurred into one.
Momentum: The coded momentum signal — recent winners keep winning — ranking names inside solver baskets.; The coded momentum signal ranks the basket's universe by recent price strength — winners keep winning over a months-long horizon. Inside relative-value baskets it is one of the default blend's two legs, versioned in the registry. Distinct from the Momentum style, which forms whole books from a momentum ranking of its own and re-forms them on a clock — this entry names the signal leg.
Reversal: The coded reversal signal — sharp recent losers tend to bounce over about a month.; The coded reversal signal leans toward names that fell hardest over the recent window — short-term overreactions tend to partially unwind over about a month. The default blend pairs it with momentum so the two horizons balance each other, versioned in the registry like every leg.
Earnings drift: The tendency of a stock to keep moving in the direction of an earnings surprise in the weeks after the report.; The tendency of a stock to keep moving in the direction of an earnings surprise in the weeks after the report; as a signal leg it tilts an equity basket toward recent positive surprises. Off unless you switch it on — a method parameter on your book, not a suggestion.
Pair: Two orders that belong together — an event contract and a related stock position, reviewed and placed as one trade.; One trade in two legs on two venues — an event contract and a related stock position — reviewed and placed with one confirm, booked as one package, and watched through the event's resolution.
Net edge: The gap after fees between the trade's stated view and what the price implies — the amount it is trying to capture.; One meaning everywhere (ADR-0028/ADR-0033): the gap, after fees, between a stated probability view and what the price being paid implies. On a pair the view is the related stock's implied odds; on a single-venue evaluation it is the stated house view against the fee-adjusted breakeven. Server-computed on every recommendation that shows it, and shown as-is.
Basis risk: The part a hedge can't remove: the stock moves for its own reasons, not only the event's.; The part of a pair a hedge can't remove — the stock can move for reasons other than the event, so the two legs don't offset perfectly. Every pair recommendation states this in plain words.
Implied probability: The chance the market's own price implies — a contract priced at 34¢ implies about a 34% chance.; The probability read off the executable price of the exact contract being evaluated: paying 34¢ for a contract that pays $1 implies roughly a 34% chance of it paying out. It is the market's collective read at that moment — not the desk's view, and not advice.
Breakeven probability: The chance at which this trade is exactly fair after fees — your view must beat this to expect a profit.; The probability at which the trade's expected payout exactly covers its cost including the venue's fee — computed for the exact order (price, size, and this venue's fee schedule), never a generic figure. If your honest view of the odds is above the breakeven, the trade has positive expected value; below it, it doesn't.
Catalyst clock: The countdown to the dated event this book keys on — always stated with where the date came from.; Time remaining to the book's catalyst — the scheduled event its method trades around. The clock names its source (an earnings-calendar entry with its estimated-vs-confirmed status, your configured date, or the market's own close time) and says plainly when the date is missing, unparseable, or already past — it never guesses.
Spread: The gap between the buy and sell price — the built-in cost of a round trip.; The difference between the price you'd pay to buy and the price you'd receive to sell at the same moment. It is the least cost of entering and exiting: the market must move at least the spread in your favor before a round trip breaks even, before fees.
Expected carry: The carry the desk expects over the horizon: the pair's ten newest resolved funding days with the current rates.; A pair's carry over the book's horizon at a size, in a structure, at an instant: each leg's next payment at the venue's estimate inside the statement's lead, else its latest print, then each leg's remaining funding instants at its own run rate, the mean of its payments over the pair's ten newest resolved funding days. One owner states it, with its instant, each rate's kind and the size it prices; the entry, the switch and the swap read it, and a row ranks on it. A pair under the record minimum has none, and its row says why.
Paid carry: The carry at the funding instants the venues paid on the pair, on the desk's stored marks.; Each leg's payment at each funding instant the venue paid, at the venue's posted rate on the desk's stored mark for that instant, at the contracts held then, signed for the leg's side. One series owns it: the Close reads its run of intervals under their bar, the record books its sum over the holding beside the account's own ledger, and the funding notice states it to date.
Latest print: A venue's newest posted rate at its instant: shown beside expected carry, read by no decision that stages an act.; Each venue's newest posted funding rate, with the instant it paid at. The screen states it beside expected carry, named; no decision that stages an act reads it, and the funding notice triggers on the next payment's rate, the venue's estimate inside the lead, else the latest print, named by its kind. A rate of exactly zero counts as observed, “zero k of n” over the window; Kalshi documents a dead band, a rate under 0.01 % in absolute value posted as zero, and the screen gives no other zero a cause.
Standing direction: The side a pair trades: the sign of its record's net carry that has held three resolved days of its newest ten.; The direction a carry pair trades and is stated on: the sign of its realized net carry over its newest ten resolved funding days, once that sign has held three resolved days; a turn pending for fewer days leaves the standing direction as it was, with the turn's start and age beside it. A held pair keeps the held structure until the standing direction turns against it, and the reversal then states the turn's start. Before a pair has a record, the newest rates seed it. One print never turns it.

Availability

How a theme's chip reads in every theme list — derived from what you hold, what's connected, and the mode (ADR-0024, ADR-0032).

Running: You hold a book minted from this program — flat or deployed, it reads running.; A program reads running when you have a minted, non-archived book on it — deployed or flat (adopting a thesis is what readies placement; the book card carries the flat truth). The chip names the workspace mode it speaks for — “running · paper” — because each mode lists its own books. Running programs sort first in every list, and front their theme on the shelf.
Book can open: A venue covers one of this program's markets in this mode — a book could open and run here. Access, not advice.; Book can open means at least one of the program's published markets has a fill source in the named workspace mode — “book can open · paper”. It states access only: whether a book could open and run here, never whether the desk thinks you should — the desk's view lives in the theme's brief and its stance. Minting a book is still your explicit action.
Needs a connection: The program's market has a shipped runner, but its venue isn't connected — the chip names it and links Settings.; Needs a connection names the exact venue whose connection would make the program runnable (Kalshi for prediction markets, Alpaca for stocks). Connecting is one step on the Settings page; nothing runs until you mint a book.
Live mode only: This program's market can only fill in Live mode — no fill source exists for it in Paper mode.; Live mode only appears in Paper mode on a program whose market has no fill source there — no venue paper environment and no desk-run paper account — so no connection could make it runnable, and Settings cannot unlock it. Today no shipped program reads this chip: stocks fill in Alpaca's paper environment, and prediction markets fill on the desk-run Kalshi paper account (simulated by the desk against live Kalshi markets). The chip stays defined for a future market without a paper fill source. While Live mode is open to enter, the action is the mode switch; before launch, these pages state 'Live mode opens at launch' instead of naming a switch (ADR-0064).
Roadmap: On the desk's roadmap — not open for new books yet; greyed and actionless.; Roadmap marks programs and styles not open for new books yet — either no shipped runner exists for their markets, or the style is held back while the desk fronts its launch styles (existing books keep running either way). They stay visible greyed under their theme's story, with no actions — never a hidden promise.

The mode

Paper or Live — which money the whole workspace reads (ADR-0032). One mode at a time, chosen on the badge — the app only ever falls back to Paper, stated (ADR-0064).

Mode: Which money the app shows: Paper (simulated) or Live (real). You choose it; ended live access lands back in Paper.; The workspace mode selects which of your two environments every page reads — paper accounts in Paper mode, live accounts in Live mode. The badge in the corner names the current mode and carries the switch (entering Live asks you to confirm). The selection is remembered across sign-ins and devices, and the app never switches you INTO a mode on its own. One stated exception runs the other way (ADR-0064): if your live access ends while Live still holds positions or working orders, the live view stays — close-only, labeled on the badge — until Live is flat and order-free; then the workspace reads Paper and says so. Your stored choice is never rewritten, and both environments keep accruing whichever mode you're in; switching changes what you see, never your data.
Paper mode: Simulated money — practice with the full desk, stocks and prediction markets both. Not real dollars.; Paper mode shows your paper (simulated-money) accounts only: positions, day P&L, headroom, and books are Paper's. Stocks trade in your connected Alpaca paper account; prediction markets trade on the Kalshi paper account the desk itself runs — built in, no connection, fills simulated by the desk against live Kalshi markets. Orders you confirm in Paper mode reach only your paper accounts.
Live mode: Real money. Every figure is a live account's, and orders you confirm commit real dollars.; Live mode shows your live accounts only — Kalshi and a live Alpaca account if connected. Entering Live mode asks for an explicit confirm; leaving it doesn't. Orders you confirm in Live mode reach only live accounts, and every placement still gets your per-order review first. Live mode opens at launch (ADR-0064): until then the whole desk runs on paper, and the badge states it — a wound-down Live stays visible close-only until it is flat and order-free.

Plans & billing

Free on paper, paid to open live (ADR-0038). What a plan grants, and how your subscription is managed.

Plan: What an account can do — the Free plan (the whole desk on paper) or the Live plan (opening real-money positions).; A plan is a named set of capabilities, not your account's state. Both plans run the whole desk on paper, keep full sight of live, and let you cancel orders and close live positions. The Live plan adds two things: opening or increasing real-money positions, and a higher daily chat limit. Your account's instance of a plan is a subscription.
Subscription: Your account's active Live plan — it unlocks opening real money. Cancel anytime; exits and paper stay free.; A subscription is your account's instance of the Live plan: it turns on opening and increasing real-money positions. Without one you're on the Free plan — the whole desk on paper, full sight of live, and you can always cancel orders and close live positions. It's managed from Settings through the payment provider's hosted checkout and portal; card details never touch Quantreno.

The style catalog

A practical style catalog — the styles relevant to the desk's current and planned venues, grouped into six families. Shipped styles open new books today — the three launch styles, Event-Driven, Relative Value, and Momentum; Roadmap styles are committed but not open for new books yet (a few already run for existing books — their rows say so); Recognized styles are named here and buildable when prioritized; Out of scope styles need a venue or cadence we deliberately don't have. Each running style also states how its Run happens: auto-pipeline — the app constructs and sizes the recommendation in code — or agent-led — the run is commissioned to the desk's background research worker: the chat acknowledges right away, the desk screens and researches the book's markets against its rules, and the report — with any recommended order cards — arrives in the same chat, typically within a few minutes. Either way, nothing is placed without your review.

Event-DrivenTrade a known catalyst.

Event-DrivenShippedAgent-led: Trade around known catalysts (data releases, votes, hearings, decisions) — research the catalyst harder than the market and position before or just after.; The family's generic catalyst style.
Merger ArbitrageRecognized
DistressedRecognized
Special SituationsRecognized
Earnings PlaysRecognized: PEAD enters first as a signal-blend leg.

Relative ValueMispricing between related instruments.

Relative ValueShippedAuto-pipeline: Find mispricing between mathematically related instruments — prediction-market pairs, or the flagship market-neutral equity baskets the solver constructs.; Neutral solver baskets — the flagship.
Pairs / Stat ArbRoadmap
Volatility ArbitrageRecognized: Prediction-market probability vs matched option-implied event odds — needs options execution.
Cross-Venue ArbitrageRecognized: Gated on cross-instrument books.
Convertible ArbitrageOut of scope: No convertible-bond venue.
Capital-Structure ArbitrageOut of scope: No credit venue.
Fixed-Income ArbitrageOut of scope: No fixed-income venue.

Systematic DirectionalRide or fade price behavior.

MomentumShippedAuto-pipeline: Hold a theme's momentum winners against its losers — a coded signal ranks the constituents by trailing return, the book forms from the ranking's top and bottom, and the desk re-forms it on the book's rebalance clock, staging every rebalance ticket for your review.; How this style usually behaves: many small gains, with occasional sharp losses when trends snap back.
Mean ReversionRoadmapAgent-led: Fade overreactions — the desk's researcher investigates sharp moves against the book's rules in chat and proposes trading back the ones the catalyst doesn't justify; an agent-research method, not a coded signal.; Runs today for existing books; not taking new books yet.
BreakoutRecognized

MacroTop-down programs and causal chains.

Macro ThematicRoadmapAgent-led: Map a macro program's causal chain and trade its second- and third-order effects across markets as each link triggers.; Runs today for existing books; not taking new books yet.
EM / CountryRecognized
Systematic Macro (CTA)Out of scope: No futures venue.

Volatility & ConvexityOwn optionality.

Tail RiskRoadmapAgent-led: Buy protection for your pool on a premium budget — the book's research proposes each contract that pays in the adverse scenario that the book names, and you read its terms and confirm it. Most settle at zero; a payout offsets part of the pool's loss in that scenario.; Runs today for existing books; not taking new books yet.
Long VolatilityRecognized: Needs options execution.
Gamma ScalpingOut of scope: Rehedging cadence conflicts with per-action approval.

Carry / IncomeHarvest steady premia.

Premium SellingRoadmap: Selling richly priced contracts for frequent small gains — opens the family.
Covered CallsRecognized: Needs options execution.
Cash-Secured PutsRecognized: Needs options execution.
Dividend CaptureRecognized
FX CarryOut of scope: No FX venue.

Two words we deliberately keep apart: a Deployment cap bounds what a book may commit of the mandate pool, while Book attribution is which book an order or position belongs to. And a Position is one holding, while your Portfolio is every open position you hold, grouped by book.