Kalshi & Alpaca paper trading — in early access

AI-powered trading.
Desk-grade discipline.

An AI agent researches, sizes, and drafts every trade; you approve each order, and the risk limits you set are enforced at the moment of placement. In early access the desk paper trades across equities and prediction markets — starting with Alpaca and Kalshi, on simulated money; live accounts open at launch.

Early access — free on paperYou approve every orderRisk limits enforced at placementNot a black box

Free in early access — simulated money, no credit card required.

Three styles front the desk

We ship the methods we can stand behind end-to-end. The other shipped styles return as each earns its keep.

Launch style

Event-Driven

Trade around known catalysts — Fed decisions, CPI prints, hearings — researched before they're priced in. Kalshi-first, where your view on an event trades directly.

Launch style

Relative Value

The quant flagship: market-neutral long/short baskets on Alpaca, and consistency scans across related Kalshi markets. Paid for being right on relationships, not market direction.

Launch style

Momentum

A coded ranking holds a theme's recent winners against its losers and re-forms the book on a clock. Every rebalance waits as one ticket for your review — nothing trades unattended.

RoadmapTail RiskMean ReversionMacro Thematic

Runs today for existing sleeves; not taking new sleeves yet.

Not a chatbot. A trading desk.

Professional desks separate research, execution, and risk oversight. Now you get that structure too — free on paper in early access.

Research

Triangulated web + X/Twitter scanning surfaces catalysts before they're priced in. Never trade on a single source.

Discipline

Kelly-criterion sizing and a mandatory thesis on every trade, plus a take-profit / stop-loss exit plan the desk watches for you. No FOMO, no YOLO.

Risk officer

Set exposure and daily-loss limits once, plus a same-instrument concentration warning — every opening order you confirm is checked against your whole book before it places. A limit that would be broken blocks the open; exits are never blocked.

Memory

Every order, thesis, and result persists across sessions — an auditable track record the agent carries into every conversation.

The quant engine

Institutional portfolio construction, retail account.

Most "AI trading" tools just pick tickers. Quantreno's finance engine builds market-neutral baskets, measures beta and R² against the market, and tilts toward a momentum + reversal signal — the construction a hedge-fund desk runs, automated on your account.

  • Target near-zero market exposure, so you're paid mainly for being right on the names — not market direction.
  • A per-name signal tilts the basket toward what it expects to outperform, keeping it neutral.
  • Beta, R², and net exposure on every recommendation — you see the risk before you confirm.
recommended basket
1you say it
you > Build a market-neutral AI-chip spread, $10k
2the solver's inputs
universeMU · AMD · NVDA · TSM (semis)
signal σ+1.5 · +0.6 · −0.9 · −0.6
budget$10,000 gross
targetsmax signal tilt · net $0 · β-$ ≈ 0
3the recommended basket
Chip-vs-chip spreadtarget: net β-$ ≈ 0
LONGMUsig +1.5$3,000
LONGAMDsig +0.6$2,000
SHORTNVDAsig −0.9$3,000
SHORTTSMsig −0.6$2,000
Gross $10,000 · long $5,000 / short $5,000 · net $0. The bet is MU/AMD vs NVDA/TSM on a +1.9σ signal spread — hedged name-vs-name toward the solver's neutral target. Confirm to trade.

From idea to executed trade, in one chat

The full workflow — discovery, research, sizing, execution, and monitoring — in a single conversation, on simulated money during early access.

01

Connect Alpaca paper

Link an Alpaca paper account in seconds — credentials encrypted at rest, never stored in plaintext. Kalshi and live connections open at launch.

02

Tell the agent an idea

Say "hedge my tech exposure" or "bet on a March rate cut with $200." The agent shapes it into one of the desk's launch styles, researches, and sizes the trade.

03

Review and execute

Every recommendation shows the edge, sizing, net beta, and risk. You confirm — the app places exactly what you reviewed. Nothing trades without your confirm.

Built so you stay in control

It's your money and your own venue accounts. The architecture treats that as the constraint that matters most.

The AI can't touch your orders

On every venue the agent recommends and reads — it has no path to an order. Only the app can place, only the exact trade you confirmed on the review sheet, and only after it clears the risk limits you set.

Every leg checked before anything places

A hedged basket is validated leg by leg at review — if any leg can't place, nothing is sent. Cross-symbol orders aren't atomic once submission starts, so if a leg is rejected mid-basket, placement stops and everything already filled is journaled and shown — never hidden.

Everything on the record

Every venue instruction is journaled verbatim — request and response — and closes are verified against actual fills, not order acceptance.

Latest briefs

The desk’s freshest research — the actual output, not a sample. Each brief is dated, links its sources, and says so out loud when it has gone stale.

Free in early access.

Early access runs on simulated money — research, sizing, and approval-gated paper trading across equities and prediction markets, starting with Alpaca and Kalshi. Live trading and the paid plan open at launch; early access stays free.

Stop trading on gut feel.
Start trading with a desk.

Your AI quant is ready. Run your first sleeve on paper — simulated money, real markets — in minutes.