Kalshi arbitrage: how it works — and its limits

"Arbitrage" gets thrown around loosely in prediction markets — usually by people selling something. What follows is the precise meaning, the three forms the trade takes in practice, the fee and execution math that eats most of it, and what a realistic, risk-disciplined version looks like — including the relative value trade next to it, which accepts a little risk in exchange for edges that actually persist.

What arbitrage means in an event market

A Kalshi contract pays $1 if an event happens and $0 if it doesn't, and trades between $0 and $1 on a per-market price grid (most markets quote in whole cents; some use finer increments). Prices are probabilities. That gives arbitrage a precise meaning: whenever a set of prices implies probabilities that are mathematically inconsistent — probabilities that must sum to 100% but don't, or an "easier" threshold priced below a "harder" one — there is, in principle, a combination of positions that locks in a profit no matter the outcome.

Note what that definition excludes. Buying a contract you think is mispriced is not arbitrage — it's a directional view that can lose. Arbitrage is only the case where every outcome pays you. Keeping that line sharp is the single most useful habit this topic teaches.

Form 1 — bucket consistency inside one event

Kalshi often lists an event as a set of mutually exclusive, collectively exhaustive ranges — "Where will CPI land?" split into buckets. Exactly one bucket settles YES, so the YES prices should sum to about $1.00. When they don't, the inconsistency is tradable.

This is real and it appears — most often in the minutes after news moves one bucket and the others lag. It is also exactly what market-making firms watch for, so the windows are short and the size available at the quoted prices is usually small.

Form 2 — monotonicity across thresholds and dates

Threshold markets must be ordered: "above $70 by Friday" cannot be less likely than "above $72 by Friday", and "before July" cannot be less likely than "before April" for the same event. When a harder condition trades above an easier one, buying the easier and selling the harder locks the inversion. These show up across calendar series and strike ladders, again mostly when one leg reacts to news faster than its neighbors.

Form 3 — the same risk priced on two venues

The widest — and least clean — form: the same underlying risk priced differently on Kalshi and somewhere else. A Fed-decision contract against rate-futures-implied odds; an oil-price contract against energy markets; an event contract against the equities that would move on the same outcome. True cross-venue arbitrage — simultaneous, offsetting, guaranteed — is rarely available to a retail account: settlement terms differ, timing differs, and you usually can't perfectly offset an event contract with anything else.

What exists instead is cross-venue relative value: when Kalshi's implied probability and the probability implied by a related market genuinely diverge, you can buy the cheap expression and hedge with the related one — accepting basis risk, the possibility that the two legs don't move together. It's not riskless, and honest practitioners don't call it arbitrage. It is, however, where the more durable edge in this family tends to live, because it doesn't vanish the moment one market-maker updates a quote. That paired-leg trade is the shape Quantreno's cross-market desk is built around — one reviewable trade, both legs stated, basis risk named on the recommendation.

The math that eats the edge

Before any of this is profit, three costs come out:

• Trading fees. Kalshi charges a fee that scales with price — largest near 50¢, smaller near the extremes. It is charged once per order rather than per contract and rounds up to a whole cent, and some series carry different multipliers, so check the current fee schedule. Fees dominate this trade: at a 100-set order size, the worked bucket set above pays about 4.6¢ per set in taker fees across its three NO legs — over three-quarters of the 6¢ gross — leaving roughly 1.4¢ net per set before slippage. One set at a time, the whole-cent rounding takes the entire 6¢.
• The spread and depth. The prices that make the inconsistency are the quotes on top of the book. Filling real size means walking the book, and the edge shrinks with every level.
• Legging risk. Multi-leg trades don't fill atomically. If the second leg moves before you're filled, the "locked" profit becomes an open position you didn't plan.

This is why "kalshi arbitrage betting" content that promises steady riskless income is misleading: the pure form is a scavenger hunt for small, fast-closing gaps, not an income stream.

A disciplined way to trade the family

Treat consistency scanning as a screen, not a strategy: scan related markets for incoherent pricing, then ask why it exists. Sometimes it's a genuine lag (tradable), sometimes a settlement-rule subtlety you missed (not a mispricing — read the rules twice), and sometimes stale quotes with no real size behind them. Size small, compute the fee drag before entering, and prefer the relative-value framing — a stated edge with a stated risk — over the arbitrage framing that pretends the risk away. How Quantreno runs event markets follows exactly that discipline: consistency scans surface candidates, every trade is sized under your mandate's limits, and nothing places without your review.