Slippage on Kalshi: Why Your Fill Was Worse Than the Price You Saw

The one-cent tick is bigger than it looks

This is the part that surprises people arriving from equities. If a $200 stock slips by a cent, that is 0.005% and nobody notices. Kalshi contracts settle at $0 or $1, so they trade between roughly 0.01 and 0.99, and the tick is also one cent.

Buy a contract at 0.33 and slip one cent to 0.34 and you have paid 3% more for the same position. If your strategy's edge is two or three cents per contract — which is a perfectly respectable edge in prediction markets — a single tick of slippage can consume between a third and all of it.

That is why traders who never thought about execution costs before start thinking about them on Kalshi. The market did not become more hostile; the denominator got smaller.

How the yes/no book creates it

Kalshi shows two sides, and both are bids. There are bids to buy yes and bids to buy no. There is no separate list of yes offers, because a no bid at price p is arithmetically an offer to sell yes at 1 − p.

That means the price you cross when buying yes comes from the no side. In a real recorded book from a BTC 15-minute market, the best yes bid sat at 0.32 while the best no bid sat at 0.67 — so the implied yes ask was 1 − 0.67 = 0.33. The market was one cent wide, and a buyer paid 0.33 rather than the 0.32 they might have quoted as "the price".

Reading only one side of the book, or reading a last-trade price, hides that entirely. The first surprise is often not slippage at all but the spread — paying the ask when you had mentally anchored on the bid.

What a real Kalshi book actually looks like

Generic explanations of slippage cannot tell you how much depth is there, because that is a measurement rather than a definition. On DepthFeed's recorded archive, the flagship BTC 15-minute markets typically show a one-cent (single-tick) spread with around 96 price levels on each side (up to the full 100) and roughly 156,000 contracts of resting depth across both sides, measured as of June 21, 2026.

A book that deep sounds like slippage should be impossible, and for small orders it very nearly is. The catch is distribution: resting size is not spread evenly across those levels. It clusters, and the levels immediately behind the touch are frequently far thinner than the total suggests. Total depth tells you the market is liquid; the shape of the first few levels tells you what your order will actually cost.

A worked example on a real ladder

Take the recorded book above. Converting the no side into yes offers gives roughly 635 contracts available at 0.33, about 1,351 at 0.34, and about 2,645 at 0.35.

Buy 2,000 contracts of yes. The first 635 fill at 0.33, the next 1,351 at 0.34, and the remaining 14 at 0.35. Total cost is about $674, so the average price paid is roughly 0.337 — not the 0.33 on the screen.

That is about 0.7 cents of slippage per contract, roughly $13.80 on the order, or a little over 2% of the cash committed. A backtest that assumed it filled 2,000 contracts at 0.33 would have recorded $660 and quietly invented $14 of profit that never existed.

Why your backtest did not warn you

Almost every prediction-market backtest is built on a price series — last trade, midpoint, or a periodic close. A price series records that a trade happened at a price, not that your size could have traded there.

This produces a specific and consistent failure: the backtest is most optimistic exactly where it matters most, because slippage grows with size and with urgency, and those are the moments a strategy makes or loses its money. Strategies that look strong on midpoint fills and disappear in production are usually not overfit — they were priced with a fill model that never existed.

The fix is to replay against the book that was actually displayed at the decision time, walk the ladder for the intended size, and record the unfilled remainder rather than assuming the whole order went through.

Reducing it, in order of effectiveness

• Size to the book rather than to conviction — check what is resting before choosing a quantity.
• Use limit orders and accept partial fills; a market order is an instruction to pay whatever the ladder asks.
• Avoid the minutes around settlement on short-dated markets, when spreads widen and resting size thins.
• Measure the cost of crossing the spread separately from slippage — on a one-cent market the spread is often the larger of the two.
• Make your backtest walk the ladder, so the cost shows up in research instead of in your account.