Prediction-market order books look like stock ladders, but the instrument is different: a binary (or multi-outcome) claim that pays on a defined event.
This chapter translates bid, ask, spread, and depth into YES/NO contracts and shows what traders actually pay. Regulated venues expose a classic ladder; many crypto-native markets led with an automated quote and added limit books where makers provide CLOB liquidity. The read skills below apply wherever your fill came from a bid/ask stack.
The contract first
On a headline binary, YES pays $1 if the event happens and $0 otherwise; NO pays $1 if the event does not happen. You can often buy YES or buy NO (economically similar to selling YES in a complete market). Prices should satisfy YES + NO ≈ $1, minus spread and fees—arbitrage keeps that relationship tight when rules and capital allow.
If you buy YES at 64¢, your maximum loss is 64¢ per share and your maximum win is 36¢ per share before fees. Buying NO at 38¢ when YES is expensive is another way to express a bearish view on the same event.
Multi-outcome elections extend the same ladder idea: each candidate (or party) has its own book or linked books, and prices across outcomes should sum to roughly 100% when markets are complete. The binary YES/NO case is the teaching default because math and UX are clearest there.
Bid, ask, and your execution
Suppose the YES book shows a best bid of $0.61 for 5,000 shares and a best ask of $0.64 for 3,200 shares. A market buy YES pays $0.64 (you lift the ask). A market sell YES receives $0.61 (you hit the bid). The mid is $0.625, which media may quote as the “market probability.”
Your edge must clear a three-cent spread plus fees before you profit versus the mid. The mid at 62.5¢ reads as roughly 62.5% implied; if you lifted the ask at 64¢, your breakeven is higher than the headline number.
Implied probability with friction
Buying YES at 64¢ means you need the event to happen often enough that the $1 payoff justifies entry after fees. Selling YES at 61¢ means you believe true odds are lower than 61% or you are closing a long. A round-trip buy then sell loses the spread plus fees unless your view changed in between.
Breakeven thinking is the bridge between trading and forecasting: the ask is the price at which the market will sell you conviction; the bid is the price at which it will buy it back.
Depth: size at each level
Depth answers how much you can trade before price moves. If only 500 shares sit at the best ask and you buy 5,000, you walk the book—paying 64¢, then 65¢, then 67¢ on subsequent levels. Your average fill (VWAP) can be 66.1¢ when the headline ask was 64¢, shrinking edge versus a model that said 70%.
Election week often brings deep books on major contracts; obscure props may offer one level and punishing slippage. Always check cumulative size within a few cents of the mid before sizing up.
NO side and equivalence
If YES trades near 64¢, NO often trades near 36¢, though the relationship is not always exact at every second. Arbitrageurs keep YES price + NO price ≈ 1. If YES mid is 62¢ but NO mid is 30¢ (sum 92¢), someone can buy both for 92¢ and collect $1 at settlement—gross profit minus costs. Persistent gaps usually mean fees, transfer limits, or different resolution rules, not free money.
Cross-venue comparisons only work when the resolution sentence matches. A regulated book at 58% and an offshore pool at 65% on the “same” headline invite bots when capital and rules align both legs.
Spread as uncertainty tax
Wide spreads appear when volatility is high, liquidity is thin, resolution risk is murky, or the event is imminent (gamma risk). A ten-cent spread on a 50¢ contract is enormous relative to a one-cent spread on a liquid equity—prediction markets are often wider than large-cap stocks.
| Mid | Spread | Spread as % of mid | Practical read |
|---|---|---|---|
| $0.50 | $0.02 | 4% | Tradeable for small discretionary size |
| $0.50 | $0.10 | 20% | Need a strong edge or patient limits |
| $0.15 | $0.06 | 40% | Long-shot book—extreme caution |
Last trade vs mid vs mark
Platforms may display last trade (can be stale), mid (average of best bid and ask), or mark (a risk system’s fair value for margin). For forecasting, prefer mid with spread context or recent volume-weighted trades. A last trade at 55¢ with a 48¢/52¢ book is stale noise; a mid at 60¢ with no size at the touch is a fantasy quote.
Limits in fast markets
A limit buy below the ask rests in the queue until someone sells into you—useful when you have a model price but not urgency. A limit sell above the bid waits for buyers. During fast news, limits may fill only after the mid has already moved; the saved spread is the compensation for patience. Market orders are the right tool when the information half-life is shorter than your queue wait.
If your limit never fills while the mid drifts away, you did not “miss” the market—you chose not to pay the spread. That is a forecasting discipline decision, not a platform bug.
Reading the tape
Useful patterns include bid stacks thickening (buyers accumulating), ask walls (large sell orders capping price until absorbed), spread tightening (competition among makers), and spread blowing out (news risk or liquidity pull). A burst of aggressive buys that lifts last trade but not resting depth may be a thin sweep rather than durable conviction—check whether size traded at each level.
Worked example — edge vs book
You believe true YES probability is 70%. The book shows bid 58¢ and ask 60¢. Buying at 60¢ leaves ten cents of edge versus your model before fees—attractive. If the ask is 68¢, edge shrinks to two cents—maybe pass. If the ask is 60¢ but only 100 shares sit there and you need 10,000, a simulated walk might average 66¢ and erase the edge.
Holding 1,000 shares bought at 60¢ to settlement wins 40¢ per share if YES wins and loses 60¢ if YES loses. The same position entered at a 66¢ VWAP wins only 34¢ on YES—slippage is not abstract.
Fees and holding to settlement
Entry price is only half the story. Taker fees on aggressive orders, maker rebates if you provide liquidity, and exit fees when you flatten before resolution all enter breakeven. Holding to settlement removes exit spread but locks capital until the oracle or exchange rule fires. A good book price with a two-point fee round-trip can be worse than a mediocre pool fill with a one-point fee—compare all-in, as in the AMM chapters.
Platform differences
Regulated CLOBs show official bid and ask. Pool-first crypto flows often start from a single percentage with optional limits; hybrids may fill you on either layer. Always know which engine printed the trade—not only the brand on the app icon.
Correlated and basket contracts
Some platforms list baskets (e.g., “party wins presidency and senate”) with their own books. Spreads can be wider because hedging is harder. The YES/NO binary skills still apply: read the resolution sentence first, then the ladder, because basket language is where expensive mistakes cluster.
Closing discipline
Good CLOB traders write down intended hold, max spread paid, and exit plan before entry. Prediction markets punish impulse market orders more than equity day-traders expect, because event variance is discrete and spreads are wide. The book rewards patience more than hero clicks.
Before you click: confirm the resolution sentence matches your model; read bid, ask, and spread in cents—not only mid percent; compare your size to size at the touch; sanity-check YES plus NO near one dollar on-platform; glance at a second venue if you care about consensus; and include fees on entry and exit.
Time in force
Good-til-cancel limits rest until filled or canceled; day orders expire at session end on venues with sessions. Prediction events often run 24/7—know whether your limit survives overnight or disappears, especially around resolution blackouts.
Key ideas
You trade at bid or ask, not the mid shown on TV. Depth determines slippage for real size. YES and NO are linked by arbitrage near $1. Spread is a first-class variable in forecast quality—not noise.
Next: systematic pros and cons of order-book design for event markets.