Sharp Markets primer — prediction markets 101

A sportsbook hides probability inside a price you have to decode. A prediction market puts the probability on the screen. Same question, cleaner instrument — and once you can convert between the two notations, the two venues become one market you shop across. This lesson covers what a prediction market is, why price is probability, how order-book pricing differs from house pricing, what fees replace vig, and the two risks worth worrying about.

What a prediction market actually is

A prediction market lists a question with a yes-or-no answer and a fixed resolution date: "Will the Fed cut rates in March?" "Will this team win the championship?" "Will CPI print above 3.0%?"

Each is a binary contract. Buy one and at resolution it pays exactly $1.00 if the answer is Yes and $0.00 if the answer is No. No partial credit, no push. Because the payoff is fixed at a dollar, the only variable is what you paid: buy at 62¢ and you risked 62¢ to make 38¢; buy at 8¢ and you risked 8¢ to make 92¢.

You can also sell. Selling a Yes contract at 62¢ means collecting 62¢ now and owing $1.00 if the answer is Yes — economically identical to buying the No contract at 38¢. Two sides, one instrument, either one click away.

Price is implied probability

A contract trading at 62¢ is the market saying the event has a 62% probability. Not "roughly 62 after some arithmetic." Exactly 62, because a contract with probability p of paying $1 has a fair value of p dollars.

Compare the work a sportsbook makes you do. American odds branch — negative odds convert as −O / (−O + 100), positive as 100 / (O + 100) — and then the two sides sum above 100%, so you devig before either number means anything.

On a prediction market, EV is a subtraction. If your read says an event is 66% and the contract is offered at 62¢:

EV per contract = (your probability × $1) − price = $0.66 − $0.62 = $0.04

Four cents of expected value on 62 cents at risk: +6.45%. One line, no devig, no odds format to translate — and that clarity is the whole reason sharps trade these.

Converting a contract price to American odds and back

You still need the conversion, because the other quote on the same outcome often lives at a sportsbook. Take the contract at $0.62 → 62.0% implied probability. Decimal odds are the reciprocal: 1 / 0.62 = 1.613.

For a side above 50%, American odds are −100 × p / (1 − p):

−100 × 0.62 / 0.38 = −163.2, call it −163

Check it in reverse: 163 / (163 + 100) = 61.98%. Back where we started.

The No side sits at $0.38 → 38.0%. For a side below 50%, American odds are +100 × (1 − p) / p:

+100 × 0.62 / 0.38 = +163.2, call it +163, and 100 / (163 + 100) = 38.02%. Also correct.

Note what happened. The two sides came out at -163 and +163 — perfectly symmetric, summing to exactly 100%. No sportsbook posts that, because a sportsbook has to pay for the building.

Order book vs house pricing

That symmetry is a structural difference, not a promotion.

A sportsbook is the counterparty to your bet. It posts a price, takes the other side, and builds margin in so the implied probabilities sum above 100%. That surplus is the hold, or vig. A market at -180 / +150 implies 64.29% and 40.00%, summing to 104.29% — a 4.29% hold. Devigged proportionally, the favorite's honest price is 64.29 / 104.29 = 61.64%.

A prediction market is an exchange, matching you against another trader rather than the house. No posted margin, so a liquid contract's mid-price reflects where buyers and sellers actually agree — 62.0% here, versus the sportsbook's devigged 61.64% on the same event.

Two consequences. First, you see a bid and an ask, not one number. A contract might show Yes bid 61¢ / Yes ask 62¢. Buying at the ask and later selling at the bid costs the full 1¢ spread — roughly 1.6% of a 62¢ position. On thin contracts that spread is the real cost, and it can exceed any sportsbook's vig.

Second, you can leave a resting order. Post a bid at 61¢ and wait instead of paying 62¢. Sometimes you get filled; sometimes the market runs without you. No sportsbook offers that — there, the posted price is the only price.

Fees vs vig — the same cost, differently packaged

Exchanges charge fees rather than embedding margin. Fees are visible and charged once; vig is invisible, priced into every number you see. Run the same 66% read through both venues.

Prediction market. Buy 100 contracts at 62¢ = $62.00. Say the platform charges $1.70 on the trade — fee schedules vary by platform and contract, so check the one you are on. Total outlay $63.70 for a $100 payoff if Yes.

EV = (0.66 × $100) − $63.70 = +$2.30, or +3.61% on capital at risk. Note what the fee did to break-even: without it you needed 62.0%, with it you need 63.70 / 100 = 63.7%.

Sportsbook. The same outcome at -180 (decimal 1.5556):

EV = (0.66 × 1.5556) − 1 = 1.0267 − 1 = +2.67%

Same read, 3.61% versus 2.67%. The exchange wins here, but by less than "no vig" suggests, and on a wide-spread contract it would lose. Price the fee every time.

What gets listed

Anything with a verifiable outcome and a date:

Sports — championships, season win totals, some game-level and award markets, though depth trails sportsbooks badly
Economics — Fed decisions, CPI and jobs prints, GDP, recession calls. The category where prediction markets are genuinely the best price anywhere
Politics — elections, confirmations, legislative outcomes, domestic and foreign
Entertainment and culture — awards, box office, releases
Climate and technology — temperature records, launch dates, regulatory approvals

The sports overlap with your sportsbook stack matters most here: the same outcome gets priced twice by two different crowds. Lesson #20 walks through trading that gap.

Liquidity risk

A price is only a price if you can transact on it. A headline contract on a major election or Fed meeting may absorb six figures without moving. A niche contract may show a beautiful 62¢ mid with $200 of depth per side and a 5¢ spread — there, the mid is decoration.

Look at the book, not the last price. Check the size resting at the best bid and offer, how far you would walk the book to fill your size, and whether you could exit fairly if your read changes before resolution. Thin contracts are a one-way door.

Resolution risk

The risk newcomers underrate. A binary contract pays on what the resolution criteria say, not on what you think happened. Ambiguous wording, an edge-case outcome, a source that stops publishing — any of these can resolve a contract against a position you thought was obviously correct. Regulated venues run formal dispute processes; crypto-native venues use oracles with dispute windows running days or weeks. Either way, the text governs.

Three defenses:

Read the resolution criteria in full before every trade, including the source of truth and the tie-break language. If it is ambiguous, skip it — there is always another contract.
Prefer high-volume contracts. A misresolution on a widely held market gets contested loudly. One on a $3,000 market may not.
Never size a cross-venue hedge as though the prediction-market leg is certain to settle your way. If it misresolves, your covered position becomes a one-sided one.

How Sharp Markets fits

Sharp Markets is Sharp's prediction-market surface, covering Kalshi and Polymarket. It solves the comparison problem: the two venues state the same probabilities in incompatible notations, so you cannot eyeball which has the better number.

Sharp normalizes both into implied probability and shows them together, turning "is +163 better than 38 cents?" into a subtraction. The rest of the stack applies unchanged — devig calculators for the sportsbook side, your bet log recording the position so CLV still measures whether you beat the close, a price alert when either venue crosses a threshold you set.

Which venue for which question is Lesson #19; trading one against the other is Lesson #20. Carry this out of the primer: price is probability, fees are the cost, the contract text is the contract.