Player Props vs Game Lines: Trading the Correlation on Polymarket
How player props embed game-script assumptions, how to price a prop against the moneyline, and what liquidity, void rules and stat sources do to the trade.

A player prop is not an independent market. It is a bet on a game script. A quarterback's passing yards, a running back's carries, a receiver's targets and a basketball star's points all depend on the same things that decide the moneyline: pace, margin, minutes and how long the game stays close. Trading a prop against the game line means trading that dependence, and the first step is to write down the script each contract assumes.
The two markets can disagree because they are priced by different people at different times. The moneyline is usually deep and fast. The prop is usually thin and slow. That is where a relative-value trade can exist, and it is also why execution is the hard part.
Props embed a game script
A few examples make the dependence concrete.
- Passing yards over: assumes the team keeps throwing, which usually means trailing or playing in a shootout. A big lead means fewer attempts in the fourth quarter.
- Rushing yards over: assumes the team protects a lead and runs out the clock. A trailing team abandons the run.
- A basketball star's points over: assumes minutes. A blowout cuts the fourth quarter, and foul trouble cuts the first half.
- A pitcher's strikeouts over: assumes innings and a lineup that chases. A rain delay or an early hook ends the bet early.
- A soccer player's shots on target over: assumes the team is chasing the game or dominating territory. A team protecting a one-goal lead may take fewer risks.
The prop's fair probability is a weighted average over game states. P(over) equals the sum over states of P(state) multiplied by P(over given that state). The moneyline prices the distribution of states, or at least the win and loss states. If you price the prop with a single season average and ignore the state distribution, you will be wrong in exactly the games where the prop is most likely to be mispriced.
A worked example
An NBA star's points over 24.5 is offered at 58 cents. The team moneyline is 62 cents. Your model says the star averages 26 points, so the over looks cheap. The average hides the script. In close games he plays 36 minutes and goes over 70% of the time. In blowouts he sits the fourth quarter and goes over 30% of the time. If 60% of games are close, the fair probability is 0.6 x 0.70 + 0.4 x 0.30 = 0.54. At 58 cents the over is rich, not cheap, and the correct side is under at 42.
The fee on 100 under shares at 42 cents is 100 x 0.05 x 0.42 x 0.58 = $1.22. Buying 100 shares costs $42 plus $1.22, and pays $100 if he stays under. The break-even probability is 43.22%, so a 46% fair probability, one minus 0.54, leaves about 2.8 points of edge, or 6.4% of capital. That is a real trade, but it depends on the state weights, not on the season average.
The same arithmetic works in reverse for a favorite's passing prop. The market may price the quarterback's yards as if the game will be close, while the moneyline says a blowout is likely. The prop is then too high, and the under is the trade. The disagreement between the two markets is the signal, and the moneyline is usually the more reliable of the two prices.
Correlated positions and hedging
A book of props on the same game is one view. Over on the star's points, over on the team total, over on the moneyline and over on the game total are four expressions of a high-scoring, competitive game. If the game turns into a blowout, three of them can lose together. Treat the group as a single position with a single maximum loss.
Hedging a prop with the moneyline is possible but imprecise. If you bought the over and the risk is a blowout, the natural hedge is the favorite's moneyline or the under on the opposing team's total, not a one-to-one short of the prop. The hedge ratio comes from the conditional probabilities. How much does the prop price move per cent of moneyline move? A regression of prop price changes on moneyline price changes over a season gives a beta. If beta is 0.6, a $1,000 prop position needs about $600 of moneyline exposure to neutralize the game-script component, and it leaves the player-specific risk unhedged.
Hedging also costs. Each leg pays a taker fee and a spread, and the hedge can turn a positive-EV trade into a negative one. Hedge only the risk you cannot underwrite, not every fluctuation.
Mispricing between props and moneylines
The two markets are linked, but the link is not enforced by an arbitrage that a retail trader can execute. The NBA order-book study found that combinatorial opportunities do exist, mostly in the final minutes, with a median return of about 101 basis points. It also found that 76.9% of them were limited to an average executable size of 14.8 shares. At a 50-cent price that is cents of profit per episode, before fees. The fee alone can exceed it.
Single-market anomalies were even rarer in that study. Only 7 executable in-game episodes appeared across 173 games, with a median duration of 3.6 seconds. A prop that is 3 cents off the model is not automatically an opportunity. It is an opportunity only if you can fill it before the market corrects and at a size that covers the cost.
The realistic edge is not a pure arbitrage. It is a pricing disagreement: your model of the game script is better than the prop market's. That is a research edge, and it decays as the market matures and as more traders build the same state models.
Execution constraints
Props are thinner than moneylines. The book may hold a few hundred shares at the touch, and the spread can be three to five cents. Some sports markets use 0.25-cent ticks and minimum order sizes that are larger than the displayed depth. Pre-game orders are cancelled automatically at the official start time, which matters if you plan to trade the prop after lineup news.
Injury and lineup news is the biggest execution risk. A late scratch can void a prop or change its subject. Official injury reports are published before games, and outlets such as ESPN track them, but the timing varies by league and a coach's pre-game comment can move a prop before the report updates. If the rule says the prop voids when the player does not play, the contract is not a pure bet on performance. It is a bet on participation plus performance.
The resolution source matters for the same reason. A prop that resolves on the official box score can differ from one that resolves on a broadcaster's stat feed if the two disagree about a late scoring change. Read the rule before the game, not after.
How to execute
- Price the game first. Build a distribution over margin, pace and minutes from the moneyline and the total.
- Price the prop conditional on that distribution, not on the season average.
- Compare the result with the executable ask and add the taker fee at your size.
- Check the prop's depth within one cent and size to that, not to the moneyline's depth.
- Post the prop leg as a maker if you can. The spread is the largest cost in a thin market.
- If hedging, compute the beta and enter the hedge after the prop fills, not before.
- Confirm the void rule and the stat source before the game starts.
Risks
- Correlation blowup. A blowout or an early injury can hit several props at once.
- Role risk. A trade, a benching or a coaching change can alter minutes and usage.
- Injury and void rules. A non-starter may void the market or resolve it No, depending on the wording.
- Stat corrections. The official box score can differ from the broadcast.
- Liquidity. The exit can be far worse than the entry, especially after a slow start.
- Model risk. Your game-script weights are estimates, and the market's may be better.
- Time decay. A prop can be right about the player and wrong about the game, and there is no way to hold it past the final whistle.
Checklist
- What game script does this prop need?
- Does the moneyline agree with that script?
- What is the conditional probability in each state?
- What is the fee at my price and size?
- How deep is the prop book within one cent?
- What happens if the player does not start?
- Is this position already in my book through another prop?
- Can I exit at a reasonable price if the script changes?
Sources
- 1.Markets & Events— Polymarket Documentation
- 2.Resolution— Polymarket Documentation
- 3.Sports Channel— Polymarket Documentation
- 4.Arbitrage Analysis in Polymarket NBA Markets— arXiv
- 5.The Anatomy of a Decentralized Prediction Market: Microstructure Evidence from the Polymarket Order Book— arXiv
- 6.NBA Injuries— ESPN