Strategies·November 17, 2026·7 min read

Pair Trading Correlated Markets: Championship vs Conference Logic

How to trade logical implications between Polymarket contracts, set a hedge ratio, price rule mismatch and leg risk, and size a relative-value pair by its worst state.

pair tradingrelative valuecorrelated marketspolymarket
Calculator numbers — illustrative image
Photo: Martin Vorel / StockSnap · CC0Source

Two markets on the same team, one for the conference title and one for the championship, look like the same bet at two prices. They are not. The championship contract is a subset of the conference contract, because a team cannot win the title without first winning its conference in the usual playoff structure. The gap between the prices is the market's implied probability of winning the final conditional on getting there. The pair trade is a bet on that conditional, not on the team.

Correlated markets are everywhere on Polymarket. Nominee versus president, division versus conference versus championship, series winner versus championship. Some pairs are linked by a strict implication and some only by a statistical tendency. The first job is to state the link in one sentence. The second is to check whether the two rulebooks enforce it.

The implication trade

Suppose the conference contract trades at 55 cents and the championship at 34. The implied conditional is 34 / 55 = 61.8%. The market says the team would win the final 61.8% of the time it reaches it. If your model of the final says 70%, because the team would be a meaningful favorite, the championship is cheap relative to the conference.

The trade is to buy championship Yes at 34 and buy conference No at 45. The pair costs 79 cents. It pays $1 if the team wins the title, $1 if it fails to win the conference, and nothing if it wins the conference and loses the final. Maximum profit is 21 cents, maximum loss is 79 cents, and the break-even probability of that middle state is 79%.

The pair is short the middle state, so it needs that state to be unlikely. The market's implied probability for the middle is 55 minus 34, or 21%. At a 70% conditional, the middle is 0.55 x 0.30 = 16.5%, and the expected value is 0.835 x 21 - 0.165 x 79 = +4.5 cents per pair on 79 cents, about 5.7%.

Fees change the ranking. At the sports rate of 0.05, 100 championship shares at 34 cents cost 100 x 0.05 x 0.34 x 0.66 = $1.12. The 100 conference No shares at 45 cents cost 100 x 0.05 x 0.45 x 0.55 = $1.24. Total fees are $2.36 per 100 pairs, or 2.36 cents per pair. That is 11.2% of the 21-cent maximum profit and 3% of the capital. The 5.7% edge becomes roughly 2.7% after fees, before slippage.

The pair is not a forecast that the team is good. It is a forecast that the market's conditional is wrong. A trader who likes the team will naturally want to buy both contracts, which is a directional bet with no relative-value content. A 2026 paper on relationship discovery built relation graphs from resolution text and found that linked pairs co-move more than matched controls, which suggests the links are real, not that any pair is mispriced.

Hedge ratios for pairs that are not strict implications

Strict containment gets a 1:1 share ratio because both legs are binary and pay in dollars. A pair like division winner versus championship is containment only if the playoff format guarantees it. A pair like popular vote versus election winner is not containment at all.

For statistical pairs, set the ratio from price changes. Regress the daily change in market A on the daily change in market B over 60 or 90 closes. The slope is the beta. If beta is 1.4, 100 shares of A are hedged with 140 of B for a price-neutral position. The alternative is dollar matching: 100 shares at 20 cents is $20, so hedge with 40 shares at 50 cents. Beta matching targets typical co-movement; dollar matching targets mark-to-market stability for small moves. Neither protects you if the logical link breaks.

Whichever method you use, compute the payoff in every state. A pair can have positive expected value and a worst case larger than the account allows.

Rule mismatch is where the trade lives and dies

The pair pays at least $1 in every state except conference Yes and championship No. That payoff is robust to many rule differences. Rule differences usually change the probability of the losing state, and in rare cases the payoff itself.

Consider a college basketball team. The conference market may resolve on the regular-season title, the conference tournament, or a share of the title. Those are three different events. A team can win the regular season, lose the tournament, and still win the national championship. If the conference contract resolves on the regular season, the middle state is smaller than it looks. If it resolves on the tournament, it is larger.

Other mismatches to check: whether the championship includes a play-in path the conference market does not, whether a vacated result counts, and whether a shortened season triggers a 50/50. The last one is the payoff breaker. If one leg can void while the other does not, the pair no longer pays $1 in the bad states. A 50/50 on the conference leg turns a 100-cent payoff into 50 cents.

Leg risk and execution

Two legs mean two fills, and one can fail. If you buy the championship first and cannot buy the conference No, you own a naked long championship position, which is the directional bet you were trying to avoid. Start with the leg that has the less reliable book, usually the championship Yes, and abandon the trade if the first leg fills and the second does not.

  1. Check the depth within one cent on both legs before sizing.
  2. Use a limit order at or inside the touch, never a market order.
  3. Decide in advance whether a partial pair is acceptable. Usually it is not.
  4. Use FOK for the second leg if you need the full hedge, or FAK if a partial hedge is better than none.
  5. Do not chase the second leg more than a defined number of ticks.

Negative-risk events add a wrinkle. In a multi-outcome event, buying No in one outcome is economically close to buying Yes in every other outcome, and the depth on the two sides can differ. A pair that looks expensive on the No side can sometimes be built more cheaply through the complementary Yes positions. Check both paths before assuming one price is the only price.

Sizing the pair

Size by the worst state, not the notional. If the pair risks 79 cents to make 21, and your rule is to risk 1% of a $10,000 bankroll per idea, the maximum is $100 / 0.79 = 126 pairs. That is $9,954 of capital, which may be most of the account. Capital is often the binding constraint.

Then check liquidity. A 126-pair order is 126 shares on each leg, which may be several times the visible depth in a thin championship market. The NBA order-book study found that 76.9% of combinatorial opportunities were capped at an average executable size of 14.8 shares. The lesson transfers: the size that clears at the quoted price is the only size that matters.

Finally, check correlation with the rest of the book. If you already own the team's moneyline or a future, the pair adds exposure to the same roster, the same injury risk and the same playoff path.

How to execute a pair

  1. Write the implication in one sentence. If you cannot, it is not a containment trade.
  2. Check both rulebooks for definition, source and void conditions.
  3. Compute the implied conditional from executable prices, not midpoints.
  4. Compare it with your own conditional and estimate the middle state.
  5. Compute the payoff in all three states, including fees and slippage.
  6. Size by the worst-state loss and the depth of the thinner leg.
  7. Enter the hard leg first and set a maximum chase.
  8. Cap exposure per team and per rule template.

Risks

  • The conditional is your estimate. If the market's 61.8% is right and yours is 70%, the pair loses on average.
  • Rule mismatch can raise the probability of the middle state, which is what the pair is short.
  • A void or 50/50 can change the payoff, not just the probability.
  • Leg risk can leave a naked directional position.
  • Thin depth and a 5% sports fee can turn a positive edge negative.

Checklist

  • Is the implication strict, or only statistical?
  • Do both contracts use the same event definition and source?
  • What is the implied conditional, and what is mine?
  • What does the pair pay in each state?
  • Does it survive both taker fees and slippage?
  • Is the size within the depth of the thinner leg?
  • What happens if one leg voids or resolves 50/50?
  • Is this pair already in my book through another position?

Sources

  1. 1.Negative Risk Markets— Polymarket Documentation
  2. 2.Markets & Events— Polymarket Documentation
  3. 3.Fees— Polymarket Documentation
  4. 4.Agentic AI for Clustering and Relationship Discovery in Prediction Markets— arXiv
  5. 5.Arbitrage Analysis in Polymarket NBA Markets— arXiv
  6. 6.Pairs Trading: Performance of a Relative Value Arbitrage Rule— National Bureau of Economic Research

More from the journal

All articles