Cheap tail outcomes near resolution: no executable edge found
Football match outcomes priced 3-6c won 4-5% of the time, an expected value near zero to -20%. A first run showing +126% came from timing and selection errors; priced from real trades, the 0-0 tail returned -11% to -15%.
What we asked
A "tail" is the cheap side of a market shortly before it resolves, priced between 1 and 15 cents. Buying it is a lottery ticket: it pays $1 if the outcome happens. The question was whether such tails win more often than their price implies. At a price of 5 cents, a buyer needs the outcome to happen more than 5% of the time to break even, before fees and spread.
This was a paper measurement on historical data, not live trading. Expected value (EV) below means payout divided by price, minus 1.
Data
-
Main sample. 12,561 closed markets with at least $10,000 of volume, from 1 September to 13 October 2026. The selection was the top 300 markets by volume per day. Prices are the CLOB price history of the YES token. For each market we took the cheaper side at snapshots 24, 6 and 1 hours before the reference time.
-
Sports sample without a volume threshold. 75,592 Over/Under and Spread markets, taken in 6-hour windows, of which 8,296 were "Over/Under 0.5" markets.
-
Trade tape. The public trade feed for 500 football Over/Under 0.5 markets, used to see where buyers actually traded.
Result
Our first run reported that sports tails priced around 5 cents won 12% of the time, an EV of +126%. Two errors produced it, described below. After correcting them:
| Sample | Win rate | Price | Result |
|---|---|---|---|
| Match outcomes (win or draw), snapshot from the scheduled start | 4-5% | 3-6c | EV about 0 to -20% with a 1c spread |
| Crypto and other markets | not recorded | not recorded | EV from -100% to +13% at +1c, noise |
| Football Under 0.5, unselected pool of 2,392 markets, one hour before kickoff | 9.7% (0-0) | 6.8c | +43% at the mid price |
| Same markets, where Under was actually bought in the 6 hours before (309 of 500) | 5.5% (0-0) | 6.3-6.5c | EV -11% to -15% |
| Same sample, no Under purchases at all (191 of 500) | 11.5% (0-0) | 7.8c mid | nobody to buy from |
The mid price is halfway between the best bid and best ask. It is not a price at which you can trade.
Why the edge disappears
There were two timing and selection errors in the first run. First, for esports markets the recorded end date sat 6 hours after the start of the match, so our "pre-event" snapshot landed inside the game. Second, most of the wins were Under 0.5 outcomes, meaning a 0-0 result.
A third effect is selection by volume. Among Under 0.5 markets, 0-0 finished 1.9% of the time when lifetime volume was under $1,000, 6.4% between $1,000 and $10,000, and 15.9% at $10,000 or more. Volume accumulates during the match while the score stays 0-0, so filtering for high final volume selects the markets that ended 0-0. With the $10,000 threshold, 0-0 appeared in 16-22% of markets priced 4.5-7.5c.
In the unselected pool the 0-0 rate was 9.7%. Across 12 leagues in a football-data reference set, the base rate of 0-0 was 5.4%. The top leagues (La Liga, Champions League, Ligue 1, English Championship, Eredivisie, Bundesliga) showed about zero or negative EV. The positive results came from women's, youth and lower leagues, with volume of $400-10,000.
The trade tape explains that last group. Where buyers existed, 0-0 happened 5.5% of the time at purchase prices of 6.3-6.5c. The apparent edge sat in 191 markets where no one had bought Under, so the cheap mid price was quoted in a book with nothing to buy.
Limits of this measurement
-
The sample is a six-week window, 1 September to 13 October 2026. We did not record a confidence interval for the individual rates.
-
Prices come from price history and trade tape, not from a recorded order book, so the exact spread at each entry is not known.
-
The trade-tape check covers 500 football Over/Under 0.5 markets only. Other sports and market types were examined less closely.
-
An earlier check of executable prices found 5.9% wins against the 10% needed. We kept no further detail on that check.
What it means for a trader
We found no executable edge in cheap tails. Match-outcome tails appear fairly priced. Apparent edges in football Under 0.5 markets came from quoting a mid price in a book with no sellers, and from filtering markets by volume that accumulated after the outcome was in doubt. A backtest on tail outcomes should use prices someone actually traded at, and should select markets by what was known at the entry time.
Sources
- sourcePolymarket Gamma API
- sourcePolymarket CLOB API
- sourcePolymarket Data API
Polytrading is an analytics tool. It does not trade for you, hold your keys, or promise returns.