--- title: "Over or Under? 82,080 Settled Polymarket Game Totals and 45,662 Large Bets, Checked Against the Result" canonical: "https://0xinsider.com/research/over-under-polymarket-sports" datePublished: "2026-09-13" dateModified: "2026-09-13" description: "The Under settled 52.7% of 82,080 Polymarket game totals, and the price already knew: Under buyers paid 58.4 cents and won 57.9%, Over buyers paid 56.4 and won 55.8%. Neither side beat its price, in any sport, line or phase." --- # Over or Under? 82,080 Settled Polymarket Game Totals and 45,662 Large Bets, Checked Against the Result Last updated: 2026-09-13 Canonical: https://0xinsider.com/research/over-under-polymarket-sports Every sportsbook bettor has heard that the public bets the over and the sharps bet the under. Polymarket lists a total for every game with a fixed half-point line and a price per side, and every large buy is public, so the claim can be checked. Across 82,080 settled game totals since April 2, 2026, the Under hit 52.7% of the time, and more in soccer and esports. The price already knew. Across 45,662 large buys, Under buyers paid 58.4 cents and won 57.9%; Over buyers paid 56.4 and won 55.8%. Neither side beat its price, and no sport, soccer line, price band, phase or grade cohort has an interval that clears zero. The graded wallets lean Under, 60.5% of their totals buys against 50.7% for the poorly graded ones, and that lean is the one number in the study worth watching. ## Over or under is a question about the line A total is a bet on how many points, goals, runs or rounds a game produces, against a number the market sets. At a sportsbook the number moves and the price stays near -110 on both sides, and the folklore is old: the public bets the over because rooting for points is more fun, the book shades the total up to meet them, and the disciplined bettor takes the under. There are studies of it and there are counter-studies, and all of them read the book's side of the counter. On Polymarket a game total is a market. The line is fixed in the market itself, always a half point, and the two outcomes, Over and Under, are priced in cents that sum to about a dollar. Buying Over at 56 cents pays a dollar a share if the game clears the number. There is no line to shade because there is no line to move, only a price. So the question splits into two that can be answered separately: how often does the Under actually settle, and did the price know? 0xinsider holds both halves. Every settled totals market since April 2, 2026 is on record with its result, and every large buy on one is on record with its price and its buyer. This study reads both. ## 82,080 settled game totals and 45,662 large buys Two universes. The markets: every Polymarket game total, a market whose type is totals with outcomes Over and Under, with a kickoff between April 2 to September 13, 2026 and a settled result, counted once, whether or not anyone bought it large. 82,080 markets. Every one of the 80,848 with a recorded line carried a half point, so no total in the sample could push. The buys: every buy of $1,000 or more that 0xinsider recorded on one of those markets, settled after the trade, at 2 to 98 cents. 45,662 buys from 3,064 wallets across 7,120 markets, $810.0 million in stakes. Overs were 48% of the buys and 49.1% of the dollars. The window starts on April 2 for the reason the calibration study at 0xinsider.com/research/favorite-longshot-bias-polymarket-sports documents. Edge is the share of buys that won minus the average price paid, in percentage points; the dollar return is stake-weighted and held to settlement, before fees. Intervals on the buy-level edges resample markets, because the buys on one game are one bet in disguise. Intervals on the settled shares are plain binomial half-widths. ## The Under settles 52.7% of the time Of the 82,080 settled game totals, 52.7% went Under, with a half-width of 0.34 points. That is a real lean, established to a third of a point, and it is the folklore's premise confirmed on a market with no bookmaker: the numbers Polymarket lists sit a little above where the games land. By sport the lean has a direction and a size. Soccer, 49,291 markets, went Under 54.3% of the time. Esports, 8,766 markets, 55.3%. Baseball, 14,519 markets, 49.8%, a coin flip. Hockey, 889 markets, 49.8%. College football, 4,500 markets, went Under 46.1% of the time, so the Over hit more often there. The NBA's 1,388 totals went Under 44.7%. MMA's 942 round totals went Under 37.8%: fights went past the listed number more often than not. None of these is a betting edge on its own, for the reason the next section measures. A market that lists Unders that hit 54% of the time and prices them at 54 cents is a market with a lean and no mispricing. The share tells you which way the numbers sit. The price tells you whether anyone is paying for it. ## The price already knows Under buyers paid 58.4 cents on average and won 57.9% of the time: -0.5 points, with a market-resampled interval from -3.17 to plus 2.23. Over buyers paid 56.4 cents and won 55.8%: -0.61 points, interval -3.37 to plus 2.21. Both sides sit a fraction of a point below their price, which is what crossing a one-cent spread and paying a fee looks like, and neither interval comes near excluding zero. So the two-cent gap in what buyers paid, 58.4 for the Under against 56.4 for the Over, is the settled lean already in the price. Buyers of the Under were buying the side that hits more often and paying more for it, in about the right proportion. There is no free Under on Polymarket, and there is no Over tax either. The dollar returns differ more than the edges do, 1.1% for the Under against -2.74% for the Over. Dollar return is stake-weighted, so a few large losing Overs move it; the point edge weights every buy the same. Read the edge as the finding and the dollar figure as what it felt like. ## Sport by sport, neither side clears zero Soccer is 25,307 of the buys and its Over is the closest thing to the folklore in the data: Over buyers paid 60.1 cents and won 58.0%, minus 2.11 points, with a dollar return of minus 7.52%. The interval runs from minus 6.49 to plus 2.36. Soccer Under buyers paid 62.8 and won 63.0%, plus 0.14. The pattern is in the direction the folklore predicts and the interval does not confirm it. Baseball runs the other way: Over buyers plus 1.03 points and plus 7.06% on the dollar, Under buyers minus 2.30 and minus 5.09%, both intervals across zero. The NBA's Over came in at plus 2.75 points on 4,078 buys, interval minus 3.25 to plus 8.79. Hockey's 1,368 buys are too few to say anything and its intervals span eighteen points. The wide intervals are the point, and they come from clustering. Totals buys pile onto a few markets: the 21,938 Over buys sit on 5,130 markets, an average of four large buys per game, and the big games carry dozens. Resampling markets rather than buys is the honest interval, and it says that a two-point lean in soccer over 2,030 markets is inside the noise of one summer. ## Soccer, line by line A soccer total is mostly 2.5 goals, and the price of Over 2.5 tells you what the market thinks of the two teams. Over 2.5 buyers paid 54.6 cents and won 53.5%, minus 1.08 points; Under 2.5 buyers paid 54.2 and won 54.8%, plus 0.51. At 1.5 goals, a line reserved for games the market expects to be tight, Over buyers paid 73.8 and won 71.3%, minus 2.44. At 3.5, Over buyers paid 51.0 cents and won 46.1% of the time: minus 4.92 points and minus 20.81% on the dollar, across 2,168 buys. That last number is the one a bettor remembers, and its interval runs from minus 15.97 to plus 6.94 over 415 markets. Over 3.5 is a bet that a game produces four goals, most large buyers of it in this window were wrong, and the sample cannot say whether that is a mispricing or a bad summer for high-scoring games. Under 3.5 buyers, at 66.1 cents, won 65.2%, on the price. The rest of the lines are calibrated in the way a thin sample allows: Over 0.5 at 87.9 cents won 84.2%, Under 4.5 at 78.3 won 77.7%, Under 5.5 at 86.2 won 86.2%. The market knows what a 4.5-goal game costs. ## Cheap sides and expensive sides Split by the price paid and the two sides track each other. Between 40 and 60 cents, where two thirds of the buys sit, Over buyers came in at minus 0.12 points and Under buyers at minus 0.07. Between 60 and 80 cents both sides lost two to three points to the price, minus 2.86 for Overs and minus 2.18 for Unders. At 80 cents and up the Over lost 1.43 and the Under 0.02. Under 40 cents, 2,202 buys, the Over came in at plus 0.91 points and the Under at minus 1.49, and both returned badly on the dollar, minus 6.82% and minus 15.92%, because the cheap side that lost was bought bigger than the cheap side that won. The favorite-longshot study found the same shape on every sports market: the cheap contract is where the dollar return and the edge come apart. ## Before kickoff and during the game Before kickoff, Over buyers paid 53.9 cents and came in at minus 0.86 points; Under buyers paid 54.2 and came in at minus 0.99. Both sides were near a coin flip and both lost about a point to the price, with intervals from minus 4 to plus 2. In-play the prices move away from the middle, because time passes and goals either come or do not. In-play Over buyers paid 62.0 cents and won 62.0%, minus 0.04. In-play Under buyers paid 64.5 and won 64.7%, plus 0.22, and they were 59% of the in-play buys. A live Under is a bet that the game stays where it is, and the market priced it to the tenth of a point. The timing study at 0xinsider.com/research/when-large-sports-bets-land found the in-play price fairer to the buyer than the pre-game price across every market type. Totals are no exception. The one thing the phase split adds is that the lean toward Unders is a live phenomenon: before kickoff the buys split 52 to 48 for the Under, and in-play 59 to 41. ## Graded wallets lean Under From June 1, when grade coverage widened, each buy carries the grade its wallet held that day. Wallets graded S, A or B put 60.5% of their totals buys on the Under. Wallets graded D or F put 50.7% on it. The graded cohort leans the way the settled shares lean; the poorly graded cohort splits the two sides evenly. On the Under, S, A and B wallets paid 63.5 cents and won 65.7%: plus 2.12 points and plus 8.15% on the dollar across 4,640 buys, interval -2.78 to plus 6.91. On the Over they came in at -1.54 points and -18.06% on the dollar, the dollar figure driven by a handful of large losing buys. D and F wallets lost on both sides, -4.1 points on Overs and -3.18 on Unders, with intervals across zero. This is consistent with the sharp money study at 0xinsider.com/research/sharp-money-polymarket-sports, which found totals the market type where the poorly graded cohort did worst, and with the fade study at 0xinsider.com/research/fading-the-crowd-polymarket-sports, which found totals the market type where a one-sided crowd lost most. The graded wallets' Under edge here has an interval that includes zero. Their preference for the Under does not need an interval: it is a count, and it is ten points wide. ## Month by month, the sign flips April, the largest month with 16,963 buys, was calibrated on both sides: Over plus 0.15, Under minus 0.17. May, Over plus 2.99 and Under minus 0.82. June, the second-largest month, Over minus 3.19 and Under minus 0.16. July, Over plus 0.22 and Under minus 2.53. August and September are a few hundred buys each after the alert floor rose to $10,000. A bias that is real shows the same sign every month. This one changes sign three times in four months of real size. The soccer Over's minus 2.11 for the window is mostly June, and June is mostly the World Cup, and a tournament is one draw from the distribution of summers. Nothing in the month rows says the Over is systematically overpriced. ## Corners, team totals and first halves Polymarket lists other Over/Under markets on the same games, and the study reads the ones with 300 or more large buys on each side, without intervals because the samples are too small to resample. First-half totals: Over plus 1.99 points, Under minus 3.82, on 467 and 652 buys. Soccer team totals, one team's goals: Over plus 2.85, Under plus 0.80, with dollar returns of plus 27% and minus 19% that a single game could produce. Total corners: Over minus 8.69 and Under plus 9.38, on 344 and 364 buys. The corners number is the one that would be a finding if it held. Nine points on a few hundred buys across a few dozen markets is a number to re-check on a larger sample, and the study records it so that a rerun has something to compare. It is not evidence yet. ## What this study does not show It does not compare Polymarket's lines with a sportsbook's. The claim that books shade totals toward the Over is about books, and this data has no book in it. What it can say is that on a market with no book the Under still hits more often than not and the price reflects it. The buy-level intervals are wide because totals buys cluster on popular games. A two-point lean that a bettor might feel over a season is inside those intervals. A study of every totals trade rather than the $1,000-and-up buys would have narrower intervals and the same clustering problem. The settled share is a fact about lines and results, not about bettors. Polymarket's totals lines come from the market it lists; how they are set is outside this data. And the alert floor rose from $1,000 to $10,000 in August, so August and September are a few hundred buys of the largest bets. Every position is scored as held to settlement at the buyer's own price, before fees. A wallet that sold out of a losing Over at 30 cents is scored as if it held to zero. ## What it means for a sports bettor There is no free Under on Polymarket. The folklore's premise is true here, the Under hits more often, and the folklore's conclusion is false here, because the price is set by bettors who know it. A bettor who takes every Under at 58 cents is paying 58 cents for a 58% chance. The edge on a total, where there is one, comes from the same places it comes from on a moneyline. The counterparty: the fade study found the poorly graded crowd loses most on totals, minus 4.85 points when it agrees on a side before kickoff. The timing: in-play totals were priced to the tenth of a point, and the pre-game price was a point worse for the buyer. And the number itself: a bettor who can say why a game produces four goals has a bet on Over 3.5 that the market, in this window, priced at 51 cents for a 46% outcome. The graded wallets' Under lean is the one habit in this study worth copying as a default. They buy the side that hits more often, six times in ten, and in this window they were paid for it on the Under and not on the Over. That is a lean, not a rule, and its interval says so. ## How to use it on 0xinsider Every game on a 0xinsider sports board, 0xinsider.com/sports/soccer, 0xinsider.com/sports/mlb, 0xinsider.com/sports/nba, 0xinsider.com/sports/esports and the rest, lists its total beside the moneyline and the spread, with the Over and Under prices in cents. Open the total and its market page lists the wallets holding each side with the grade each has earned, so the cohort split in this study is visible for tonight's game. The whale alerts feed at 0xinsider.com/whale-alerts shows each large buy as it lands with the buyer's grade, and a filter on the sport shows the totals buys among them. A run of D and F buys on the Over of one game before kickoff is the fade study's pattern on the market type where it was widest. The guide at 0xinsider.com/learn/over-under-betting-explained covers the bet itself, and 0xinsider.com/learn/how-to-find-sharp-money-polymarket-sports walks through the surfaces in order. ## How to check it yourself Every number on this page comes from one read-only run of a set of queries against the production database at 16:51 UTC on September 13, 2026. The query text, the raw output, the market-level export and the bootstrap output are public at github.com/0xinsider/research, under polymarket-sports-markets, in the files whose names carry the words over-under. The export is committed, so the intervals reproduce with the bootstrap alone and no database. The buys universe is the calibration study's universe restricted to markets of type totals with Over and Under outcomes. The markets universe is every such market with a kickoff in the window and a settled outcome. Outcome 0 is Over and outcome 1 is Under on these markets; the settled share is the share with outcome 1. Expect your counts to grow as open markets settle. The settled share should hold within its half-width; the buy-level edges should stay inside their intervals, which is a low bar, and a rerun that puts the soccer Over outside its interval is a finding this study did not make. ## FAQ ### Does the under hit more often than the over on Polymarket? Yes. Of 82,080 Polymarket game totals that settled with a kickoff between April 2 to September 13, 2026, 52.7% went Under, with a binomial half-width of 0.34 points. Soccer went Under 54.3% of the time and esports 55.3%; baseball and hockey were coin flips; the NBA, college football and MMA totals went Over more often than not. ### Is betting the under profitable on Polymarket? No more than betting the over. Under buyers paid 58.4 cents on average and won 57.9% of their buys, -0.5 points against the price; Over buyers paid 56.4 and won 55.8%, -0.61 points. The Under's higher hit rate is already in its higher price. ### Do Polymarket bettors overpay for the over? Not measurably. Soccer Over buyers came in at minus 2.11 points with a market-resampled interval from minus 6.49 to plus 2.36; baseball and NBA Over buyers came in above zero; and the Over's sign flipped three times across April to July. No sport, price band, phase or soccer line has an Over interval that excludes zero. ### Can a Polymarket total push? Not in this sample. Every one of the 80,848 settled game totals with a recorded line carried a half-point line, so each market settled Over or Under with no push. Markets without a recorded line settle the same way; the study excludes nothing on that basis. ### Which side do the sharp wallets bet on totals? The Under. From June 1, 2026, wallets 0xinsider grades S, A or B placed 60.5% of their game-total buys on the Under, against 50.7% for D and F wallets. On those Unders the graded cohort came in at plus 2.12 points with an interval from -2.78 to plus 6.91; the poorly graded cohort lost on both sides.