Win Rate Analyzer

The Win Rate Analyzer turns a closed sample of bets into a hit rate and a P&L under a constant stake and average decimal odds. Four inputs: Wins (default 55), Total Bets (default 100), Stake (default $10), Avg Odds (default 1.91). Win Rate = wins/total × 100 = 55.00%. Net Profit = wins × (stake × odds − stake) − (total − wins) × stake. On the defaults, 55 wins pay 55 × (10×1.91 − 10) = 55 × 9.10 = $500.50; 45 losses cost $450; profit = $50.50.

That $50.50 on $1,000 turned over (100 × $10) is a 5.05% yield on stake turnover, not a 55% “I win more than I lose” story. At 1.91 you need more than 1/1.91 ≈ 52.36% to break even. 55% clears that hurdle by 2.64 points, which is why the P&L is positive. Drop Wins to 52: profit = 52×9.10 − 48×10 = 473.20 − 480 = −$6.80. The win rate still looks like a majority; the odds made you a slight loser. That is the whole analyzer.

Stake is flat. If your actual stakes varied, type the average stake and accept a sketch, or do not use this page for P&L. Avg Odds is a straight arithmetic mean of decimals in the default interpretation; a volume-weighted average of odds would be better if big bets sat on short prices. The form does not weight.

Closing-line value is not included. Beating 55 of 100 at 1.91 tells you nothing about whether you beat the close. A square who lucked into 55–45 at bad prices and a sharp who went 53–47 at +EV numbers can print similar $50 here. Pair this with CLV tracking elsewhere.

Total is clamped to at least 1; wins are clamped to ≤ total. 55 wins in 50 bets becomes 50 wins in 50. Read 55.00% and $50.50 together; a win rate without the odds line is a sports-radio statistic.

Win rate is the most abused statistic in betting. Casual players quote “I win 60% of my bets” without the price. A 60% win rate at 1.50 decimal is a disaster (break-even is 66.7%); a 45% win rate at 2.40 can be a profession. SorteCalc forces the odds into the same form as the count so the $50.50 default can contradict the 55% glow if you change the price.

The sample of 100 bets is a teaching size: large enough for a percentage, small enough that binomial noise is huge. Standard error of a 55% rate is √(0.55×0.45/100) ≈ 5.0 percentage points. A 95% interval is roughly 45–65%. The $50.50 profit is one realization, not a Sharpe ratio. Professional syndicates care about closing-line value (CLV) — whether their number was better than the number the market closed at — because CLV predicts; a 100-bet P&L barely does. This analyzer omits CLV on purpose: it is a P&L identity, not a skill model.

History: as soon as betting shops kept ledgers, clerks computed hits / attempts. The conversion to profit requires the payout function. Decimal odds make that payout (o−1) per unit won. American odds would need a branch; this site standardizes on decimal. Thorp and Wilson would still ask for expectation going forward, not the last 100 bets; the last 100 are a random variable with mean n × EV if the process is stationary, which it is not if you shop lines and tilt.

Limits: flat stake, one average price, no pushes, no voids, no bonuses, no commission, no parlay mix (a parlay is one “bet” with a high decimal). Mixing -110 sides with +400 underdogs into “avg odds 1.91” is a mush. If your book is 1.91 across a homogeneous sides sample, the default is fair.

Use it to translate 55/100 at 1.91 into $50.50, to find the break-even win rate 1/o, and to stop quoting win rate alone. The $50.50 is arithmetic on the inputs, not a forecast of the next 100.

A closed sample of bets becomes a hit rate and a profit only after you pair the count with a price. Divide wins by the total number of bets and scale to a percent; that is the hit rate. Profit is then wins times the net payout on each winner, minus losses times the stake. Net payout on a winner at decimal odds is stake times odds minus stake, which is stake times (odds minus one). The teaching defaults are fifty-five wins in one hundred bets, a stake of ten dollars, and average decimal odds 1.91. Hit rate is therefore 55.00 percent. Fifty-five winners pay 55 times 9.10, which is 500.50 dollars; forty-five losers cost 450; profit is 50.50 dollars.

Break-even against a constant decimal price is the reciprocal of the odds. For 1.91 that hurdle is about 52.356 percent. The sample sits 2.644 points above it, which is why the ledger is positive. Drop wins to fifty-two and the same stake and price print 52 times 9.10 minus 48 times 10, namely 473.20 minus 480, a loss of 6.80 dollars. A majority of winners is then still a losing book. Huygens already wrote expectation as probability times gain minus complementary probability times stake; Bernoulli's law of large numbers only says the sample hit rate concentrates if the trials are identical. One hundred bets are not that limit.

Turnover on the defaults is one hundred times ten, or 1,000 dollars. Yield on turnover is 50.50 divided by 1,000, which is 5.05 percent. Yield on a starting bankroll is a different fraction if that bankroll is not 1,000, and this identity never sees a bankroll. Standard error of a 55 percent rate in one hundred Bernoulli trials is the square root of 0.55 times 0.45 over 100, about 4.97 points. A crude ninety-five percent interval therefore spans roughly 45 to 65 percent. The 50.50 dollar profit is one realisation of a random sum whose session standard deviation is hundreds of dollars around a 50 dollar mean.

Closing-line value is omitted on purpose. Beating fifty-five of one hundred at 1.91 says nothing about whether the prices beat the close. Thorp would still ask for the forward expectation, not the last hundred tickets. Assumptions: a flat stake, a single arithmetic-mean decimal, no pushes, no voids, no commission. A push would need wins plus losses plus pushes to equal the total. Commission on winnings would replace (odds minus one) with that quantity times one minus the rate. Mixing short favourites with long underdogs into a mushy average of 1.91 is a measurement error, not a formula error.

Clerks in nineteenth-century shops already computed hits over attempts; converting the count to money requires the payout rule. Decimal odds make that rule linear. American prices would need a branch. Read 55.00 percent and 50.50 dollars together. A hit rate without the odds line is a radio statistic, not a ledger.