How to Track Your Betting Results Honestly
Win rate, ROI and units measure three different things, and quoting whichever one looks best is how most betting records get inflated without anybody technically lying. Here is what each actually says and what to write down.
9 min read
The short answer
- Win rate alone is meaningless without the prices it was earned at.
- ROI on turnover — profit divided by total staked — is the honest headline figure.
- Return on bankroll flatters a record by shrinking the denominator; say which you mean.
- Units make records comparable across bankrolls, which is why every serious record uses them.
- Log the price you took and the closing price on every bet, not just the result.
- Record pushes separately: they are not wins, not losses, and they distort a win rate if counted as either.
Three numbers that get confused
A record can be summarised three ways and they answer different questions.
| Measure | What it is | What it answers |
|---|---|---|
| Win rate | Wins ÷ graded bets | How often you are right |
| ROI on turnover | Profit ÷ total staked | What each dollar risked returns |
| Units | Profit ÷ one standard stake | Size-independent profit |
Win rate is the one that misleads. It says nothing on its own, because it has to be read against the prices behind it. 55% at −110 is a strong edge; 55% on −140 favourites is losing money, since that price needs 58.33%. Any record quoted as a bare percentage is missing the half that determines whether it was profitable.
ROI on turnover is the honest headline. It folds the prices in automatically — profit against everything you risked to earn it — and it is comparable across bettors, sports and stake sizes. The ROI calculator computes it from a record.
Units are how a record travels.“Up 32 units” means the same thing to someone betting $10 and someone betting $1,000, which is why every credible public record — including this site’s — reports in them.
The ROI denominator trap
“ROI” gets quoted against two completely different denominators, and the gap between them is enormous.
ROI on turnover = profit / total staked ROI on bankroll = profit / starting bankroll $2,500 profit from 500 bets of $100, on a $5,000 bankroll on turnover: 2,500 / 50,000 = 5% on bankroll: 2,500 / 5,000 = 50%
Both are legitimate figures. Only one of them is what people assume when they read “ROI”.
Same season, same profit, and one number is ten times the other. A 5% ROI on turnover is an excellent long-run result. A “50% ROI” sounds like something else entirely, and it is the version that tends to appear in marketing.
What to log for every bet
The result is the least useful column. These are the ones that let you diagnose anything later.
- Date, sport, league, market — so you can slice the record and see where the edge actually lives.
- The price you took, and the sportsbook. Without it your win rate cannot be judged against anything.
- The closing price. The single most valuable column in the sheet — it is what makes closing line value computable, and CLV reads years faster than results do.
- Stake, in units and in currency. Both, because units make it comparable and currency makes it real.
- Result, including pushes and half-settlements — Asian lines settle as half wins and half losses, and a sheet with only win/loss columns cannot record them.
- Bankroll after settlement, so drawdowns are visible rather than inferred.
The record calculator turns a tally into win rate, ROI and profit, and the units calculator converts between units and currency in both directions.
Pushes, voids and half-settlements
A push returns the stake. It is not a win and it is not a loss, and counting it as either distorts the win rate — as a loss it understates you, as a win it overstates you, and either way the number stops being comparable to a break-even rate.
The convention that works: exclude pushes from the win rate and report them separately as a count. Profit and ROI are unaffected either way, since a push contributes zero to both.
Half wins and half losses from Asian lines need a sheet that can hold fractional results. Rounding a half loss to a loss is a quiet, systematic error that accumulates over a season.
Reading your own record without fooling yourself
Three habits that separate a record that teaches you something from one that just makes you feel a way.
Put a confidence interval on it. A 55-45 record is consistent with a true win rate anywhere from about 45% to 64%. Until the interval clears your break-even rate, the record is not evidence — the confidence calculator shows where you actually are, and the sample-size guide explains why the answer takes so long to arrive.
Do not slice until you have volume. Splitting a 300-bet record by sport, market and day of week produces a dozen sub-samples of twenty bets each, one of which will look spectacular by chance. That is not a discovery, it is arithmetic — and acting on it is how a real edge gets replaced with a pattern in noise.
Compare against the expectation, not against zero. The expected profit calculator shows what your edge should have produced over the bets you actually made, and how wide the plausible range around it is. Being below the average is normal. Being outside the range is information.
Common questions
- Is win rate or ROI the better measure?
- ROI on turnover, because it accounts for the prices automatically. A win rate has to be read against the odds behind it — 55% is a strong edge at −110 and a losing operation at −140 — so it means nothing quoted on its own.
- What is a good ROI in sports betting?
- On turnover, a sustained 2–5% is a strong long-run result. Be careful which denominator is being used: $2,500 profit from 500 bets of $100 is 5% on turnover and 50% on a $5,000 bankroll, and only the first is what “ROI” normally means.
- How should I count pushes in my record?
- Exclude them from the win rate and report the count separately. A push returns the stake, so it contributes nothing to profit or ROI — but counting it as a win or a loss distorts a win rate you want to compare against a break-even rate.
- What should I log for every bet?
- Date, sport, market, the price you took, the closing price, stake in units and currency, the result including pushes and half-settlements, and the bankroll afterwards. The closing price is the most valuable column — it makes CLV computable.
- Why should I not break my record down by sport?
- Not until you have real volume. Splitting a few hundred bets into a dozen small sub-samples guarantees that one looks excellent by chance. Acting on that replaces a real edge with a pattern in noise.