Betting ROI is your profit divided by the total amount you've staked, times 100. If you've staked $5,000 across 250 bets and you're $150 up, your ROI is 3%. Most bet trackers call the same number yield. The harder part is knowing when to believe it: at ordinary head-to-head odds, luck alone can produce a healthy-looking ROI over a few hundred bets, and it usually takes thousands before the number tells you much about skill.
This guide covers how to calculate it, the ways it gets inflated, and a table showing how far luck can move it at each sample size and price.
How to calculate betting ROI
The formula, as Covers sets it out, is:
ROI = (profit ÷ total amount staked) × 100
Here's a short invented run of five bets at $20 each:
| Bet | Odds | Stake | Result | Profit or loss |
|---|---|---|---|---|
| 1 | 1.91 | $20 | Won | +$18.20 |
| 2 | 2.10 | $20 | Lost | -$20.00 |
| 3 | 1.85 | $20 | Won | +$17.00 |
| 4 | 3.40 | $20 | Lost | -$20.00 |
| 5 | 1.95 | $20 | Won | +$19.00 |
| Total | $100 | +$14.20 |
ROI is $14.20 ÷ $100 × 100 = 14.2%. It looks great, and it means almost nothing yet. Five bets is a coin toss with extra steps. More on that below.
ROI, yield and return on bankroll are different numbers
Three numbers get called "return", and mixing them up is the most common way results get oversold.
- Yield (ROI on turnover): profit ÷ total staked. This is what the formula above gives you, and what most trackers show.
- Return on bankroll: profit ÷ the bankroll you started with.
- Units won: profit ÷ your standard stake.
Say you start with a $1,000 bankroll and place 500 bets of $20. That's $10,000 of turnover. Finish $300 up and your yield is 3%, your return on bankroll is 30%, and you're up 15 units. Same results, three very different-sounding numbers. When anyone quotes a return, ask which one it is and over how many bets.
Strike rate on its own tells you nothing
Winning 55% of your bets sounds good until you check the odds. The win rate you need to break even is 1 ÷ odds (see our implied probability guide):
| Average odds | Break-even strike rate |
|---|---|
| 1.70 | 58.8% |
| 1.91 | 52.4% |
| 2.50 | 40.0% |
| 4.00 | 25.0% |
A 55% strike rate at 1.70 returns 0.55 × 1.70 = 0.935 per dollar staked, which is a 6.5% loss. Yield already includes the odds, which is why it's the better headline number.
Track in units with flat stakes
If your stakes jump around, a couple of big bets can decide your whole ROI. Flat stakes, or at least logging everything in units, keep the number honest. Our bankroll management guide covers sizing.
Why a few hundred bets can't tell you much
Every bet either wins or loses, so it follows what statisticians call a Bernoulli distribution, with variance p × (1 - p), where p is the chance of winning. For a one-unit bet at decimal odds d, that works out to a standard deviation of about d × √(p × (1 - p)) units per bet. At odds of 2.00 with a 50% chance, that's 1 unit: each bet swings your result by roughly your whole stake.
Average that over n bets and the luck in your ROI shrinks with the square root of n. That's the standard error: the per-bet standard deviation divided by √n. Here's what it looks like across sample sizes and prices:
| Typical odds | Luck band after 100 bets | 500 bets | 1,000 bets | 2,500 bets |
|---|---|---|---|---|
| 1.50 | ±7.1% | ±3.2% | ±2.2% | ±1.4% |
| 1.91 | ±9.5% | ±4.3% | ±3.0% | ±1.9% |
| 2.00 | ±10.0% | ±4.5% | ±3.2% | ±2.0% |
| 3.00 | ±14.1% | ±6.3% | ±4.5% | ±2.8% |
| 4.00 | ±17.3% | ±7.7% | ±5.5% | ±3.5% |
| 10.00 | ±30.0% | ±13.4% | ±9.5% | ±6.0% |
Under the 68-95-99.7 rule, your ROI will land within one band of your true edge about 68% of the time, and within two bands about 95% of the time. So after 500 bets at around 1.91, a bettor whose true long-run ROI is exactly 0% will show anything from about -8.6% to +8.6% on a normal run.
A worked example
An invented NBA bettor logs 300 bets at an average of 1.91, $20 flat. That's $6,000 staked. They finish $360 up, a 6% ROI.
The luck band at 300 bets and 1.91 is about ±5.5%. So 6% sits only just over one band above zero. A bettor with no edge at all would post 6% or better over 300 bets roughly 1 time in 7. The result is consistent with a real edge, and it's also consistent with none. It isn't proof either way.
Good bettors have long losing runs too
The maths cuts both ways. Take a bettor with a genuine +3% edge betting at 1.91. The chance they are still behind after:
| Bets placed | Chance of being behind at 1.91 | Chance of being behind at 4.00 |
|---|---|---|
| 100 | 38% | 43% |
| 500 | 24% | 35% |
| 1,000 | 16% | 29% |
| 2,000 | 8% | 22% |
| 5,000 | 1% | 11% |
These come from the same standard-error maths. Over a few hundred bets, a real edge loses money about a quarter to a third of the time. That's why staking discipline matters more than any single month's figure.
Longer odds need far more bets
Racing and outright markets sit at the bottom of the first table. At 10.00, the luck band is still about ±9.5% after 1,000 bets. A few dozen bets through the Spring Carnival tell you almost nothing about whether your approach works, in either direction.
How many bets do you actually need?
For a true edge to sit two luck bands clear of zero, you need roughly (2 × standard deviation per bet ÷ edge)² bets. For a 3% edge:
| Typical odds | Bets needed for a 3% edge | Bets needed for a 5% edge |
|---|---|---|
| 1.50 | about 2,200 | about 800 |
| 1.91 | about 4,000 | about 1,500 |
| 3.00 | about 8,900 | about 3,200 |
| 4.00 | about 13,300 | about 4,800 |
| 10.00 | about 40,000 | about 14,400 |
That lines up with the benchmark Covers mentions of about 5% over roughly 3,000 wagers. These figures assume flat stakes, similar odds across the sample, and bets that don't depend on each other. Same-game multis and several bets on one game are correlated, which makes the swings bigger.
What realistic numbers look like
Covers says great long-term bettors sit around 5% to 7%, and that anyone expecting 15% or more needs a reality check. Without an edge, the expected ROI is negative, because the bookmaker's margin is built into every price (see what vig is).
Even a large, fully logged record moves around month to month. BetBank publishes its pick record on its bet tracker page, with every pick logged before kickoff and settled against official results. As shown on 12 October 2026 it covers 42,117 settled bets, and its monthly ROI from June to September ran between +0.2% and +7.3%. Each month is a few thousand bets, and the maths above says swings of that size are normal. Past results don't predict future ones.
The long run can also be cut short. Researchers who staked real money on a statistical strategy found that bookmakers use discriminatory practices against successful clients, so a winning sample at one bookmaker may simply stop.
What to check before your ROI settles
Profit and loss is the slowest signal you have. Two faster ones are worth logging from day one.
Closing line value
Closing line value compares the odds you took with the final odds before the event starts. Each bet's CLV usually varies by a few per cent, not by your whole stake, so its average settles in far fewer bets than ROI does. Beating the close consistently suggests you're getting good prices. It doesn't mean you'll finish ahead. Our closing line value guide explains how to calculate it.
Expected value at the time of the bet
Record the fair price, with the margin removed, when you place each bet, and the expected value that implies. If your average logged EV is +2% and your ROI after 400 bets is -3%, the table above says that gap is well inside normal luck. The BetBank calculators include no-vig, implied probability and EV calculators for this.
Split results, but read the splits carefully
Breaking results down by sport, market and odds band is useful, but each slice has fewer bets, so its luck band is wider. A market showing +12% on 80 bets is almost all noise.
A simple routine for your tracker
- Log every bet before the event starts, losers included.
- Record the odds you took, the closing odds, your stake in units, and the fair price.
- Report yield on turnover, in units, never return on bankroll on its own.
- Next to your ROI, write the luck band: standard deviation per bet ÷ √(number of bets).
- Judge your process (CLV and logged EV) monthly, and your results over seasons, not weeks.
- Set a loss limit and a time limit before you start, and stick to them.
If you'd rather not build this in a spreadsheet, our bet tracker comparison covers what to look for in an app.
Track your ROI and CLV in one place
The BetBank bet tracker logs your bets and tracks your ROI and bankroll, so the numbers in this guide are there without a spreadsheet. Start with the 7-day free trial and log your next bets there.
BetBank is for adults 18+ only. Betting carries real risk, no tool or tracker makes it profitable, so set limits and gamble responsibly. If betting stops being fun, BetStop lets you self-exclude from every licensed Australian online and phone bookmaker for free.