Think Like a Professional Sports Bettor: The Realistic Roadmap
No income figure appears here, because none has a source. What follows is the arithmetic of the job, and a map of this part of the site.
A professional sports bettor beats the margin a sportsbook builds into its prices, or the commission a betting exchange takes from winnings, over a sample large enough for the edge to show through the luck. At odds of 1.98 a bet must win 50.51% of the time to break even. And a bettor with a real edge of 2.56% of each stake still finishes behind after 500 bets in about 28 of 100 runs (calculated, hypothetical). This page gives no income figure and no share of bettors who profit, because no source it uses carries one. It gives the arithmetic: the bar, the sample, the variance, the stake, then a map of the subjects that go deeper.
Reality check
Margin and commission are costs, so a bettor with no edge is expected to lose. Most bettors end behind. A winning run is not an edge. This page explains how the job works. It does not predict your results, and it is not a plan for winning. Decide what you can afford to lose first, and read the responsible gambling page.
The bar you have to clear
How many bets does it take to know you are any good? More than any good month will tell you.
Break-even is one divided by the odds, so at 1.98 a bettor must win more than 1 ÷ 1.98, about 50.51%, of bets to come out ahead on average. A bookmaker puts its charge inside the price, as a margin. An exchange takes a share of net winnings in a market, as commission, which lifts the bar. Say an exchange shows 2.00 and charges an example rate of 3%, chosen for the sum. The price you keep is 1 + (2.00 - 1) × (1 - 0.03) = 1.97, and the bar for that bet is 50.76%.
Odds in other formats go through a converter, the mechanics sit in how commission is charged, and a bookmaker price against an exchange price shows what each costs. I would not make any big decision on a record shorter than the bar in the chart below that matches the edge I think I have, and I would still be careful with a record that long, because luck can clear the same mark.
How many bets before a record means anything
A win rate measured over n bets has a standard error: the square root of (p × (1 - p) divided by n). (Side note: "standard error" is a statistician's way of saying how far luck can drag a result. Anyway.) For the average record to sit two standard errors above break-even, n = 4 × p × (1 - p) divided by (p - bar)², with p the true win rate and bar the break-even rate, 1 ÷ 1.98 unrounded.
Hypothetical: flat stakes at odds of 1.98, independent bets, break-even taken as 1 ÷ 1.98 exactly, not the rounded 50.51%. Even then, about half of the runs fall short of the mark.
Show the data table
| True win rate | Bets |
|---|---|
| 51% | 40,804 |
| 52% | 4,467 |
| 53% | 1,601 |
| 54% | 813 |
| 55% | 490 |
What variance does to a good bettor
Picture a bettor whose only question is the price. No account is capped, no stake is trimmed. What is left is the bar, the sample and the swings, and lifting the caps changes none of them.
Give that bettor a real edge: a 51.8% win rate at 1.98 is worth +2.56% a bet (0.518 × 1.98 - 1). Hypothetical, like every figure. Take the edge away, a win rate right at break-even (50.51% rounded), and about one run in two still ends above the start after 500 bets. A positive run says little about an edge.
- Edge at 1.98 with a 51.8% win rate
- +2.56% a bet
- Runs above the start after 500 bets
- 72.0%
- Runs above the start after 2,000 bets
- 87.3%
- No edge, runs above the start after 500 bets
- 50.1%
Bankroll variance simulator
A made-up bettor with a win rate you choose places the same stake on every bet. The simulator plays 1,000 runs of those bets and shows how far their bankrolls wander, how many end below the start and how deep the dips go. The runs are random but repeatable: they come from a seed, not from anyone's betting. Everything is worked out in your browser and nothing is sent anywhere.
Same seed, same paths: the figures change only when you change an input or press Draw again.
A win rate you type is an assumption. Nobody knows their true win rate in advance, and a run of results only estimates it.
The expected figures are averages. No single run follows them, and some runs finish below the start even when the win rate is above break-even.
The 80% line is only a marker. A run that falls below it carries on with its remaining bets, so the dip row counts the made-up bettors who sat through a drop of more than a fifth of the bankroll at some point.
A run that cannot cover one stake stops betting and keeps what is left.
Example shown: a win rate of 51.8% at odds of 1.98, 500 bets of 1 unit, a bankroll of 100 units, seed 1. Replace them with your own.
Twelve of the runs, drawn as paths
Each line is one run, starting from your bankroll. The dashed lines are the runs that did not finish above where they began.
The straight line is the starting bankroll. Solid lines with a filled dot finish above it; dashed lines with a hollow dot finish at or below it. The chart shows one point for every 4 bets; the table uses every bet.
Final bankrolls of the paths shown
Bankrolls are in units. Lowest is the lowest bankroll the run reached, start included. The table uses every bet, one row per run.
| Run | Final | Lowest | Result |
|---|---|---|---|
| 1 | 128.66 | 95.94 | Above |
| 2 | 114.80 | 92.58 | Above |
| 3 | 100.94 | 82.88 | Above |
| 4 | 126.68 | 98.66 | Above |
| 5 | 116.78 | 96.92 | Above |
| 6 | 93.02 | 85.30 | Below |
| 7 | 122.72 | 90.32 | Above |
| 8 | 118.76 | 99.00 | Above |
| 9 | 81.14 | 71.86 | Below |
| 10 | 122.72 | 95.26 | Above |
| 11 | 33.62 | 27.58 | Below |
| 12 | 148.46 | 97.84 | Above |
The simulator draws 1,000 runs from a seed, so its percentages land near these figures, not on them.
Stake size: a rule, not a mood
A rule fixes each stake as a share of the bankroll. One rule is the Kelly fraction: f = (b × p - q) divided by b, where b is the odds minus 1 and q is 1 - p. At a 51.8% win rate and odds of 1.98, f is 2.62% of the bankroll. Because p is an estimate, one response is a fraction of Kelly: half of it is 1.31% here. The formula is arithmetic, not a recommendation.
What to log for every bet
- Price taken, in decimal odds.
- Price at the close.
- Stake, as a share of the bankroll.
- Venue, since the cost is paid there.
- Cost: the margin or the commission you paid.
Win rate alone hides the price. Using the closing price as a yardstick is a subject of its own.
Where a winner can stay
A bookmaker sets a price and manages the risk on each bet, so which bets it takes, and for how much, is its own decision. On an exchange, users face each other, and the question is whether someone offers the other side at your price. For the limits themselves, see the limits guide, which sites accept winners and Orbit's own limit messages.
AsianConnect writes "Certain bookmakers limit winning bettors". Its blog also says "Our team understands the needs of serious punters, arbitrage traders, and professional bettors." Its conditions say "The Player may make use of the Services strictly in his/her personal non-professional capacity". BetInAsia (clause 3.9) and MadMarket (clause 3.10) each write, in their conditions, that the client is not acting on behalf of another party "or for any commercial purposes, but solely on your own behalf as a private individual in a personal capacity".
This page is best for: a bettor who keeps records and sizes stakes by rule. Not ideal if: you expect a good month to settle the question.
A roadmap in seven steps
Count your costs
Find what the margin or the commission takes from each bet.
Find your bar
Turn every price into a break-even rate: one divided by the odds.
Record every bet
The log above, row by row.
Size stakes as a share of the bankroll
Fix the share before the bet, not after the last result.
Wait for the sample
Count the bets first. The chart above shows how long even a real edge takes to show, and a record that long can still be luck.
Choose where a winner can stay
How a broker passes a stake on shows what sits between you and the venue.
Know when to stop
Write down where you stop before the first bet. The responsible gambling page is a place to start.
The rest of this part of the site
Five subjects belong to this part of the site; the cards marked Soon are not published yet.
Arbitrage across venues
Prices that disagree, the stakes that cover them and the legs that fail, with a calculator.
Open the calculator pagePrices against the close
The closing price as a yardstick for the prices you take.
Read the guidePrice against probability
Estimating a fair price and the maths of expected value.
Read the guideA working routine
The method side: costs, records, staking.
Read the guideDigging into data
Data, variables and testing out of sample.
Read the guideShort answers
How many bets prove an edge?
No number settles it, because a lucky run can clear any mark. The chart above shows only the scale: at odds of 1.98, two standard errors over break-even takes 490 bets at a true 55% and 40,804 at 51%, and even there about half of the runs fall short of that mark.
How much do professional sports bettors make?
This page gives no income figure, because none of its sources carries one. It shows the bar and the sample.
Do professional bettors get limited by bookmakers?
AsianConnect writes that certain bookmakers limit winning bettors; the section "Where a winner can stay" has the mechanism and the pages on limits.
Do professional bettors pay tax?
Tax treatment depends on where you live and how the activity is classed. A qualified adviser can answer it; this page does not.
Orbit Exchange opens through a broker, and the sign-up steps take it from there.