Learn
How to win at sports betting, answered without the sales pitch
There is a real answer to this question and it is narrower than the phrase suggests. You cannot make winning happen. You can build a process that wins on average over a long enough sample, and you can arrange your stake and your bankroll so that the long sample arrives before the bankroll does not.
Educational only. quant+ does not accept wagers and makes no claim of profit.
- 52.381% The win rate at which a -110 price returns exactly nothing. Everything above this is profit, everything below is loss, and the margin is thin.
- about 48% How often a bettor sitting exactly at that break-even rate is behind after 2,000 bets. A fair game is not a kind one.
- 1 to 3 points. The size of the edge a genuinely profitable bettor usually has over the closing price. Not a trick, not a system. A small, persistent advantage.
Those three numbers are the honest frame for the whole question. Winning at sports betting is a small edge, held for a long time, sized so that you are still there at the end. Every other answer is selling something.
Four things have to be true at once, and three of them are not about picking
Most material on this subject addresses only the first. The other three are what separate a bettor with a real edge from one who had a good season.
1. You have an edge
Your probability estimates are better than the price on enough bets to matter. This is the part everyone focuses on and it is the hardest to obtain and the hardest to prove. A one to three point advantage over the closing line is a large edge.
2. You can get the price
The advantage has to be available at the moment you act, at a stake worth taking. An edge you can only obtain for a stake of a few units is a real edge and not a living.
3. Your stake does not end the story early
Even a genuine edge passes through long losing runs. Sizing is what decides whether a losing run is an inconvenience or the end of the exercise, and it is entirely within your control, unlike the edge.
4. The market lets you keep it
A book that limits or closes the accounts that beat it converts a long-run edge into a short-run one. This is an operational constraint on the whole enterprise, not a detail.
A bettor who has the first and not the other three does not win. That is the ordinary failure, and it is the reason this page spends most of its length on the last three.
Variance does not care that you are right
This is the table that most betting material never prints. It shows, for a bettor who genuinely wins at the given rate and bets -110 prices, how often they are behind after a given number of bets.
| True win rate | Behind after 100 | After 200 | After 500 | After 1,000 | After 2,000 |
|---|---|---|---|---|---|
| 52.4% (break-even) | 50.7% | 48.3% | 48.2% | 48.7% | 49.1% |
| 55.0% (large edge) | 30.7% | 21.7% | 11.3% | 4.6% | 0.9% |
| 58.0% (very large) | 13.3% | 5.0% | 0.5% | 0.0% | 0.0% |
The first row is the important one. A bettor at exactly the break-even rate is behind about half the time no matter how long they play, because their expected profit is zero and drift never rescues them. This is why a break-even bettor can play for years and never see the long run arrive: there is no long run to arrive.
The second row shows what a real edge looks like in practice. A bettor with a large five-point advantage is still behind after 200 bets more than one time in five. Nothing about that contradicts the edge. It is the edge working, with noise on top.
The practical consequence is that a bad run is not evidence against your process, and a good run is not evidence for it. Judge the process by whether the prices you took beat the closing prices, which is a property of your decisions rather than of your results.
Long losing runs are not bad luck, they are the schedule
A win rate is an average, and averages hide the arrangement. At any realistic success rate, losing five in a row is close to certain over a season, and losing ten is not rare.
| True win rate | Bets | At least 5 in a row | At least 7 in a row | At least 10 in a row |
|---|---|---|---|---|
| 52.4% | 200 | 93.4% | 43.9% | 5.8% |
| 52.4% | 1,000 | 100.0% | 94.8% | 26.8% |
| 55.0% | 200 | 88.0% | 33.4% | 3.5% |
| 55.0% | 1,000 | 100.0% | 87.5% | 17.0% |
A profitable bettor at 55% will, over a thousand bets, almost certainly lose seven in a row, and has about a one in six chance of losing ten in a row. Neither of those events is a signal that the process broke.
What matters is what those runs do to the bankroll, and that is a staking question rather than a betting one. At a flat two percent stake, ten consecutive losses take the bankroll down 18.29 percent. Twenty in a row takes it down 33.24 percent. Both are survivable at that stake. At a ten percent stake, ten in a row is a 65 percent drawdown, which is the point at which most people stop, and the drawdown is the only part of that sentence they will remember.
This is why stake sizing is not a detail of a winning approach. It is the mechanism that converts a real edge into a result, and there is no edge that survives being sized as though losing runs do not happen.
How much to stake, and why the theoretical answer is uncomfortable
There is a standard answer to stake sizing, and it has a property that makes almost nobody follow it exactly. That gap between theory and practice is worth understanding rather than ignoring.
The growth-optimal stake for a bet at decimal price d with win probability p is a fraction of the bankroll, and it is the one that maximises the long-run growth rate rather than the expected profit on any single bet.
f* = (p x d - 1) / (d - 1)
at d = 1.9091 (a -110 price):
p = 53.0% f* = 1.30% EV = +1.18% per unit
p = 55.0% f* = 5.50% EV = +5.00% per unit
p = 58.0% f* = 11.80% EV = +10.73% per unit
Two things are visible in those numbers. The first is how quickly the optimal stake rises: a three-point improvement in win rate multiplies it by more than four. The second is the problem. That 58% is your estimate of your own edge, and if it is wrong the optimal fraction is wrong with it, in the expensive direction.
| True win rate | Full optimal stake | Half | Quarter |
|---|---|---|---|
| 55%, growth per bet | 0.1378% | 0.1033% | 0.0602% |
| 58%, growth per bet | 0.6365% | 0.4762% | 0.2773% |
Halving the stake costs about a quarter of the growth rate and removes roughly half the drawdown. Betting a quarter keeps only 44 percent of the growth and makes the ride much flatter. That is the trade, and it is why bettors who understand the maths still bet well under the full fraction: the penalty for underbetting is a slower curve, and the penalty for overbetting is ruin.
How long before you can tell whether you are winning
Whether a record shows a real edge or a coin landing well is a sample size question, and it has an answer. The answer is usually longer than the streak that prompted the question.
| True win rate | Edge per bet | Bets before a 95% lower bound clears zero |
|---|---|---|
| 60% | +14.55% | about 112 |
| 58% | +10.73% | about 209 |
| 55% | +5.00% | about 977 |
A bettor with a real five percent edge over the price needs roughly a thousand bets before a statistical test can separate their record from noise. At a hundred bets a season, that is a decade. Most people who ask how to win are asking after fifty bets.
There is a way to get a verdict sooner, and it does not involve the results at all. Record the price you took and the price at which it closed. If you consistently take prices longer than the close, you are getting value, and that becomes visible in dozens of bets rather than thousands. It is a leading indicator of edge, and it is a stronger signal than a profit and loss number over the same span.
None of this makes a losing record into a winning one. It makes a short record legible, which is the only thing a short record can be.
The market has a ceiling, and it is not set by your skill
Everything above is about you. This section is about the other side of the transaction, which has its own interests and its own tools.
Winning bettors get restricted
Sportsbooks are businesses that profit from the margin, and a customer who reliably takes value from them is a cost. The common response is to reduce the maximum stake, to remove promotional eligibility, or to close the account. None of this requires the book to be doing anything wrong, and none of it is unusual.
The practical consequence is that an edge in a soft market may have a short working life, and a plan that assumes you can keep taking the same stake forever is a plan that assumes the book does not notice.
The ceiling is a stake limit, not a win rate
Multiply a five percent edge by the stake you are actually permitted to place, and the result is the money the edge can produce. If the limit is small, the edge is real and the income is small, and nothing about the analysis changes that.
This is the reason serious bettors talk about limits more than about models. A modest edge at a usable stake beats a large edge at a stake that gets refused, and the second situation is far more common than the first.
Other constraints that do not show up in the maths
- Time to act. Prices move. An edge that exists for thirty seconds is only yours if you are at the screen for those thirty seconds, which is a constraint on how you spend your day.
- Rule differences. Two prices that look comparable can settle differently, and the difference belongs to the rule rather than to your edge.
- Tax treatment. Where winnings are taxed, the effective edge is smaller than the market edge, and a thin edge can be thin enough that this decides the outcome.
- Your own tolerance. A staking plan you abandon during a losing run is worse than a more conservative one you keep, and only one of those two facts is in the arithmetic.
Why nothing can be guaranteed, stated as arithmetic rather than as caution
Guaranteed-profit claims are not merely unlikely. Each of them describes something that would break the market it is offered in, which is a stronger objection than saying it is risky.
- A system that wins every week. A method that won reliably would be copied and priced away, because the people best placed to spot it are the ones setting the prices. What survives is a small edge that is hard to scale, not a rule that keeps working as more money follows it.
- A sure thing, a lock, a can't-lose bet. Every bet has a losing outcome by definition, or it would not be offered at a price. A price is the market's statement that the other side is possible, and no amount of information you hold makes an outcome certain in a sport.
- Turn a small bankroll into a large one in a few weeks. Growth comes from an edge applied to a stake over time. A rate of return high enough to do that in weeks implies a stake fraction that would be wiped out by the losing runs in the table above, which are near certain over that many bets.
- An 80 percent strike rate with a positive return. A high strike rate is easy to produce by taking short prices, and short prices need a very high strike rate to be worth taking. An 80 percent hit rate at a price of -500 loses money, so the strike rate on its own says nothing at all.
The honest version of this page's title is that you can win at sports betting in the sense that a long-run positive expectation is achievable, and you cannot win in the sense the phrase usually implies, which is reliably and quickly.
Anyone promising the second is either describing a period of good variance, or selling something, or both. The tell is the same in every case: a claim about results with no statement of the sample, no statement of the price taken, and no statement of the stake.
Measure the decisions, not just the results
A journal that records your price against the closing price gives a verdict on your process in dozens of bets, rather than waiting a thousand for the results to speak.
Questions people ask
Can you actually win at sports betting long term?
A long-run positive expectation is achievable and it is rare, because it requires probability estimates that beat the market's on a recurring basis. Even when it exists, the edge is usually small, around one to three points over the closing price, and it is subject to stake limits. What is not achievable is a reliable, fast, or guaranteed return, and any presentation that implies one of those is describing variance or selling a product.
Why do I keep losing if my picks are good?
Two arithmetic reasons and one operational one. At a genuine 55% win rate you are still behind after 200 bets about 22 percent of the time, and losing seven in a row over a thousand bets is close to certain. Operationally, a price that was available when you picked it may not have been available at the stake you took. A losing run at a real edge is the expected shape of the curve, not evidence against the picks.
What percentage of bets do you need to win to profit?
At a -110 price the break-even win rate is 52.381%, because the decimal price is 1.9091 and one divided by it is 0.52381. Anything above that is profit and anything below is loss. At shorter prices the required rate is much higher, and at longer prices it is lower, so the answer always depends on the price rather than on the sport or the market.
How much of my bankroll should I stake per bet?
The growth-optimal fraction for a bet at a -110 price is about 1.30% at a 53% win rate, 5.50% at 55%, and 11.80% at 58%. Those figures depend entirely on your estimated edge being correct, which is the reason many bettors use half or a quarter of the optimal stake. Halving the stake costs roughly a quarter of the growth rate and removes about half the drawdown, which is usually the better trade for anyone who answers to a real bankroll.
How many bets before I know whether I am winning?
For a 95% lower bound on the return to clear zero, a true 55% win rate needs roughly 977 bets, a true 58% needs about 209, and a true 60% needs about 112. Most bettors asking the question are looking at far fewer. Recording the price you took against the closing price gives a usable verdict in dozens of bets, because it measures your decisions rather than your results, and it is the leading indicator rather than the lagging one.
Is a high strike rate the same as winning?
No, and the two can point in opposite directions. A strike rate is only meaningful alongside the price. Winning 80% of bets at -500 loses money, and winning 40% of bets at +150 makes money. A service promoting a strike rate without stating the average price has told you nothing about whether it profits.
