Binomial Distribution in Sports Betting: Why Winning Bettors Still Have Losing Streaks
By: JV Miller | Professional Sports Bettor and Publisher
A bettor can be genuinely good at picking winners and look absolutely terrible for weeks at a time. That isn't a contradiction. It's probability.
Suppose you are a legitimate 55% sports bettor. Over a sufficiently large number of wagers, you expect to win approximately 55 out of every 100 bets. At standard -110 odds, that's an excellent long-term result.
Now here's the dangerous assumption: If you're a 55% bettor, you might expect your next 21 wagers to look something like 12 wins and 9 losses.
They might. But probably not.
A 55% expectation is not a prediction of your next 21 bets
If every wager has an independent 55% probability of winning, the number of winners in a 21-bet sample follows what mathematicians call a binomial distribution.
The expected number of wins is:
21 × 55% = 11.55 wins
Since we can't win 0.55 of a bet, 12–9 is the single most likely record. But here's the important part:
The probability of finishing exactly 12–9 is only about 17%.
In other words, even though 12–9 is the most likely individual outcome, there is roughly an 83% chance that something else happens.
That “something else” can be considerably better—or considerably worse.
Being a 55% bettor does not mean winning 55% every week
This is where sports bettors get themselves into trouble. They confuse a long-term expectation with a short-term forecast.
A genuine 55% bettor can go 15–6. He can also go 8–13. Neither result, by itself, proves that his underlying handicapping ability changed.
In fact, with a true 55% win probability, the chance of winning 8 or fewer of 21 wagers is about 9.1%. That's roughly once in every eleven non-overlapping 21-bet samples.
And truly ugly stretches aren't impossible either. The probability of winning 7 or fewer is about 3.8%. Rare? Yes. Impossible? No. More like inevitable.
Bankroll management is mathematics, not psychology
That is why we recommend sensible unit sizes and why we strongly oppose changing wager sizes simply because a bettor has recently been in a winning or losing streak.
After all, you never know if you're in a winning or losing streak. No matter how many wagers you won or lost in a row, anything can happen now. All you know is you were in a streak. But that was yesterday. Anything can happen today.
A streak may tell you something about your handicapping. Or it may tell you almost nothing. The binomial distribution holds the key to understanding whether or not you have an edge.
Why Losing Streaks Are Virtually Guaranteed Over a Full Season
The 21-wager example becomes even more important when we stop looking at a single group of bets and start looking at an entire season.
Suppose you make 210 wagers during a football season. It is tempting to divide those wagers into ten neat groups of 21: Bets 1–21, bets 22–42, bets 43–63, and so forth.
But that isn't how a bettor actually experiences a season. At any point, you can look back at your most recent 21 wagers. Bets 1 through 21 form one 21-bet window. Bets 2 through 22 form another. Bets 3 through 23 form another.
Continue through the season and 210 consecutive wagers contain 190 different overlapping 21-bet windows. That changes the way you should think about streaks.
A rare result gets many opportunities to occur
We just established that a true 55% bettor has about a 9% chance of winning eight or fewer of any particular 21 wagers. Nine percent doesn't sound terribly frightening when you're considering only one group of 21 bets.
But over a long season, you don't get only one opportunity to encounter an ugly stretch. You get many. Those overlapping windows are not statistically independent, so it would be incorrect to treat 190 windows as 190 separate coin flips and simply multiply the probabilities.
But the practical conclusion is unavoidable:
Over hundreds of wagers, a good bettor should expect to experience stretches that look very bad.
The same thing happens in the other direction. At some point during a long season, a 55% bettor may produce a spectacular run that makes him appear to be a 70% handicapper.
He isn't.
And when he later goes 8–13, he hasn't suddenly forgotten how to handicap.
Hot streaks can be just as dangerous as losing streaks
This may sound strange, but an unusually good run can be one of the most dangerous things that happens to a bettor.
A bettor who begins with a losing streak may conclude that his handicapping method doesn't work. A bettor who begins with a spectacular winning streak can make an even more expensive mistake:
He may conclude that he is better than he really is.
That's when wager sizes start increasing. A one-unit play becomes two units. The bettor starts assigning "three-star" and "five-star" ratings to selections. A larger bankroll becomes justification for larger and larger bets.
Then ordinary variance eventually moves in the other direction. The handicapper didn't suddenly become stupid.
The probabilities simply caught up with him.
Don't resize your bankroll strategy around a streak
This is one of the reasons we are so adamant about conservative, consistent staking. Your bankroll strategy should be capable of surviving the bad stretches that probability says will eventually occur.
It should also prevent you from becoming reckless during the good ones.
If your long-term expectation is 55%, you should build your staking strategy around being a 55% bettor—not around whatever winning percentage happens to appear in your last 10, 20 or 30 wagers.
A hot streak does not automatically justify a larger bet. A losing streak does not automatically justify a smaller bet. And neither one justifies chasing.
Variance is not an emergency. It is part of the business.
That is why professional sports betting requires enough capital—and small enough units—to remain in action long enough for a genuine mathematical edge to reveal itself.
Check These Topics Out, Too:
Don’t Judge a Handicapper by the Last 20 Bets
One of the strangest things about publishing sports selections is watching how differently bettors react to the exact same handicapper depending on when they happen to start watching.
We have seen this for decades.
A new subscriber joins during a winning streak and may conclude that we are smarter than we really are.
Another joins during a losing streak and may conclude that we are dumber than we really are.
Neither conclusion is necessarily justified.
They may simply have entered the same long-term sequence at different points along the probability curve.
A winning streak can create dangerous confidence
Suppose a new subscriber happens to arrive just as a 55% handicapper goes 16–5.
That's a terrific run.
But the subscriber hasn't yet experienced the hundreds of wagers that define the handicapper's actual long-term performance. He has experienced 21 unusually successful ones.
The danger is that he mistakes the streak for the norm.
Suddenly a 2% wager seems too conservative.
Why bet $200 when these guys are killing it?
Why not $500?
Why not $1,000 on the plays we really like?
This is precisely how a favorable run of variance can encourage a bettor to increase his exposure at the worst possible time: after recent results have convinced him that normal risk no longer applies.
It still applies.
The 16–5 handicapper can be the exact same handicapper who later goes 8–13.
A losing streak creates the opposite illusion
Now imagine joining at the other end of the distribution.
Your first 21 selections go 8–13.
The natural reaction is obvious:
These guys can't pick winners.
Maybe that's true.
But 21 wagers don't establish it.
As we demonstrated earlier, even a bettor with a genuine 55% win probability has a meaningful chance of producing an 8–13 or worse stretch through ordinary variance.
This creates one of the fundamental problems in evaluating sports handicapping:
Good results do not necessarily prove skill, and bad results do not necessarily prove its absence—not over a small sample.
The smaller the sample, the louder luck speaks.
Evaluate the process over a meaningful sample
This is why we have historically encouraged prospective clients to track our selections before drawing conclusions.
Not five picks.
Not ten.
Not one spectacular week.
The more wagers you observe, the less influence any single hot or cold streak has on the overall record.
Even 100 wagers can produce considerable variation. Two hundred tells you more. Hundreds more tell you more still.
There is no magical sample size at which luck suddenly disappears. It doesn't.
But as the sample grows, sustained performance becomes increasingly difficult to explain as nothing more than short-term variance.
That is the perspective serious bettors need.
Don't let recent results dictate your bet size
This brings us back to bankroll management.
If you increase your wagers because a handicapper has recently been hot, you are allowing a short-term statistical fluctuation to determine your capital allocation.
If you double your wagers because you're trying to recover from a losing streak, you're doing the same thing in reverse.
Neither changes the probability of the next wager winning.
Your staking plan should be based on your bankroll and your demonstrated long-term edge—not on the emotional temperature of your last 20 bets.
A winning streak doesn't make you invincible.
A losing streak doesn't necessarily make you wrong.
Sometimes, probability is simply being probability.
Wo cover all the principles of money management in our artricle Sports Betting Money Management: Professional Bankroll & Staking Strategies.
Professional Gambler Newsletter
Advanced +EV Strategies. Syndicate-Grade Selections.
JV Miller
JV Miller Reveals the 5 Most Overrated Factors in Handicapping Football