Five winning sports bets in a row can do interesting things to your confidence.
The first win is satisfying, the second feels encouraging, and by the time the fifth comes around, you might find yourself looking at your betting history with a new appreciation for your own judgement. Perhaps you've become better at reading the games, or finally learned which teams are worth trusting. It's difficult to know how much credit to take when the results keep going your way.
There's another possibility that probably wouldn't occur to most people checking their betting history. The winning streak might have less to do with an improvement in their predictions than with the kinds of bets they've started choosing.
In 2014, researchers Juemin Xu and Nigel Harvey examined 565,915 online sports bets placed by 776 gamblers [1]. They discovered that bettors who had been winning were more likely to win again, while those who had been losing were more likely to continue losing. As the researchers looked more closely at the records, they also noticed differences in the odds selected following wins and losses.
After winning, bettors tended towards shorter odds, which generally correspond to outcomes with a higher probability of success. Following losses, they tended towards longer odds, where the potential payout is greater but the likelihood of winning is lower.
The pattern suggested that a run of successful bets might partly reflect changes in what people chose to wager on. Someone who appeared to be predicting results increasingly well could instead be selecting outcomes that were more likely to happen.
It was an intriguing finding, although the explanation soon became the subject of a scientific disagreement that has continued well beyond the original study. The difficulty lay in determining whether winning actually changed people's choices, or whether the apparent relationship was partly produced by the way their betting records had been analysed.
Why a Winning Streak Feels Different
People are generally comfortable with the idea of randomness until it begins producing results that look meaningful.
A fair coin, for example, has approximately a 48% chance of producing at least four consecutive heads somewhere in twenty flips. That sort of run can feel unusual when we encounter it, despite being an entirely ordinary possibility in a random sequence.
Our expectations about randomness don't always match its behaviour. We tend to imagine chance producing a fairly even alternation of outcomes, so a sequence of repeated successes can attract more attention than the countless unremarkable combinations surrounding it.
Psychologists Thomas Gilovich, Robert Vallone and Amos Tversky explored a related belief in their influential 1985 study of basketball's hot hand [6]. They examined the widespread assumption that a player who had made several consecutive shots was more likely to make the next one. Their original findings challenged that belief, although later research has revisited aspects of the statistical analysis and identified circumstances in which genuine hot-hand effects may occur.
It's easy to understand why the idea appeals to us. When someone performs well repeatedly, we naturally look for an explanation in their confidence, concentration or ability. The sequence seems to tell us something about the person producing it.
Sports betting introduces an additional complication because the probability of winning can vary considerably from one selection to another. A bettor backing a strong favourite faces a different prospect from someone wagering on an unlikely outsider, even when both are placing bets on the same sport.
Xu and Harvey found that, in their dataset, the proportion of successful subsequent bets increased with the length of the preceding winning streak. After two consecutive wins, the next bet won approximately 57% of the time. Following three wins, that figure rose to around 67%, and after six it reached roughly 76% [1].
These were observed proportions within the study, rather than probabilities that could be applied to any individual bettor. Still, the pattern was striking enough to warrant a closer look at the selections being made.
The Choices Behind the Results
Think of odds as a way of expressing how likely an outcome is considered to be. If a bookmaker offers decimal odds of 1.25, a £10 bet would return £12.50 if it wins, including the original stake. The payout is relatively small because the outcome is considered more likely to happen. At odds of 5.00, the same £10 bet would return £50, but the chances of winning are much lower.
Those prices translate into implied probabilities of 80% and 20%, respectively. The calculation is simply one divided by the decimal odds. These aren't exact predictions of what will happen, since bookmakers include a margin in their prices, but they give us a useful way to understand the difference between a relatively safe selection and a much riskier one.
The difference matters because someone who begins selecting shorter-priced outcomes will generally win a greater proportion of their bets, even if their ability to assess sporting events hasn't improved.
Imagine a bettor who initially enjoys backing outsiders but gradually moves towards strong favourites after a few successful wagers. Their winning percentage may increase as a consequence of those choices. Looking at the results alone, however, it would be difficult to distinguish that change from an improvement in judgement.
Xu and Harvey observed an association between previous results and the odds selected for subsequent bets. They proposed that the gambler's fallacy might help explain it [1].
The gambler's fallacy is the mistaken belief that a run of one outcome makes the opposite outcome more likely, even when successive events are independent. It's the reasoning behind someone expecting red to appear on a roulette wheel because black has come up repeatedly.
The researchers suggested that bettors might respond to winning streaks with a similar expectation of reversal. Someone who feels that a loss is approaching could become more cautious and select a safer bet. Conversely, a person who has experienced several losses might expect a win to be due and choose a longer-priced outcome offering a larger payout.
Under that interpretation, a belief that luck was about to change could influence the choices that made a streak more likely to continue. It was an interesting connection between the gambler's fallacy and the hot-hand effect, two ideas that appear contradictory at first glance.
The betting records couldn't reveal whether the participants actually held those beliefs. They showed the bets people placed and the outcomes they experienced, but not the reasoning behind their selections. A shift towards shorter odds might reflect caution, a change in available opportunities or an entirely different motivation.
There was also a more fundamental question about whether the observed differences represented changes in individual behaviour at all.
When the Data Creates Its Own Pattern
Consider two people with different betting habits. One regularly backs strong favourites, accepting modest potential returns in exchange for a higher probability of success. The other prefers outsiders, knowing that most selections will lose but that an occasional win could produce a larger payout.
Over a long period, the first bettor will generally experience more frequent winning sequences. The second is likely to encounter longer runs of losses.
If a researcher combines their betting records and groups the observations according to the length of each streak, the longer winning sequences will naturally contain a greater proportion of selections from the person who usually backs favourites.
That relationship could emerge even if neither bettor ever changed their approach after a win or a loss.
This is the problem of selection effects, and it became central to the debate surrounding Xu and Harvey's findings.
In 2015, Heath Demaree, James Weaver and Jessica Juergensen published a commentary in Cognition questioning whether the original results could be explained by the composition of the groups being compared [5]. They argued that differences between bettors might create the appearance of a relationship between previous outcomes and subsequent choices without requiring those outcomes to influence behaviour.
Xu and Harvey responded in the same journal, defending their analysis and rejecting the proposed selection-effect explanation [2].
The disagreement concerned something quite specific. Both interpretations could account for an association between winning streaks and shorter odds. What they disputed was whether the association demonstrated that bettors had changed their choices in response to winning.
The distinction matters because the same collection of results can support very different explanations depending on how the observations are grouped and what assumptions are made about the people producing them.
In 2026, Bálint Bakó and M. C. Sándor returned to the issue with a further critique of the original findings [3]. Their analysis argued that patterns resembling the reported hot-hand effect could emerge from differences between bettors and the statistical properties of betting sequences, without requiring the proposed behavioural response.
That doesn't establish that previous wins never influence subsequent decisions. People may well become more cautious, confident or ambitious after a run of results, and gambling research has documented numerous ways in which recent outcomes can affect behaviour.
What the dispute demonstrates is the difficulty of identifying a particular psychological mechanism from patterns in betting records. A relationship that appears consistent with one explanation may also arise under different conditions.
It's an unusually revealing problem for researchers studying human behaviour. We can observe what people choose and what happens afterwards, sometimes across hundreds of thousands of decisions, while the reasoning connecting those events remains much harder to establish.
For bettors themselves, the uncertainty is greater still. Most people looking back over a successful run have only a handful of results to consider, and even those don't necessarily reveal whether the choices were financially worthwhile.
Eight Wins, and Still Losing Money
Imagine placing ten bets of £10 each, all at decimal odds of 1.20. Eight selections win and two lose, giving you an 80% success rate.
It sounds like a good session, particularly if the wins arrive consecutively. But each successful bet returns only £12, including the original stake. Eight winning bets therefore return £96 against a total outlay of £100, leaving a net loss of £4.
At those odds, the break-even win rate is approximately 83.3%. Winning eight out of ten bets isn't quite enough.
The example illustrates why the frequency of successful predictions can be a misleading measure of financial performance. Short-priced selections may win regularly, but their smaller payouts leave less room for losses.
Research into the favourite–longshot bias adds another consideration. In many betting markets, long-priced selections have historically produced worse average returns than favourites, partly because the prices offered don't correspond perfectly to the outcomes' underlying probabilities [4].
Karl Whelan's research into competitive fixed-odds betting markets examines how pricing and risk preferences contribute to that relationship [4]. The findings help explain why two bettors with very different success rates might also experience different financial outcomes.
None of this means that backing favourites guarantees a profit. Even a highly probable outcome can be a poor-value bet if the offered odds are too low relative to its actual chance of occurring. Bookmakers' margins also affect the returns available to players.
A bettor might therefore improve their win rate by selecting shorter odds without improving their overall profitability. Looking back at a sequence of successful wagers, the satisfaction of having been right repeatedly can be much more apparent than the financial consequences of the prices accepted.
What We Remember About Winning
A winning streak has a way of making earlier decisions seem more deliberate than they felt at the time. Once the results are known, the successful selections sit together in the betting history, and it's tempting to see them as evidence of a particularly good period of judgement.
The record rarely preserves the uncertainty that accompanied each choice. It doesn't show which alternatives were considered, how confident the bettor felt or whether their approach gradually changed as the session progressed.
That gap is part of what makes the debate over Xu and Harvey's research so interesting. The original study identified patterns across hundreds of thousands of real bets, yet subsequent researchers have disagreed about what those patterns reveal about the people who placed them.
There may be genuine behavioural changes hidden within the data. There may also be relationships created by the different habits of the bettors being compared. The visible sequence of wins and losses cannot settle the question on its own.
Outside gambling, we encounter a similar difficulty whenever we look back at a series of successful decisions. Knowing that something worked can make the reasoning behind it seem clearer and more consistent than it was when the outcome remained uncertain.
We remember the winning selections because the results are easy to see. The small adjustments, competing possibilities and occasional guesses that preceded them are much less likely to survive in the story we tell ourselves afterwards.
A streak can be a satisfying thing to look back on. Understanding how it happened is considerably more complicated.
References
- Xu, J., & Harvey, N. (2014). Carry on winning: The gamblers' fallacy creates hot hand effects in online gambling. Cognition, 131(2), 173–180.
- Xu, J., & Harvey, N. (2015). Carry on winning: No selection effect. Cognition, 139, 171–173.
- Bakó, B., & Sándor, M. C. (2026). Approaching the Hot Hand with a Cool Head: A Comment on Xu and Harvey (Cognition, 2014). Journal of Comments and Replications in Economics, 5(12).
- Whelan, K. (2024). Risk aversion and favourite–longshot bias in a competitive fixed-odds betting market. Economica, 91(361), 188–209.
- Demaree, H. A., Weaver, J. S., & Juergensen, J. (2015). A fallacious "Gambler's Fallacy"? Commentary on Xu and Harvey (2014). Cognition, 139, 168–170.
- Gilovich, T., Vallone, R., & Tversky, A. (1985). The hot hand in basketball: On the misperception of random sequences. Cognitive Psychology, 17(3), 295–314.