CORTENWhen a result shows up often in past matches played at the same odds, it is usually chance. Look at many odds ranges at once and some will look “more frequent than expected” by accident, and small samples make the effect larger. Before trusting a pattern, check the sample size, the interval and whether it repeats in a separate period.
Looking for repeating odds means drawing conclusions about today’s match from the results of past matches played at the same or very similar prices. Looking is instructive, but a few traps make chance look like a pattern.
Why the idea is tempting
The same price means the market expected a similar probability. If one result appeared more often than expected in past matches at that price, it is tempting to think it will happen again. The problem is that “more often than expected” very often shows up by chance alone.
Multiple testing: one range out of 20
A synthetic thought experiment: take a result whose true probability is exactly 50% and examine it in 20 different odds ranges with 30 matches each. The chance of seeing it in 20 or more of the 30 matches (66.67%+) in any single range is about 4.94%. But the chance of seeing that in at least one of the 20 ranges is about 63.67%. In other words, you will usually find a “surprisingly good” range, even though no range has a real edge.
Small samples
The more narrowly you define a price (say, only 2.15), the fewer matches you find. For 20 out of 30, the 95% Wilson interval is roughly 48.78%–80.77%: perfectly consistent with 50%. The sample size guide and the calculator show this uncertainty.
How to check an observation
1) Write down the rule and the odds range before looking at the results. 2) Split the history in two: find the rule in the older period and test it in the newer one. 3) Compare the hit rate with the margin-free probability derived from those matches’ prices, because seeing a result the price already expected is not information. 4) Record how many rules you tried and report the failures too.
Conclusion
Historical odds patterns are a useful starting point for investigation; on their own they are not evidence about the future. In our own research, a pre-registered hypothesis did not replicate in new leagues — we wrote it up in why beating football odds is hard, and describe these limits on the methodology page.
Cortenscore members can review past prices and results together with the sample size; access requires a membership.
Create account Membership and pricing (Turkish) ↗
Published 28 October 2026 · Last updated: 28 October 2026 · Responsible: Cortenscore editorial team · Spotted an error? Report a correction. This page is for statistics education; it is not a prediction or betting advice. Example numbers are synthetic.