Why the question is hard to answer
Most games mix chance and choice. A good result can come from skill, luck or both, and from a single outcome you usually cannot tell which. The way to separate them is to look at a sample of results and ask how likely that pattern would be if only chance were operating.
This is the core of the luck versus skill question: not “did I win?” but “how surprising would this record be if there were no skill at all?”
A baseline: what pure luck looks like
Start with a game that is entirely luck, a fair coin. Suppose you call heads or tails 10 times and get 7 right. It feels like evidence of a knack. Check it.
The number of ways to get exactly 7 correct out of 10 is 120. For 8, it is 45. For 9, it is 10. For 10, it is 1. Together that is 120 + 45 + 10 + 1 = 176 ways to get 7 or more. The total number of possible sequences is 2 to the power 10, which is 1024. So the chance of 7 or more correct by luck alone is 176 / 1024, about 17%.
That is roughly one in six. If six people each tried this, you would expect about one of them to get 7 or more without any ability. This is a basic use of the binomial distribution, and it shows how often chance produces results that look impressive.
Now suppose you got 70 correct out of 100 calls. The proportion is the same, but the situation is very different. With a fair coin the typical spread in 100 flips is about 5 around an average of 50, so 70 is around four such steps above the average. That is rare by chance alone, well under one in a thousand. Same percentage, much stronger evidence, simply because the sample is bigger.
Why small samples mislead
Small samples swing widely. In a handful of games, a player with no edge can win most of them, and a skilled player can lose most of them. This is why a short record, whether a winning streak or a bad week, says little by itself. Small sample sizes also exaggerate extremes: the best and worst records almost always come from the smallest samples.
As the number of results grows, the average settles toward the true rate, which is what the law of large numbers describes. A confidence interval is one way of showing how much uncertainty remains: a wide interval means the data cannot yet tell a strong player from a lucky one.
Signs that skill is probably involved
Skill leaves a more consistent footprint than luck. Look for:
- Repeatability. Results that stay good across many separate sessions, opponents and conditions.
- Persistence. Players who do well in one period tend to do well in the next. In pure luck games, past performance does not predict future performance. Where it does, regression to the mean is weaker.
- Process evidence. You can point to specific decisions, such as better estimates of odds or sensible use of position, rather than just the final score.
- Rating systems. Systems such as Elo ratings measure how well results predict later results. They can reveal skill, but they also need enough games to be reliable.
Signs that luck is doing the work
- A great record based on few games.
- A strategy that “works” only in a particular streak or session.
- Explanations invented after the fact, which can fit any outcome.
- Success stories that ignore the many people who tried the same thing and failed, a problem known as survivorship bias.
- Belief in a hot hand or a “due” outcome. In independent random events, previous results do not change the next one; the gambler’s fallacy and the hot hand are both ways of reading patterns into noise.
Games sit on a spectrum
Pure luck games, such as fair dice or roulette, offer no way to improve the odds; no strategy, pattern or record changes them. Pure skill games, such as some abstract strategy games, have no chance element. Most games lie between, and the share of each differs. Designers often tune this deliberately, and a game with more luck needs a larger sample before skill becomes visible.
For casino games, the point is blunt: they carry a house edge, and results over time reflect that edge rather than skill, with the exception of a few formats where player decisions affect the odds, and even then outcomes remain uncertain. If you notice you are explaining a few wins as proof of a system, treat that as a signal to slow down. Never spend more than you can afford to lose, and if gambling is causing worry or harm, contact a local support service.
A short checklist
- How many results do I have? If it is fewer than a few dozen, treat conclusions as tentative.
- What would pure chance produce here, roughly?
- Does the record repeat across different periods and opponents?
- Can I explain the edge before looking at the result?
- Am I ignoring failures, or counting only the cases that worked?
A strategy based on probability is one you can justify with odds and payoffs. A strategy based on luck is one whose success depends on a record too short to rule out chance.