The paradox in one sentence
Each unlikely result is unlikely, but unlikely results as a group are common. That is the heart of what looks like a paradox: people are repeatedly surprised by outcomes that, taken together, are expected to turn up regularly.
Several separate effects feed this, and each is worth understanding on its own.
Many chances for something rare
Rare events happen because there are many opportunities. Take twenty independent events that each have a 5 percent chance of happening. The chance that none happen is 0.95 to the 20th power, which is about 36 percent. So the chance that at least one happens is about 64 percent.
The same arithmetic applies to a game with many players. If a one-in-a-thousand outcome is available to 10,000 players, the expected number who get it is 10. When one of them posts about it, it can seem like a miracle. It is just a large number of tries.
For this reason, remarkable stories from a huge community are weak evidence that anything is wrong with the odds. The question to ask is how many opportunities existed, not only how unlikely the event was for one person.
Small samples exaggerate
Short records are noisy, as discussed in more detail in the sections on repeated trials. A player whose true win rate is 55 percent is expected to win 5.5 games in ten, with a standard deviation of about 1.6 wins. A record of 8-2 is within about 1.6 standard deviations of that, which is unusual but not startling, and 3-7 is similarly plausible.
Human intuition tends to treat small samples as more informative than they are. A short run of results seems to reveal the underlying truth, so a new build, a new player or a new team is declared strong or weak on very little evidence. This is why sample size is the first thing to check when someone claims that a change made a difference.
Favourites are not locks
A 70 percent chance sounds like near certainty to many people, but it means the other side wins nearly one time in three. If a team is a 70 percent favourite in each of three independent matches, the chance they win all three is 0.7 cubed, which is 0.343, or about 34 percent. Even with an edge in every game, a clean sweep is less likely than not.
This catches people out in tournaments, where a favourite must survive many rounds. The format matters here: single elimination multiplies the risk of losing across several matches, so upsets are an expected feature, not a failure of the system. Individual match odds can be very good and the chance of winning the whole event can still be modest, which is what estimates of championship odds typically show.
Misreading what a probability promises
Probabilities are statements about frequency over many cases, not guarantees about one case. Three common misreadings follow.
“It was 90 percent, so it should have happened.” A 90 percent outcome fails one time in ten. If it never failed, the 90 percent figure would be wrong.
“It happened, so the odds were wrong.” A single surprising result cannot confirm or refute a stated probability. Many results are needed. If a 10 percent event happens, the 10 percent label was not shown to be wrong.
“The long run will correct this.” Averages stabilise because more data dilutes earlier results, not because later results cancel them. The idea is explained in more detail in the discussion of repeated trials.
Coincidences that are not coincidences
Some surprising outcomes involve the choice of what to count. If you notice that a rare item dropped on your birthday, the right question is how many “special” patterns you might have noticed, not how unlikely that specific one is. With many possible patterns available, the chance that one of them shows up is high. This is a quiet form of the many-chances problem, and it explains why people often find meaning in random coincidences after the fact.
Expectation drift and feel
There is also a more mundane reason expectations fail: the numbers a player sees are not always the numbers the game uses. Displayed percentages may be rounded. Some mechanics are hidden. The odds may depend on conditions that are not shown. A player who relies on a stated figure while the real system applies something different will keep being surprised. Published loot box odds describe one layer, and the effective chance may differ with other rules.
How to calibrate expectations
A few habits reduce surprise.
- Ask how many chances there were. Multiply the single-event probability by the number of opportunities to get the expected count.
- Think in ranges, not points. For a 55 percent player over ten games, expect a range of wins rather than exactly 5 or 6.
- Treat favourites as favourites. A 65 to 75 percent chance still loses regularly. Prepare for the other outcome.
- Remember the cost of multiple rounds. Chains of probabilities shrink quickly when multiplied.
- Judge decisions separately from outcomes. A good decision can lose and a bad decision can win. The habit of results thinking is a common trap, and it makes surprising outcomes feel more meaningful than they are.
- Check the rules. Hidden adjustments and rounding can explain a gap between what you expected and what you saw.
What the paradox really says
The paradox dissolves once you separate two views. At the level of one event, the surprise is real: a 5 percent event is unexpected when it lands. At the level of everything that happens across many players, many matches and many rolls, a steady supply of 5 percent events is exactly what the math predicts. Learning to hold both views at once is what makes outcomes that defy expectations more understandable, and a little less frustrating.