Playoff series length
A Socratic walk-through of playoff series length — reasoned out one step at a time, not lectured.
The question we started with
THE QUESTION #Why does the better team lose a seven-game series far more often than fans expect?
The reason a championship is decided over seven games rather than one is obvious: a single game could be stolen by a hot shooter or a bad bounce, and a longer series lets the better team assert itself. Everyone agrees. The phrase "the cream rises over seven games" is treated as settled fact.
And yet upsets are routine. Every year a favourite that dominated the regular season goes out in the first round. Fans explain this after the fact — an injury, a bad matchup, a coach outfoxed. What if we tried to work out, before reaching for explanations, how often a seven-game series should be won by the weaker team if nothing unusual happened at all?
Reasoning it through
REASONING #To answer that we need one number: how likely is the better team to win a single game? Not the season record, which includes games against everyone — the probability against this specific opponent, in this round.
Here is the first uncomfortable observation. Playoff teams are the survivors of a selection process. They are, by construction, the good ones. The gap between two playoff teams is far smaller than the gap between a playoff team and the league. A dominant regular-season side might win 65 per cent of its games overall, but against another qualified team, at neutral quality, something like 55 per cent is a realistic figure — and even 60 would be an unusually large edge.
So take 55 per cent and ask the arithmetic question. Over one game, the favourite wins 55 times in 100. Over a best-of-seven, we need the chance of reaching four wins before the opponent does. Summing the ways that can happen gives about 61 per cent.
Sit with that. Six extra games — the entire apparatus of a two-week series, home advantage, adjustments, rest — moved the favourite from 55 to 61. It shifted the odds by about six points, and left a 39 per cent chance the weaker team wins.
Is the series length doing nothing, then? It is doing something, but the shape of the something matters. Run the ladder: one game gives 55.0 per cent, best-of-three 57.5, best-of-five 59.3, best-of-seven 60.8. Each extension helps, and each helps less than the one before. The returns are diminishing sharply, because what a longer series reduces is the variance of the average outcome, and variance falls with the square root of the number of trials, not with the number itself.
That is the whole answer to the puzzle, and it is a fact about arithmetic rather than about sport. To make a series meaningfully decisive you do not need a few more games; you need many times more. Pushing the favourite to 90 per cent at a 55 per cent edge would take dozens of games — roughly a regular season, which is exactly what a regular season is and exactly why its standings are so much more reliable than its playoffs.
Now compound it. A title usually requires four such series. If each is won 61 per cent of the time, winning all four is 0.61 to the fourth power — about 14 per cent. The best team in the league, correctly identified, is an underdog against the field for its own championship. No injury or bad matchup is needed to explain a favourite losing; losing is simply the more likely outcome.
Two honest caveats. First, this treats games as independent coin flips at a fixed probability, which they are not — home advantage alternates, injuries accumulate, and teams adjust between games. Second, the direction of those corrections is not obvious; adjustment might help the better team, while a series long enough for a key player to get hurt cuts the other way. The model is a floor on how much randomness there is, not a full account.
The analogy
THE ANALOGY #Consider a slightly weighted coin — it lands heads 55 times in 100. Flip it seven times and ask whether you can tell it is weighted from the result. You cannot, really; four-three or five-two splits both ways are entirely ordinary. Now flip it a thousand times and the bias is unmistakable. A playoff series is the seven-flip test, and a regular season is the thousand-flip one — which is why we crown champions with the instrument that measures less accurately.
A coin's bias is fixed, whereas two teams genuinely change during a series — adjustments, fatigue and injuries all shift the per-game probability as it goes, so the real process is a sequence of slightly different coins rather than repeated flips of one.
Clarifying the model
THE MODEL #The most common misreading is that this proves upsets are meaningless — that the winner was merely lucky. It does not. It proves something narrower: that a seven-game result is weak evidence about which team was better. A team that wins may well have been better; we simply cannot conclude it from the series, because a 39 per cent event is not rare.
The second misreading runs the other way: if series are so noisy, why not decide titles by regular-season record? Because the noise is largely the point. A postseason that reliably confirmed the standings would be a formality nobody would watch. Tournament design trades accuracy for drama, and the trade is deliberate — which is why leagues extend series lengths for revenue and tradition rather than for measurement, and why formats like the FA Cup embrace single-elimination precisely to maximise upsets.
Worth naming too: seeding advantages exist mostly to nudge these thin margins. Home-court advantage in a deciding game is a real but small edge, and its main function is to convert a coin flip into a slightly weighted one — which, as we have seen, changes rather little.
A picture of it
THE PICTURE #How to readThe bars are a favourite who wins any single game 55 per cent of the time; the line is a stronger favourite at 60 per cent. Both curves rise as the series lengthens and both flatten quickly — extending from one game to seven buys the first favourite about six percentage points and still leaves him losing two series in five.
What became clearer
WHAT CLEARED #A best-of-seven series is a far blunter instrument than its reputation suggests. Because playoff opponents are closely matched by selection, and because random variation shrinks only with the square root of the sample, seven games barely improves on one. The favourite losing is not an anomaly requiring a story; it is the expected frequency of an ordinary outcome, and a champion is better understood as a team that was good and then survived four coin flips.
Where to go next
ONWARD #- Why regular-season standings are so much more reliable, and what sample size would actually be needed to identify the best team confidently.
- How sports differ: baseball's per-game randomness is famously higher than basketball's, which is why its postseason produces more upsets.
- Alternative tournament designs — double elimination, seeded byes, aggregate scoring — and what each trades between accuracy and drama.
Key terms
TERMS #| Term | What it means |
|---|---|
| Best-of-seven | a series won by the first team to four victories, lasting four to seven games. |
| Binomial probability | the chance of a given number of successes in repeated independent trials with a fixed success rate. |
| Regression to the mean | the tendency for extreme results to be followed by more ordinary ones, because the extreme partly reflected luck. |
| Selection effect | here, the narrowing of quality differences that results from only qualified teams reaching the playoffs. |
| Home-court advantage | the measurable but modest boost a team receives playing in its own arena. |
Every term the collection defines is gathered in the glossary.