Game Length & Format Fairness
How much the length of a game protects the better player — and why win-by-two is worth ten times less than playing to a higher number.
Every point is treated as an independent trial at the same chance, which is what makes the length effect visible in isolation. Real players tire, tighten and adjust, and a longer game gives all of that more room too — none of which is modelled here. Nothing is uploaded.
A longer game is a fairer game
Every extra point is another sample of the same edge, so lengthening the target moves the result steadily towards the better player. A 55% player wins 65.4% of games to seven and 76.4% of games to twenty-five — eleven points of difference, with nothing about either player changed.
| Format | Win by two | Sudden death | The ending is worth |
|---|---|---|---|
| First to 7 | 65.4% | 64.4% | 1.04 pts |
| First to 11 | 68.7% | 67.9% | 0.78 pts |
| First to 15 | 71.3% | 70.7% | 0.64 pts |
| First to 21 | 74.6% | 74.1% | 0.50 pts |
| First to 25 | 76.4% | 76.0% | 0.44 pts |
The benefit peaks in the middle of the skill range — around 59%, worth 14.0 points — and falls away at both ends. An even player has no edge for the extra length to protect, and a dominant one was already winning the short games. Length helps most exactly where the result was most in doubt.
And win-by-two is the same idea, much smaller
Insisting on a two-point separation is one more demand on the weaker player, so it does favour the favourite — by about 0.78 points at first to eleven. That is real and it is small: the difference between a game to seven and a game to twenty-five is more than ten times as large.
| Points won | First to 7 | First to 25 | Length is worth |
|---|---|---|---|
| 50% | 50.0% | 50.0% | 0.00 pts |
| 52% | 56.3% | 61.3% | 5.01 pts |
| 55% | 65.4% | 76.4% | 11.00 pts |
| 60% | 78.7% | 92.6% | 13.89 pts |
| 65% | 88.7% | 98.6% | 9.96 pts |
So if a format feels like a lottery, the length is the thing to change and the ending is almost beside the point. Both effects vanish entirely at even skill — length cannot help a player with no edge to protect, and the gain there is exactly zero rather than merely small, which is the check that this is measuring skill and not the arithmetic.
How to use
- Enter the share of points you win.
- Set the target the game is played to.
- Compare win-by-two against sudden death.
- Compare short formats against long ones.
Frequently asked questions
Does a longer game favour the better player?
Yes, and steadily. Every extra point is another sample of the same edge. A 55% player wins about 65% of games to seven and about 76% of games to twenty-five, with nothing about either player changed.
How much is win-by-two worth?
About three quarters of a point of win probability at first to eleven. It does favour the favourite — a two-point separation is one more demand on the weaker player — but it is a small effect.
So what actually protects the favourite?
The length, by an order of magnitude. The difference between a game to seven and a game to twenty-five is more than ten times what the ending is worth. If a format feels like a lottery, lengthen it rather than fiddling with how it finishes.
Who benefits most from a longer game?
Whoever was closest to a genuine contest. The benefit peaks around a 59% player and falls away at both ends: an even player has no edge to protect, and a dominant one was already winning the short games.
Are short formats really a coin toss?
No, and the complaint overstates it. Even first to seven gives a 55% player about a 65% chance — well clear of a coin. Short formats compress the edge; they do not erase it.
Does sudden death favour the underdog?
Slightly, relative to win-by-two, and only slightly. A sudden-death ending still leaves the better player well ahead, because the whole game before it did most of the work.
What is not modelled here?
Fatigue, nerves and adjustment. Every point is treated as an independent trial at the same chance, which is what makes the length effect visible in isolation — and a longer game gives all of those other things more room too.
Does this send anything anywhere?
No. Every figure is computed in your browser, and nothing is uploaded or stored.
🔒 This tool runs entirely in your browser. Nothing you enter is uploaded, logged, or stored.