Naar de inhoud
OneKitly

A Secret Santa Draw Fails 63 % of the Time, Whatever the Group Size

Gepubliceerd op 21-10-2026 · 3 min leestijd · Dagelijkse calculators

Lena Hoffmann

Lena Hoffmann — Redacteur wetenschap en onderwijs bij OneKitly

Wiskunde · Natuurkunde

Getoetst aan 2 bronnen

Profiel bekijken →
Kort gezegd

A derangement is a shuffle in which nothing ends up where it started, and the subfactorial counts them: !5 = 44 out of 5! = 120 arrangements, or 36.67 per cent. Push the group to ten and the count becomes 1,334,961 out of 3,628,800 — 36.7879 per cent. At twenty it is 36.78794 per cent. The proportion is converging on 1/e, about 0.367879, and it gets there so fast that by seven people it already agrees to four decimals. So a hat drawn at random works about 36.8 per cent of the time and has to be redrawn the other 63.2 per cent, and that figure barely depends on whether the office has five people or fifty.

Advertentie

Of the 120 ways five people can draw names, 44 give nobody their own — 36.7 %. With ten people it is 36.788 %, and with twenty it is 36.788 %. The proportion stops moving almost immediately.

Why the answer stops depending on the group

Adding a person adds a chance that they draw themselves, and it also adds a name that everyone else might draw instead. The two effects nearly cancel, and the residue is an alternating series whose terms shrink like the reciprocals of factorials — which is exactly the series for 1/e. Because factorials grow so fast, the tail of that series is negligible after a handful of terms, and the answer freezes. Intuition expects a large group to be safer, on the grounds that one person's own name is a smaller fraction of the hat; the arithmetic says the extra opportunities for failure arrive at the same rate.

Redrawing is not the fix people think it is

Drawing again until nobody has their own name does produce a valid assignment, and on average it takes about 2.7 attempts to get there — one divided by 0.368. What it does not produce is a uniformly random valid assignment, if the redraw is done carelessly: swapping just the offending pair, or letting the person who drew themselves pick again from what is left, biases the result in ways that are invisible in a single run and obvious over a few years of the same office party. The clean approaches are to reject the whole draw and start over, or to build a single random cycle through all the names, which is a derangement by construction.

Share
Derangements as a share of all arrangements
PeopleDerangementsArrangementsShare
492437.500 %
54412036.667 %
71,8545,04036.786 %
101,334,9613,628,80036.7879 %
Calculator voor derangementen (subfaculteit) !nBereken de subfaculteit !n — het aantal derangementen, permutaties van n voorwerpen waarbij niets op zijn oorspronkelijke plaats blijft. De tool gebruikt de exacte recursie !n = (n−1)·(!(n−1) + !(n−2)) met grote gehele getallen, en laat de verhouding !n / n! zien, die verbazend snel naar 1/e ≈ 0,3679 convergeert. Die verhouding is de kans dat een willekeurige schudbeweging geen enkel element op zijn plaats laat — het klassieke garderobeprobleem.Probeer de tool →
Advertentie

Veelgestelde vragen

What is a subfactorial exactly?
It counts the permutations of n objects that leave nothing fixed, written !n with the exclamation mark in front. It has a tidy recurrence — !n equals n−1 times the sum of the two previous subfactorials — and an even tidier shortcut: !n is the nearest whole number to n! divided by e, for every n from one upwards. That second fact is why the ratio in the table converges so obviously; the derangement count is essentially a factorial divided by e, rounded, and the rounding is the only thing standing between the sequence and the constant.
Does the same maths cover matching problems generally?
It covers the pure version and stops where the constraints begin. Derangements answer the question of nobody matching themselves, and nothing more: the moment a rule is added — couples must not draw each other, last year's pairs are barred, one person is away — the count changes and the neat convergence to 1/e is gone. Those variants are still countable, usually by inclusion and exclusion, but they no longer have a single memorable answer. The 36.8 per cent is a baseline for the unconstrained case, and it is worth knowing precisely because so many real draws start there before someone adds a rule.

Artikelen die je misschien interesseren

Alle gidsen →

Vergelijkbare tools

Bronnen

Een fout in dit artikel gezien?