Geometric distribution
How many attempts you need until the first success, when each attempt succeeds with the same chance p. Short waits are most likely, but long ones keep happening.
01The formula
- the attempt on which the first success happens
- a particular attempt number: 1, 2, 3, …
- the chance each attempt succeeds
- the chance the first k − 1 attempts all fail
02See it move
03Where it is used
- Execution traderEstimates how many times an order must be re-posted before it fills.
- Quant researcherModels the length of a losing streak: the number of trades until the next winner.
- Market makerJudges how many quote updates pass before one is hit.
04Why traders use it
It gives the odds on waiting. Traders who expect the next success soon after a run of failures are wrong: the chance is the same p every attempt.
05Worked example
Each attempt succeeds with chance 20%. The chance the first success comes on attempt 5 is 8.19%, and the chance you succeed within 5 attempts is 67.2%. On average you need 5.00 attempts.
06How it is built
- For the first success to land on attempt , the first attempts must fail and attempt must succeed.
- By independence that has probability .
- The chance of no success in attempts is , so .
- Summing over all gives the mean .
07The common mistake
Believing a success is 'due'. The geometric distribution is memoryless: after ten failures the chance of needing more than five further attempts is the same as it was at the start.
08In Python and LaTeX
from scipy.stats import geom p, k = 0.2, 5 print(round(geom.pmf(k, p), 4), round(geom.cdf(k, p), 4), geom.mean(p))
P(X = k) = (1 - p)^{k - 1} p, \qquad P(X \le k) = 1 - (1 - p)^k, \qquad E[X] = \frac{1}{p}09Where it fits
Rather not code the maths yourself? Describe your trading idea and let AI build the bot.
Build a bot with AgenKitThe worked example and the model run the same code, so every number matches. Models simplify markets on purpose. This is education, not financial advice.