Poisson distribution
The number of events in a fixed time window when they arrive independently at a steady average rate λ. Its mean and its variance are both λ.
01The formula
- the number of events in a fixed interval, such as trades in a minute
- a particular count: 0, 1, 2, …
- the average number of events per interval
- k factorial: 1 × 2 × … × k
02See it move
03Where it is used
- Market makerModels the arrival of orders that hit their quotes as a Poisson stream, as in the classic market-making models.
- Execution traderEstimates how many trades will print in the next minute, to judge how long an order will take.
- Risk managerCounts rare events such as large jumps or system faults per month.
04Why traders use it
Order flow arrives in counts. The Poisson gives a quick baseline: if what you see is far more bursty than Poisson, something is clustering the events.
05Worked example
Orders arrive at 4.0 per minute. The chance of exactly 7 in a minute is 5.95%, and of at least 7 is 11.07%. The mean is 4.0 and the standard deviation 2.00.
06How it is built
- Split the minute into tiny slices, each with chance of one event.
- The count is then binomial: .
- Let : and .
- So , with mean and variance both .
07The common mistake
Assuming market events are Poisson without checking. Real trades cluster: one large order triggers others, so the variance of counts is usually much bigger than the mean.
08In Python and LaTeX
from scipy.stats import poisson lam, k = 4, 7 print(round(poisson.pmf(k, lam), 4), round(poisson.sf(k - 1, lam), 4)) # P(X=k), P(X>=k)
P(X = k) = \frac{\lambda^k e^{-\lambda}}{k!}, \qquad E[X] = \operatorname{Var}(X) = \lambda09Where 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.