Confidence interval for a mean
A range around the sample mean that, built this way on repeated samples, covers the true mean the stated share of the time. Wider means less certain.
01The formula
- the sample mean
- the sample standard deviation
- the number of observations
- the normal quantile for your confidence level: 1.96 for 95%
- one minus the confidence level
02See it move
03Where it is used
- Quant researcherReports a range for a strategy's mean return instead of a single number.
- Risk managerPuts bounds on estimated loss rates or average costs.
- Portfolio managerChecks whether the interval for excess return excludes zero before allocating.
04Why traders use it
A backtest average without an interval invites overconfidence. If the interval includes zero, the data have not shown an edge.
05Worked example
A sample mean of 0.08% per day with volatility 1.0% over 250 days gives a 95% interval from −0.044% to 0.204%, a half-width of 0.124% using z = 1.960. It still includes zero.
06How it is built
- The standard error of is .
- For large , is close to a standard normal.
- So with probability it lies between and .
- Rearrange for : .
07The common mistake
Saying there is a 95% chance the true mean is in this particular interval. The 95% describes the method over many samples; with small samples you should also use the t quantile, not z.
08In Python and LaTeX
import numpy as np from scipy.stats import norm xbar, s, n, c = 0.0008, 0.01, 250, 0.95 h = norm.ppf(1 - (1 - c)/2) * s / np.sqrt(n) print(round(xbar - h, 6), round(xbar + h, 6))
\bar{x} \pm z_{1 - \alpha/2}\,\frac{s}{\sqrt{n}}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.