# Student's t-Distribution

> The Student's t distribution has fatter tails than the normal. Learn how it is used for small sample tests and to model fat tailed returns, with examples.

Source: https://learn.tradelabsai.com/math/students-t-distribution/  
Track: Math and Statistics · Level: Advanced · Updated: 2026-10-03  
Publisher: TradeLabs AI (https://tradelabsai.com). Education, not financial advice.  
Cite as: TradeLabs Learn, "Student's t-Distribution", https://learn.tradelabsai.com/math/students-t-distribution/

The Student's t distribution looks like a normal distribution but with fatter tails, meaning extreme values are more likely. It was published in 1908 by William Sealy Gosset, a statistician at the Guinness brewery in Dublin, under the pen name "Student". Traders meet it in two places: in statistical tests with small samples, where it gives more honest confidence intervals, and in risk models, where it captures the fat tails of market returns better than the normal distribution.

## Shape and degrees of freedom

The t distribution has one key parameter, the degrees of freedom (ν):

| Degrees of freedom | Tail behaviour |
|---|---|
| 1 | Very fat tails (the Cauchy distribution; no defined mean or variance) |
| 3 to 5 | Fat tails, similar to daily stock returns |
| 10 to 30 | Moderately fat tails |
| Above about 30 | Close to normal |
| Infinite | Exactly normal |

For ν > 2, the variance is ν / (ν minus 2), so a t distribution must be rescaled to match a target standard deviation.

## Use 1: small sample tests

When estimating a mean from a small sample, using the sample standard deviation adds uncertainty. The t distribution accounts for this, giving wider intervals than the normal distribution.

```
t = (sample mean - hypothesised mean) / (s / √n), with n - 1 degrees of freedom
```

| Sample size | 95% critical value (two sided) |
|---|---|
| 5 | 2.78 |
| 10 | 2.26 |
| 30 | 2.05 |
| 100 | 1.98 |
| Large | 1.96 (normal) |

**Example: A small sample interval**
A strategy's 10 monthly returns average 1.2% with a standard deviation of 3%. Standard error = 3% / √10 ≈ 0.95%. Using the t distribution with 9 degrees of freedom (critical value 2.26), the 95% interval is 1.2% ± 2.15%, from about minus 0.95% to +3.35%. Using the normal value 1.96 would give a misleadingly narrower interval. See [Confidence Intervals](https://learn.tradelabsai.com/math/confidence-intervals/).

## Use 2: modelling fat tailed returns

Daily returns of stocks, indices, currencies and crypto have more extreme days than the normal distribution predicts. Fitting a t distribution often gives degrees of freedom between about 3 and 6 for daily returns, capturing those fat tails.

**Example: Tail risk under normal vs t**
A portfolio has daily volatility of 1%. The probability of a daily loss worse than 4% is about 0.003% under a normal distribution (roughly once every 125 years of trading). Under a t distribution with 4 degrees of freedom scaled to the same volatility, it is about 0.24%, roughly once every year and a half. The t model is much closer to what equity markets have actually experienced. See [Fat Tails](https://learn.tradelabsai.com/math/fat-tails/).

## Where traders use the t distribution

| Use | Lesson |
|---|---|
| t tests for strategy returns | [Hypothesis Testing and P-Values](https://learn.tradelabsai.com/math/hypothesis-testing-and-p-values/) |
| Confidence intervals with small samples | [Confidence Intervals](https://learn.tradelabsai.com/math/confidence-intervals/) |
| Fat tailed value at risk and expected shortfall | [Value at Risk (VaR)](https://learn.tradelabsai.com/portfolio/value-at-risk/), [Expected Shortfall (CVaR)](https://learn.tradelabsai.com/portfolio/expected-shortfall/) |
| GARCH models with t distributed errors | [GARCH](https://learn.tradelabsai.com/math/garch/) |
| Monte Carlo simulations with realistic tails | [Monte Carlo Simulation](https://learn.tradelabsai.com/research/monte-carlo-simulation/) |
| Copulas linking fat tailed assets | Portfolio risk models |

## Limits

- **Symmetric:** the standard t distribution has no skew, while equity returns often have negative skew. Skewed t versions exist.
- **Static:** it does not capture volatility clustering on its own; combining it with GARCH helps.
- **Extreme tails:** even t distributions may understate the most extreme events; extreme value theory focuses on the very far tail.

## Frequently asked questions

### What is the Student's t distribution?

A bell shaped distribution similar to the normal but with fatter tails, controlled by its degrees of freedom.

### Why use the t distribution instead of the normal?

For small sample tests, because it accounts for uncertainty in the estimated standard deviation, and for returns, because it captures fat tails.

### What degrees of freedom fit stock returns?

Daily returns are often fitted with about 3 to 6 degrees of freedom, reflecting fat tails.

Next, learn the distribution of win counts in [Binomial and Bernoulli Distributions](https://learn.tradelabsai.com/math/binomial-distribution/).

## Continue learning

- Next lesson: [Binomial and Bernoulli Distributions](https://learn.tradelabsai.com/math/binomial-distribution/)
- Previous lesson: [Lognormal Distribution](https://learn.tradelabsai.com/math/lognormal-distribution/)
- Related: [Lognormal Distribution](https://learn.tradelabsai.com/math/lognormal-distribution/): A lognormal distribution describes values whose log is normal, like prices that cannot go negative. Learn log returns, volatility drag and its option uses.
- Related: [Normal Distribution](https://learn.tradelabsai.com/math/normal-distribution/): The normal distribution is the bell curve behind many financial models. Learn its properties, the 68 95 99.7 rule, where traders use it and why markets break it.
- Related: [Fat Tails](https://learn.tradelabsai.com/math/fat-tails/): Fat tails mean extreme market moves happen far more often than the normal curve predicts. Learn the evidence, the causes, how to measure them and how to manage them.
- Related: [Hypothesis Testing and P-Values](https://learn.tradelabsai.com/math/hypothesis-testing-and-p-values/): Hypothesis tests check whether results are likely due to chance. Learn null hypotheses, test statistics and p values, a strategy test and how p values mislead.
- Related: [Confidence Intervals](https://learn.tradelabsai.com/math/confidence-intervals/): A confidence interval gives a range of plausible values for a statistic. Learn how to calculate them for returns and win rates and how to read them in backtests.
- Related: [Value at Risk (VaR)](https://learn.tradelabsai.com/portfolio/value-at-risk/): Value at risk estimates the loss a portfolio should not exceed with a given confidence over a set period. Learn the three methods, an example and the limits.