Trends/Statistics and risk

Autocorrelation

Whether one day's return tends to be followed by a return in the same direction or the opposite direction.

Statistics and riskBest on 3Y, 5YAlso called serial correlation

Illustration of the autocorrelation shape. It is a sketch, not real price data.

What it shows

Autocorrelation is the correlation of a series with itself shifted by one bar. A positive value means up days tend to follow up days and down days follow down days. A negative value means a move tends to be followed by a reversal.

For most large stocks the value is close to zero. We compute it over the last 252 days.

How we find it

The same code that draws it on your chart follows these steps, with the same result every time.

Daily returns for 252 days.
Pair each return with the next.
Compute the correlation of the pairs.
Plot as a rolling line.

Starting values: Lag 1 bar; Window 252 bars. These are proposed and will be tuned on real price data before launch.

How to read it

The panel shows autocorrelation.
Above zero: moves tend to continue.
Below zero: moves tend to reverse.

Status labels you may see

Label	What it means
Positive	Above 0.1.
Near zero	−0.1 to 0.1.
Negative	Below −0.1.

What cancels it

Nothing cancels it.

What traders check next

Other lags.
Hurst exponent.
Window.

Limits

Small, noisy values.
Statistical noise at ±0.1.
Changes over time.

Try it in forecastviz

Open any ticker and choose Statistics and risk → Autocorrelation in the left menu.
Switch between 3Y, 5Y to see how the picture changes.
Compare with Hurst exponent and Correlation with the index on the same chart.

Open on a ticker

Questions

Why is it near zero?

Prices already include a lot of what is known.

Is this useful?

It shows whether a stock tends to trend or reverse day to day.