**AEdge Distribution Bands **(or *ADB*) is an on-chart tool that estimates the future distribution of asset returns to project where price may end up in the near future. The bands are similar to a credible interval from Bayesian statistics and show the estimated tail-ends of the predicted distribution.

Built into the tool is a mean reversion hypothesis test that tests the null hypothesis that the residuals of the estimate mean revert back to the estimate. When the bands are filled with a yellow (default) color the test fails to reject the null hypothesis, meaning the assumption of reversion of price back to the estimate does not hold.

The way ADB works is it models asset returns as a skewed, fat-tailed distribution (very non-Gaussian) to capture the behavior of most securities, including cryptocurrencies. We know that volatility tends to cluster since there is autocorrelation in squared asset returns (i.e. low volatility tends to follow low volatility and high volatility tends to follow high volatility) and so the model tries to capture this in its estimate by centering the bands around a volatility weighted moving average of price where the rolling lookback window size is continuously adjusted based on the estimated autocorrelation, as well as by scaling the distribution in a way that projects where price may end up in the future.

There are multiple possible uses for this indicator in your analysis. Here are a few examples:

- Complementing other analysis. If your analysis tells you price may end up at a certain level, you can use these bands to see if they agree about the price range (i.e., if that price is within the range of the bands or not).
- Inferring if price currently is not exhibiting mean reversion behavior. It does not mean price will not move back to the mean. It simply means that where price ends up is more random and reversion to the mean cannot confidently be assumed.
- Identifying outlier moves in price.

The indicator settings have one main parameter: *Number of Bars*, under Lookback Window Size. This parameter sets the number of bars to use for fitting the model.