# BOCPD

Bayesian Online Change-Point Detection in MQL5. Maintains a probability
distribution over how long the current regime has lasted, and updates it bar by
bar, so a break is flagged as it happens rather than after a lookback window
catches up.

Companion code for the MQL5 article: https://www.mql5.com/en/articles/23482

## What it does

The run-length posterior is the whole method. At every bar the model asks how
likely it is that the current regime started 1 bar ago, 2 bars ago, and so on.
A break shows up as probability mass collapsing from a long run length onto
zero. Nothing here is a threshold on a moving average.

`CBOCPD` in `BOCPDModel.mqh` carries the recursion. Three consumers sit on top
of it, which is the point of the article: the same signal is useful in more
than one place.

- `BOCPDRegime.mq5` plots the break probability directly, as a monitor.
- `BOCPDAdaptiveMA.mq5` is a moving average that resets itself when the model
  says the regime changed, instead of dragging stale history across the break.
- `BOCPDRiskOverlay.mq5` uses it as a meta-layer over position sizing rather
  than as an entry signal.

## Layout

```
Include/BOCPD/BOCPDModel.mqh          the recursion, CBOCPD
Indicators/BOCPD/BOCPDRegime.mq5      break probability monitor
Indicators/BOCPD/BOCPDAdaptiveMA.mq5  self-resetting adaptive average
Experts/BOCPD/BOCPDRiskOverlay.mq5    risk meta-layer
```

Copy the folders into your terminal's `MQL5` directory and compile.

## Disclaimer

Educational code. Past behaviour of any model or dataset says nothing about
future results. Test on your own data and broker conditions before drawing
conclusions.