AEdge Directional Change Probability calculates the probability that price will soon change direction from either up to down/sideways, or down to up/sideways. It does this by measuring cyclical events, taking both price and time into consideration.
For the curious: This calculation assumes these events are a type of Poisson process. We know the average number of events per time unit (which is almost constant) and the events are mutually exclusive (only one event can happen at the same time) and independently distributed (no significant autocorrelation). It is therefore a fair assumption to make.
There are multiple possible uses for this indicator in your analysis. Here are a few examples:
- Seeing when it is most likely for a move to not have much room left to develop before changing direction from up to down, up to sideways, down to up, or down to sideways. We cannot know if price will turn around or move sideways, however.
- Identifying outlier moves in price.
The indicator settings have three main parameters:
- The first parameter, Resolution, changes how sensitive the model is when defining the up events and down events. The higher the resolution, the more sensitive it becomes to define these events. The lower the resolution, the more infrequent but also more defined the events become. The resolution defaults to 50.
Note: This is not a lookback window. - The second parameter, First Significance Level (%), under Statistical Inference, determines when we want to pay attention to a move that is somewhat outside of what is considered normal. This defaults to 80%.
- The third parameter, Second Significance Level (%), under Statistical Inference, determines when we want to pay more attention to a move that is way outside of what is considered normal. This defaults to 95%.