ハスキー (also called Haski) is an statistical forecast algorithm conceived to be a stock forecaster.

The Haski's algorithm iterate through all the values in a dataset. The process of finding a pattern for a value is:

1. Calculate the deviation which is defined by the following formula:

   deviation = |((f / n * 100) - 100|
   

   Where:

   • f average of a given number of values after the current one
   • n the current value

2. Check if deviation is greater than a given number, if it's not the following steps are skipped.

3. If deviation is greater than 0 an up trend is predicted, otherwise a down trend.

4. Then, fetch a number of values previous to the current value and for each each of those values calculate the back2FrontDeviation defined by the following formula (note that for the first value the back2FrontDeviation is not calculated):

   back2FrontDeviation = round(ln(|((h / z) * 100) - 100|))
   

   Where:

   • h one of those previous values
   • z the value before h

   And:

   • round() round the number to have no decimals

5. Lastly, calculate a hash of all of the back2FrontDeviation values obtained.

I hope that I didn't forget anything 🥲

わからない！