I'm trying to implement a very simple dollar neutral long-short algo based on operating leverage (long firms with high OL, short those with little). (Based loosely on this: http://rof.oxfordjournals.org/content/15/1/103.abstract)
The strange thing is this: two times in the attached 14 year back test the algo behaves bizarrely. Once it doubles in value in a day, and later it loses 80% of its value in a day.
Is this an edge case caused by my code, or something to do with the backtest data?