This was our implementation for the algorithmic trading challenge by Optiver at HackDelft 2017.
The entire 100 days of trading prices are plotted here.
The lagplot below shows the correlation between the price at second t and t-1. The data was resampled to 30 mins interval.
The autocorrelation plot below again for the series resampled at 30 mins interval shows how the series correlates to a delayed copy of itself as function of delay.
The plot below shows the trend based on the season decomposition provided by
the tsa module in the statsmodel package.
- Greedy
- Breakout
- Moving Average
- Heuristics
- Desperate