It's not a _huge_ deal -- I imagine that particular pair is probably a money-maker (assuming you've properly subtracted out slippage and transaction costs) -- but the out-of-sample result here is deceivingly rosy. It looks like you've measured the cointegration test on the full dataset and picked the most highly cointegrated pair, which would include the "out-of-sample" data. Nevertheless, this was interesting. Thanks!
It's just a reverse heatmap of the Augmented Dickey-Fuller test. In this case, the colder (more blue) squares represent the equity pairs that are more likely to be co-integrated.
It's not a _huge_ deal -- I imagine that particular pair is probably a money-maker (assuming you've properly subtracted out slippage and transaction costs) -- but the out-of-sample result here is deceivingly rosy. It looks like you've measured the cointegration test on the full dataset and picked the most highly cointegrated pair, which would include the "out-of-sample" data. Nevertheless, this was interesting. Thanks!
RockinInRhythm 1 year ago
Awesome!!
psecondo 2 years ago
@psecondo what are that colored squares anyway? what have that to do with statiscal arbitrage?
joshuademoraes 1 year ago
@joshuademoraes
It's just a reverse heatmap of the Augmented Dickey-Fuller test. In this case, the colder (more blue) squares represent the equity pairs that are more likely to be co-integrated.
twills10000 1 year ago
Any more posts?
ip916a4bb 2 years ago