Thanks for reading my pairs trading paper.

I've seen a few papers on pairs trading (mean reversion) with currency pairs. I have not looked into it yet.

The term "data mining" is considered a "dirty word" in quantitative finance. The reason that data mining is considered to be a bad thing is that it easy to build an overfit model that is very accurate on the in-sample data with very bad results on the out-of-sample data.

If you fit a linear model on past data there is a good chance that the model will not do well on out-of-sample data.

There are several reasons for this. Financial data and statistical behavior is constantly changing. Correlation between two stocks is consistent between two adjacent time periods in only about 50% of the cases. Cointegration is persistent in only about 40% of the cases.

Currency pairs have limited data, so it is more difficult to draw conclusions. I would look at the behavior of your model with out-of-sample data.