Chapter 08 :- 02_vestorized_backtest.ipynb

[Quantuary]

Line 11

long_signals = ((predictions
                .where(predictions > 0)
                .rank(axis=1, ascending=False) > N_LONG)
                .astype(int))
short_signals = ((predictions
                  .where(predictions < 0)
                  .rank(axis=1) > N_SHORT)
                 .astype(int))

• wouldn't we want to long stock in the Top 15 highest expected return? e.g. Rank 1 to 15 with the highest expected return rank first (descending order). Hence we need rank <= 15 ?
• and short the lowest expected return stock from 1 to 15 again, due to lowest return rank first?

[stefan-jansen]

That's correct.