Girko Statistic
The second model maintains positive excess predictability for about 99% of the test period. A buy-and-hold benchmark can be substituted for random guessing to test whether the model adds value over passive exposure. The Girko statistic mea…
1 sources - 5 claims
The second model maintains positive excess predictability for about 99% of the test period. A buy-and-hold benchmark can be substituted for random guessing to test whether the model adds value over passive exposure. The Girko statistic measures how much more predictable a model's equity curve is than a random guessing benchmark. Positive excess predictability means the model is outperforming random guessing at that point in time. The random benchmark uses uniformly generated directional signals mapped into -1 and +1.