hello, i made my first submission but i am not in the leaderboard, which submissions are eligible for the leadearboard?
If you just submitted your first “daily score” will be out on Thursday, but a single round is 4 weeks long. The daily scores from now until your first round resolves don’t really matter, only the final score at the end of the 4 weeks.
The leaderboard is the weighted average over the last 20 rounds. So you are going to need to wait at least 24 weeks before you will have a full idea of your position on the leaderboard. Empty rounds have a -0.1 correlation score for every round you don’t submit.
This is a long game and the leaderboard isn’t what you should be focusing on. With the rounds being 4 weeks long, you are going to have to be patient to see the results of your modelling work.
Thank you very much for your reply!really appreciated …my next concern is how to optimize the numbers appeared in ‘‘performance’’, ‘‘risk’’, ‘‘mmc’’(3 each)?
all i found are these formulas :
def _score(sub_df: pd.DataFrame) → np.float32:
“”“Calculates Spearman correlation”""
return spearmanr(sub_df[“target”], sub_df[“prediction”])
# Calculate metrics corrs = df.groupby("era").apply(_score) payout_raw = (corrs / 0.2).clip(-1, 1) spearman = round(corrs.mean(), 4) payout = round(payout_raw.mean(), 4) numerai_sharpe = round(sharpe_ratio(corrs), 4) mae = mean_absolute_error(df["target"], df["prediction"]).round(4)
Shouldn't be a formula for each of 9 metrics (performance,risk,mmc)?
Take a read of this, should answer your questions on the definitions of the metrics