The insight library Every night a deterministic miner tests every feature against every target it is allowed to touch, with as-of joins, false-discovery control and a time-ordered holdout. Survivors land here, graded; failures stay, labelled. How it works
promising 6 weak 34 baseline_only baseline_only A game insight that predicts the result (who wins, by how much) rather than the result against the betting line. Real, but the line already prices it, so it is not an edge. glossary = predicts the result, not the result against the line what the grades meanGrade robust: significant after correction, out of sample, costs, stable, holdout agrees. promising: significant with out-of-sample excess, holdout not decisive. weak: screened but failed out of sample or the holdout. failed: was promising, no longer. Retired rows stay visible. glossary
promising Temperature predicts corn 21d return CORNBELT_NE.temp_mean_c[zscore(window=30)] lag0 → CORN.close fwd21d · effect -0.011 · OOS excess +0.82% · holdout fail · used 0× promising Home form predicts margin same-daybaseline_only NFL.form_home[game] lag0 → NFL.home_margin fwd0d · effect 0.351 · OOS excess 6.480 home_margin · holdout pass · used 0× promising Away form predicts home marginbaseline_only NFL.form_away[game] lag0 → NFL.home_margin fwd0d · effect -0.284 · OOS excess 3.647 home_margin · holdout pass · used 0× promising Home form predicts win same-daybaseline_only NFL.form_home[game] lag0 → NFL.home_win fwd0d · effect 0.013 · OOS excess 0.216 home_win · holdout pass · used 0× promising Away form reduces home winbaseline_only NFL.form_away[game] lag0 → NFL.home_win fwd0d · effect -0.006 · OOS excess 0.115 home_win · holdout pass · used 0× promising WTI 5-day vol mean reversion WTI.close[rolling_logret(window=5)] lag0 → WTI.close fwd10d · effect -0.332 · OOS excess +2.45% · holdout pass · used 54× weak Dallas wind gust predicts oil 21d DALLAS.wind_gust_max_kmh[zscore(window=30)] lag0 → WTI.close fwd21d · effect -0.009 · OOS excess +1.49% · holdout fail · used 76× weak Home rest → margin (null)baseline_only NFL.rest_home[game] lag0 → NFL.home_margin fwd0d · effect -0.038 · OOS excess -0.295 home_margin · holdout fail · used 0× weak Away rest → margin (null)baseline_only NFL.rest_away[game] lag0 → NFL.home_margin fwd0d · effect -0.049 · OOS excess -0.462 home_margin · holdout fail · used 0× weak Rest differential predicts game marginbaseline_only NFL.rest_diff[game] lag0 → NFL.home_margin fwd0d · effect 0.170 · OOS excess -1.051 home_margin · holdout pass · used 0× weak Form differential predicts marginbaseline_only NFL.form_diff[game] lag0 → NFL.home_margin fwd0d · effect 0.319 · OOS excess 8.483 home_margin · holdout pass · used 0× weak Home rest → win (null)baseline_only NFL.rest_home[game] lag0 → NFL.home_win fwd0d · effect -0.001 · OOS excess -0.016 home_win · holdout fail · used 0× weak Away rest → win (null)baseline_only NFL.rest_away[game] lag0 → NFL.home_win fwd0d · effect -0.001 · OOS excess -0.104 home_win · holdout fail · used 0× weak Rest diff → win (unstable)baseline_only NFL.rest_diff[game] lag0 → NFL.home_win fwd0d · effect 0.000 · OOS excess -0.005 home_win · holdout fail · used 0× weak Same-game form predicts home winbaseline_only NFL.form_diff[game] lag0 → NFL.home_win fwd0d · effect 0.010 · OOS excess 0.234 home_win · holdout pass · used 0× weak IA cornbelt temp z-score → corn return CORNBELT_IA.temp_mean_c[zscore(window=30)] lag0 → CORN.close fwd21d · effect -0.012 · OOS excess +1.08% · holdout fail · used 0× weak Houston max temp predicts oil 21d GULF_HOUSTON.temp_max_c[zscore(window=30)] lag3 → WTI.close fwd21d · effect -0.007 · OOS excess +0.13% · holdout fail · used 0× weak NE cornbelt temp z-score → corn return CORNBELT_NE.temp_mean_c[zscore(window=30)] lag1 → CORN.close fwd21d · effect -0.013 · OOS excess +0.83% · holdout fail · used 0× weak Port Arthur temp predicts oil 21d GULF_PORT_ARTHUR.temp_max_c[zscore(window=30)] lag3 → WTI.close fwd21d · effect -0.007 · OOS excess −0.19% · holdout fail · used 0× weak Dallas mean temp predicts oil 21d DALLAS.temp_mean_c[zscore(window=30)] lag3 → WTI.close fwd21d · effect -0.006 · OOS excess −0.37% · holdout fail · used 0× weak Kalshi price predicts marginbaseline_only NFL.kalshi_home_24h[game] lag0 → NFL.home_margin fwd0d · effect 10.648 · OOS excess — · holdout pass · used 0× weak Kalshi price predicts winbaseline_only NFL.kalshi_home_24h[game] lag0 → NFL.home_win fwd0d · effect 0.354 · OOS excess — · holdout pass · used 0× weak Home rest beats market NFL.rest_home[game] lag0 → NFL.beat_kalshi fwd0d · effect 0.014 · OOS excess — · holdout pass · used 0× weak Away rest beats market NFL.rest_away[game] lag0 → NFL.beat_kalshi fwd0d · effect 0.002 · OOS excess — · holdout fail · used 0× weak Rest difference beats market NFL.rest_diff[game] lag0 → NFL.beat_kalshi fwd0d · effect 0.035 · OOS excess — · holdout pass · used 0× weak Home form beats market NFL.form_home[game] lag0 → NFL.beat_kalshi fwd0d · effect 0.004 · OOS excess — · holdout pass · used 0× weak Away form beats market NFL.form_away[game] lag0 → NFL.beat_kalshi fwd0d · effect -0.003 · OOS excess — · holdout fail · used 0× weak Form difference beats market NFL.form_diff[game] lag0 → NFL.beat_kalshi fwd0d · effect 0.003 · OOS excess — · holdout pass · used 0× weak Kalshi home beats self NFL.kalshi_home_24h[game] lag0 → NFL.beat_kalshi fwd0d · effect -0.646 · OOS excess — · holdout fail · used 0× weak Home out injuries hurt marginbaseline_only NFL.inj_out_home[game] lag0 → NFL.home_margin fwd0d · effect -0.173 · OOS excess — · holdout fail · used 0× weak Away out injuries aid home marginbaseline_only NFL.inj_out_away[game] lag0 → NFL.home_margin fwd0d · effect 0.173 · OOS excess — · holdout pass · used 0× weak Injury imbalance unclear on marginbaseline_only NFL.inj_out_diff[game] lag0 → NFL.home_margin fwd0d · effect -0.208 · OOS excess — · holdout pass · used 0× weak Home questionable injuries boost marginbaseline_only NFL.inj_q_home[game] lag0 → NFL.home_margin fwd0d · effect 1.018 · OOS excess — · holdout pass · used 0× weak Away questionable injuries negligiblebaseline_only NFL.inj_q_away[game] lag0 → NFL.home_margin fwd0d · effect 0.234 · OOS excess — · holdout fail · used 0× weak Home out injuries reduce winbaseline_only NFL.inj_out_home[game] lag0 → NFL.home_win fwd0d · effect -0.009 · OOS excess — · holdout fail · used 0× weak Away out injuries aid home winbaseline_only NFL.inj_out_away[game] lag0 → NFL.home_win fwd0d · effect 0.008 · OOS excess — · holdout pass · used 0× weak Injury differential unclear on winbaseline_only NFL.inj_out_diff[game] lag0 → NFL.home_win fwd0d · effect -0.010 · OOS excess — · holdout pass · used 0× weak Home questionable injuries boost winbaseline_only NFL.inj_q_home[game] lag0 → NFL.home_win fwd0d · effect 0.047 · OOS excess — · holdout pass · used 0× weak Away questionable injuries negligiblebaseline_only NFL.inj_q_away[game] lag0 → NFL.home_win fwd0d · effect 0.010 · OOS excess — · holdout fail · used 0× weak Houston mean temp predicts oil 21d GULF_HOUSTON.temp_mean_c[zscore(window=30)] lag3 → WTI.close fwd21d · effect -0.006 · OOS excess −0.80% · holdout fail · used 15×