Model #1 · +5% / 5d — Data Investors

Hit rate · picks vs scored universe

graded daily

Realized hit rate of the model's top picks (prediction > 0.10) against everything it scored that day — the gap is the edge.

AUC by evaluation set

walk-forward

Each point is one retrain. Out-of-time (indigo) is the honest line; a widening gap to train-eval means overfit.

Feature importance

train June 30, 2026

CatBoost importances from the latest trained predictor; top 20 of 20 features.

Market-level · predicted vs actual

monthly, graded

Average predicted probability against the realized hit rate across everything scored — the lines tracking each other means the model is calibrated.

Top-10 picks · results

monthly, graded

Realized hit rate of the model's 10 highest predictions each day vs the market — the gap above the market line is the edge.

Top 10% of predictions · results

monthly, graded

Same view for the top decile each day — broader than top-10, so a steadier read on whether ranking works.

Feature importance trends

one line per feature

Each feature's importance across the walk-forward retrains — the line-plot view of the table below. The top 10 features draw by default; use the legend to toggle others on.

Feature importance by month

top 22 of 22

Feature	Mean	2025-06	2025-07	2025-08	2025-09	2025-10	2025-11	2025-12	2026-01	2026-02	2026-03	2026-04	2026-05	2026-06
volatility_atr	9.55	9.6	10.0	10.2	10.0	9.8	9.4	9.4	9.6	9.5	8.8	9.4	9.1	9.2
volatility_bbw	6.68	6.4	6.5	6.4	7.3	7.6	6.8	6.6	6.5	6.6	6.1	6.6	6.8	6.6
close_pe_calc	6.63	6.5	6.8	6.3	6.7	6.7	7.1	6.7	6.3	7.2	6.5	6.3	6.6	6.4
days_to_next_quarterly_filing	6.05	6.4	6.5	6.4	6.4	6.1	6.2	5.9	5.8	5.6	5.7	6.1	5.8	5.7
days_since_quarterly_filing	6.04	6.1	6.3	6.3	5.6	6.1	5.9	6.3	6.1	6.2	6.1	5.8	5.9	5.9
close_ps_calc	5.91	6.0	5.4	5.2	5.2	6.0	6.1	5.5	6.6	5.7	6.1	6.4	6.6	5.8
trend_adx	5.74	5.8	5.8	6.0	6.1	5.8	5.8	5.8	5.5	5.5	5.5	5.3	5.8	6.2
volume_stddev_20	4.79	5.2	5.2	5.2	4.6	4.3	4.8	4.5	5.0	4.8	4.9	5.1	4.2	4.5
momentum_rsi	4.79	4.1	4.3	4.7	4.5	4.8	4.3	4.8	5.1	5.1	5.2	5.3	4.8	5.2
volume_sma_20	4.75	4.7	4.5	4.3	5.0	4.8	5.1	4.9	4.9	5.0	4.7	4.5	4.6	4.7
trend_macd	4.57	5.1	4.9	4.4	4.4	4.8	4.2	4.6	4.5	4.8	4.4	4.4	4.5	4.3
trend_macd_diff	4.28	4.3	4.3	4.1	4.6	4.5	4.1	4.4	4.0	4.0	4.5	4.2	4.5	4.3
close_stddev_10	4.17	4.0	4.3	4.0	3.9	3.7	4.3	4.3	4.4	4.1	4.2	4.5	4.2	4.3
close_stddev_20	4.14	3.9	4.0	4.1	3.6	4.1	4.0	4.3	4.2	4.6	4.3	4.1	4.2	4.4
volume_sma_5	3.91	3.8	3.7	4.0	4.3	4.2	3.9	3.6	3.7	3.8	3.8	4.0	4.3	3.6
min_low_10	3.25	2.7	3.0	3.2	3.7	3.0	3.3	3.2	3.5	3.3	3.3	2.8	3.2	3.9
close_sma_20	2.88	2.7	2.3	3.3	2.6	3.2	2.8	2.5	2.9	3.1	3.4	2.9	2.8	2.9
momentum_roc	2.88	3.0	2.8	2.9	2.9	2.6	3.1	2.8	2.5	2.6	3.1	3.0	3.1	3.0
max_high_10	2.45	2.6	2.5	2.1	2.2	1.7	2.6	2.8	2.6	2.1	2.5	2.9	2.4	3.0
others_dr	2.44	2.3	2.3	2.4	2.4	2.3	2.4	2.5	2.5	2.4	2.7	2.4	2.6	2.4
close_sma_5	2.15	2.4	2.1	2.2	2.4	2.1	2.3	2.0	2.1	2.1	1.9	2.2	2.1	2.0
close_sma_10	1.97	2.3	2.6	2.2	1.7	1.8	1.5	2.4	1.8	1.9	2.0	1.8	2.0	1.6

One row per feature, one column per walk-forward retrain; rows ordered by mean importance. Watch for features that fade or spike between months — unstable importances usually mean a regime change or a data gap.

Walk-forward predictors

13 trains

Final train	Trees	LR	Top feature	OOT AUC
June 30, 2026	1000	0.070	volatility_atr	1.000
May 31, 2026	1000	0.071	volatility_atr	0.972
April 30, 2026	999	0.072	volatility_atr	0.657
March 31, 2026	1000	0.073	volatility_atr	0.156
Feb. 28, 2026	997	0.074	volatility_atr	0.768
Jan. 31, 2026	1000	0.075	volatility_atr	0.700
Dec. 31, 2025	1000	0.075	volatility_atr	-1.000
Nov. 30, 2025	1000	0.075	volatility_atr	0.770
Oct. 31, 2025	999	0.075	volatility_atr	0.728
Sept. 30, 2025	1000	0.075	volatility_atr	0.649
Aug. 31, 2025	1000	0.075	volatility_atr	0.695
July 31, 2025	1000	0.075	volatility_atr	0.746
June 30, 2025	993	0.076	volatility_atr	0.664

One predictor per retrain date; predictions always come from the newest trained one.