NKE state
Returns, volatility, drawdown, moving-average gaps, volume, range, and close location.
MODEL 3 · STEP 3Build
Build
the features.
Create 38 trailing-only price, market, peer, macro-proxy, and filed-fundamental values without backfilling future information.
MARKET CONTEXT · OUTSIDE MODEL FREEZENKE
These readings update independently. They are not model inputs, deployment evidence, or a trading recommendation.
- Latest price
- $38.40 · fallback
- 20D momentum
- -7.9% Close-to-close · 20 completed sessions
- RSI 14
- 40.1 neutral · Wilder method
- Relative volume
- 0.7× Versus the prior 20 completed sessions
TECHNICALS THROUGH · TIMESTAMPED FALLBACK
Measure only what existed by the NKE decision close.
A 63-session warm-up produces 2,450 complete rows. Scaling, missing-value handling, model tuning, and calibration are relearned inside each training window.
JOINED MARKET ROWS2,513 sessions
FEATURE-READY2,450 rows
ELIGIBLE LABELS2,250 rows
38 SELECTED FEATURES
Five families, one decision clock.
Every value can be rebuilt from completed sessions and filings available by that date.
Market & sector
SPY and XLY trend, volatility, and NKE relative performance.
Peers
LULU, DECK, VFC, and DKS return breadth and composite momentum.
Macro proxies
FXI, TLT, and UUP trailing returns on the same completed-session boundary.
Filed fundamentals
Revenue, margins, inventory, filing age, and post-filing indicators eligible from the next completed session.
Rolling values end at session t. Fundamental values begin only on the next completed trading session after the SEC filing date. Same-session market closes are used only because the prediction is made after that completed close. No later restatement is backfilled into an earlier decision date.
| # | Feature | # | Feature |
|---|---|---|---|
| 1 | nke_return_1d | 20 | nke_excess_xly_21d |
| 2 | nke_return_5d | 21 | peer_return_5d |
| 3 | nke_return_21d | 22 | peer_return_21d |
| 4 | nke_return_63d | 23 | peer_breadth_21d |
| 5 | nke_volatility_21d | 24 | fxi_return_5d |
| 6 | nke_volatility_63d | 25 | fxi_return_21d |
| 7 | nke_volatility_ratio_21_63 | 26 | tlt_return_5d |
| 8 | nke_drawdown_63d | 27 | tlt_return_21d |
| 9 | nke_ma_gap_21d | 28 | uup_return_5d |
| 10 | nke_ma_gap_63d | 29 | uup_return_21d |
| 11 | nke_volume_z_63d | 30 | filing_revenue_yoy |
| 12 | nke_intraday_range_pct | 31 | filing_gross_margin |
| 13 | nke_close_location | 32 | filing_net_margin |
| 14 | spy_return_5d | 33 | filing_inventory_yoy |
| 15 | spy_return_21d | 34 | filing_gross_margin_yoy_delta |
| 16 | spy_volatility_21d | 35 | days_since_filing_norm |
| 17 | xly_excess_spy_5d | 36 | post_filing_5d |
| 18 | xly_excess_spy_21d | 37 | post_filing_21d |
| 19 | nke_excess_xly_5d | 38 | annual_filing_flag |
POST-FREEZE TECHNICAL CHALLENGER
RESEARCH ONLYHit rate rose, while calibrated probability error worsened.
A fixed benchmark adds 20-session momentum, Wilder RSI 14 with overbought and oversold flags, prior-20 relative volume, and price-volume confirmation. It is scored on the last 252 eligible labels without changing this model's frozen nine-gate decision.
- Brier error
- 0.257081 → 0.259790 Lower is better
- Brier gain
- −0.002708 95% paired interval −0.014798 to +0.009190
- 0.5 hit rate
- 43.7% → 46.0% Threshold accuracy alone cannot approve a model
- Challenger AUC
- 0.482 Below the 0.50 no-skill line
Promotion rule: not passed. The Brier confidence interval includes zero and discrimination remains below 0.50, so the readings stay a live context layer while genuinely new sessions accumulate.
Open formulas and audit filesNEXTStep 4 · Train through time
Continue through the audit-corrected Model 3 workflow.