Xiphers Research·← Back to indexDEEN

RV3 — will today be a trend day? · ES

ES · decision at 09:30 New York · 1607 daily bars up to 2026-08-14

Verdict — ES

Can we tell at 09:30 whether today will be a trend day (RTH body ratio ≥ 0.6)?

Criterion not met — BSS -1.9 % [-3.6 ; -0.3], AUC 50.3 %, accuracy 61.6 % against a base rate of 36.2 %, n = 904 days pooled forward.

Criterion: BSS > 2 and AUC ≥ 57, pooled forward across 2023–2026

Setup
Decision time09:30 New York — bars up to t−1 complete, the opening print and every tick of session t before 09:30
Targetbody ratio br = |rc − ro| / (rh − rl) ≥ 0.6
Data window2020-05-29 to 2026-08-14 · 1607 daily bars · last day 2026-08-14
Walk-forwardDev ≤ 2022 (definitions only), test 2023 / 2024 / 2025 / 2026, expanding with a yearly refit
ModelL2 logistic regression on 15 preregistered set C features, standardised on the training window; C from inner validation (the last training year). Features: gap_atr, abs_gap_atr, on_range_atr, ln_on_relr, on_vol_rel, on_delta_norm, open_pos_on, on_ext_recency, pre_range_share, on_high_vs_rh1_atr, on_low_vs_rl1_atr, open_pos_va, prev_ln_relr_rth, prev_ln_relr_mean5, prev_atr5_atr20
HoldoutThe PX holdout from 2026-05-01 concerns the PX theses on the tick store. This study is a walk-forward over 2023 to 2026 on daily data — explicitly approved by the user for regime rounds 1 and 2. Every test year is computed from data before it only; no fit ever sees its own test slice.

Test years and pooled forward

Every row is a forecast over the same days. Base rate, majority and persistence are the baselines; the model has to beat them, otherwise it does not count.

RowYearnBase Rate %Acc %AUC %BrierBSS %
Base Rate202324639.860.250.00.23970.0
Majority202324639.860.2
Persistence202324639.860.254.70.23750.9
Model (LogReg)202324639.854.148.50.2516-5.0
Base Rate202425033.266.850.00.22620.0
Majority202425033.266.8
Persistence202425033.266.850.50.2283-0.9
Model (LogReg)202425033.265.254.60.2265-0.1
Base Rate202525137.862.250.00.23530.0
Majority202525137.862.2
Persistence202525137.862.255.30.23281.0
Model (LogReg)202525137.861.846.30.2390-1.6
Base Rate202615732.567.550.00.22260.0
Majority202615732.567.5
Persistence202615732.567.546.80.2264-1.7
Model (LogReg)202615732.567.553.60.2232-0.3
Base Ratepooled90436.263.849.90.23170.0
Majoritypooled90436.263.8
Persistencepooled90436.263.852.60.23170.0
Model (LogReg)pooled90436.261.650.30.2362-1.9

Block bootstrap (20-day blocks, 1000 runs, pooled forward): BSS 95 % interval [-3.6 ; -0.3].

Confusion matrix pooled: true 0 correct 532 of 577, true 1 correct 25 of 327 (recall 0 = 92.2 %, recall 1 = 7.6 %).

Regularisation chosen per refit — 2023: C = 0.1, n_train = 580 · 2024: C = 0.01, n_train = 826 · 2025: C = 0.01, n_train = 1076 · 2026: C = 0.01, n_train = 1327.

Calibration and sharpness

Decilenmean p %observed %difference
19130.741.811.1
29135.229.7-5.5
39136.830.8-6.1
49137.839.61.7
59038.840.01.2
69040.033.3-6.6
79041.337.8-3.5
89043.031.1-11.9
99045.736.7-9.0
109053.441.1-12.3

Sharpness: p ≥ 0.7 on 0.0 % of the days (n = 0), hitting — % there · p ≤ 0.3 on 3.2 % (n = 29), hitting 58.6 % there.

How to read this report
Base Ratethe unconditional frequency in the training window, and at the same time the constant comparison forecast.
Baselinethe number a model has to beat. Three of them here: base rate, majority (always the more frequent class) and persistence.
Persistencethe forecast “today like yesterday” — the value of the same target on the previous day, turned into a rate.
BSSBrier skill score: what percentage of the base rate constant's Brier score the model saves. 0 means equally good, negative means worse.
AUCthe probability that a random positive day is scored above a random negative one. 50 is a coin flip.
Brierthe mean squared error of the probability. Smaller is better.
Sharpnesson what percentage of the days the model says something clear (p ≥ 0.7 or ≤ 0.3) and how often it hits there. A calibrated model without sharpness is useless.
Value Areathe price band [rval, rvah] in which 70 % of the RTH volume traded. Always the previous day's here.
Trend Dayan RTH with a body ratio ≥ 0.6 — the body of the daily candle fills at least 60 % of its range.
Body Ratiobr = |rc − ro| / (rh − rl), computed on the RTH session.
relrRTH range divided by the median of the 20 previous RTH ranges. relr ≥ 1 means: today is bigger than the typical one of the last four weeks.
Walk-Forwardevery test year is computed from a model that has only seen data before it; the training window grows with every year. No fit sees its own test.

The numbers in this report are recomputed on every publish — from data/daily-bars-*.json and data/open-features-*.json, which the extractor builds from the tick store.

Run: 1607 daily bars, 0.6 s.