ProEA Lab · Research

Trading studies with methods, limits and downloadable evidence.

Start with the public monthly-bootstrap package below: its summary inputs, calculation and expected output run without a ProEA account. Our other studies describe their samples and controls, but their source checkout and frozen market data are not provided as public downloads here.

ProEA Lab publishes these studies and sells related trading software. Internal review is not independent third-party certification. Read our editorial and correction standards, cite the version and limitations, and report calculation errors to support@pulltrade.app.

New series

Edge Log: research published one wave at a time.

The rules, costs and results of our edge hunt — failures included.

Open the Edge Log →

Public download · Version 1.0 · 8 September 2026

Reproduce the monthly-bootstrap arithmetic

Download 24 published monthly summaries, standalone JavaScript, expected output and file checksums. Run node verify.mjs after extracting the ZIP with Node.js 22 or newer. No account, paid software or package installation is needed.

The inputs are separate generated-tick MT5 simulations, each starting from $10,000: 18 positive months and 6 negative months, ranging from −46.41% to +422.38%. The package reproduces the resampling arithmetic and charts; it cannot re-run the underlying MT5 tests. The sample is small, includes an extreme positive outlier and excludes unseen market conditions. Month-end drawdown omits floating losses. These outputs are not live returns or future probabilities.

Download research package (ZIP)Method and reuse terms Summary inputs (CSV)Calculation source Expected results SHA-256 checksums Read the study and sensitivity check

Suggested citation: ProEA Lab (2026), Monthly bootstrap arithmetic, version 1.0, proea.app/en/research#monthly-bootstrap. ProEA publishes this study and sells MTR; this package is not an external certification.

Study 1 · Volume profile vs its folklore

Does yesterday's POC attract price more than a random level the same distance away? Does the “80% rule” hold?

Data:: GC=F (COMEX gold futures) 5m bars with real exchange volume, 50 full CME sessions (2026-04-29 → 2026-07-10), frozen in-repo
Sample:: 50 session pairs; per-pair distance-band random-level controls (K=1,000)
Control:: Distance-band random levels [0.8d, 1.2d], real-level ±$2 excluded — redesigned after adversarial review broke the first null
Limitation:: One symbol, one 50-session window; bar-resolution volume approximation

Finding: POC touch rate 69.4% vs control 71.5% — the magnet is folklore; the “80% rule” failed all three principled readings; one narrow POC-respect effect graded “suggestive, unproven”

Read the full report →

Source-checkout command and access limits

node --experimental-strip-types scripts/research/volume-profile-study.ts

This command requires our website source checkout and the referenced frozen data, which are not public downloads on this page. Third-party market-data redistribution rights have not been established. The published report describes the method; it does not provide unrestricted independent reproduction from this page alone.

Study 2 · Our own SMC zone score, shuffle-tested

Do zones our toolkit scores higher actually resolve “respected” more often — or does a shuffled score column do as well?

Data:: Same frozen gold tape, re-exported at bar level (13,640 bars), in-repo
Sample:: 1,729 zones detected by a line-annotated port of the shipped Pine at defaults; 1,693 resolved
Control:: Scores shuffled across resolved zones ×2,000 (fixed seed) + 12-bar thinning for zone clustering + shipped age-prune robustness
Limitation:: One market and timeframe; a port, not the TradingView runtime; respect definition is mechanical, not a win rate

Finding: Composite score: no better than shuffled (+2.6 pts, p=0.21; thinned −1.9, p=0.72). Two ingredients showed exploratory associations: displacement +18.0 pts (p < 0.0005), volume +10.5 (p=0.003)

Read the full report →

Source-checkout command and access limits

node --experimental-strip-types scripts/research/smc-zone-study.ts

Study 3 · Multi-timeframe agreement vs the next hour

When five timeframes agree, does the next hour behave differently? Is “full stack” alignment rare enough to mean anything?

Data:: Frozen gold tape (observation clock + 15m stack) + frozen 60m/730d and 1d/5y series for honest higher-TF warmup, all in-repo
Sample:: 1,131 hourly stack readings (non-overlapping forward windows); 1,124 directional
Control:: Confluence column shuffled ×2,000 + session-cluster bootstrap for serial dependence
Limitation:: One market and window; UTC-bucketed 4h/weekly resamples; hourly readings remain serially dependent after guards

Finding: Full alignment held 28.6% of all hours (median run 3h) — common, not special; alignment level separated nothing (+0.4 pts, p=0.43; CI [−5.2, +10.0]); the 100% alert hit 44.8% vs 48.2% baseline

Read the full report →

Source-checkout command and access limits

node --experimental-strip-types scripts/research/mtf-confluence-study.ts

Method notes: nulls are permutation-based with fixed seeds; serial dependence is guarded with thinning or session-cluster bootstraps; sensitivity readings that flip under guards are reported and then dismissed as artifacts, in the open. Nothing on this page claims a win rate, an edge, or a prediction — these are audits of claims, including our own.