MM·V2 Backtest Workbench

build 0719-2315 · Trend Template 8/8 · RS vs SPY · micro-VCP · entry≠stop guards · 2-week forward sim
SPY baseline: not loaded
🚦 REGIME — load SPY/QQQ…

Add a ticker and an as-of date (the signal day, D0). Daily history is fetched automatically (Stooq, Yahoo fallback via CORS proxy) and all 26 instruction fields are derived as of that date — change the date and everything recomputes. Any field can be hand-edited (e.g. Prior Pivot High); edited cells turn amber and the score recalculates. The chart shows ~60 sessions of context, a D0 flag, and the shaded next 10 sessions with entry / stop / target lines so you can see whether the pick worked. If fetching fails (offline / proxy down), use CSV on the row to upload Date,Open,High,Low,Close,Volume.

entry allow %
Step 0A — Theme Lock · price momentum (auto scan)
runs at most once per trading date — after that the 🔒 lock is reused (theme-lock protocol). Paste a lock line from another chat to sync. Top-down read only; never derived from the tickers you score below.
Step 0B — Theme Lock · social buzz (Reddit / X / Threads)
Reddit/X block automated reading, so this lock comes from a Claude chat: click copy research prompt, paste it into a Claude chat with web search, then paste the returned 🔒 SOCIAL THEME LOCK line back here. Locked once per day like 0A.
US Fear & Greed
loading…
SRS — Sectors (RS vs SPY)
click rank — 11 SPDR sector ETFs, 13w & 4w relative strength
US Breakouts
daily breakout count across the universe above (vs SPY) — click scan
TV scan # history
fill the TV scan # field daily — counts plot here
S&P 500 — % above 200d MA
scans all ~500 constituents via the local proxy (≈2 min) · 🔒 locked per day
🎯 High-score finder
⚙ sampling criteria — each row is a COMBINED filter (score AND RS AND R:R AND Win%), returns that row's count, most-recent first
scans every (ticker × past date) across all available history, filtered by the sampling criteria above — click S&P 500 to load the universe, then scan · click a hit to add it as a blind backtest row
🧪 Strategy Lab — process huge samples in memory (no rows), find the winning combination
⚙ sampling criteria — each row is a COMBINED filter (score AND RS AND R:R), capped at its count
click run analysis to scan every (ticker × past date), simulate the bracket, and load the sample columnar — rows are NOT added to the table, so it stays light at 100k+ samples. Tickers are cached, so re-running or scanning more is fast.
same engine, separate study lists — ☆ on a row sends it to the Shortlist section below · high-score finder fills List 4
Score · RS ≥ · R:R · Win% ≥
TickerAs-of (D0)ScoreWin%TierTTRS ppRS 13w %Sector / ThemePattern Setup PD% EntryStopTargetR:RReward %Outcome (10 sess)P&LAlert
No tickers yet — add one above to start the backtest.
★ Shortlist (0) — potential trades from both lists
no tickers shortlisted yet — click ☆ on any row
⏱ 1-minute VCP session — intraday entry timing (same page)
⋯ drag ⋯
Σ DAILY LIST — Conclusion (swing)
press recompute — sums every revealed result in your list and states the proven best daily combination.