Research note
Should I Trade replay: unknown historical verdict
Decision buckets, equity, IC and sizing experiments keep their own methods. No frozen replay result or pass/fail is inventoried.
Backtest verdict: Unknown — no frozen result recorded in the inventory. YES/CAUTION/NO are historical decision categories, not a passing-backtest judgment for the whole page.
Open this backtest in Backtesting
The rule
The Decisions engine replays category scorers as of each sampled historical date, groups forward SPY returns by YES/CAUTION/NO and compares position-sized equity with buy-and-hold. Controls select swing/day mode, sector-ETF/S&P 500 breadth and lookback/cadence. It opens at five years with daily replay; two years uses daily and ten years uses a two-day step.
The separate public Sizing Lab retains five variants, thresholds, reweight experiments, train/test and null rows. It uses its own horizon and minimum two-day sampling. It is not an operator Apply surface. Exact fills, costs, sizing formulas and proxy details belong to the returned method and backend implementation; the inventory records no single frozen configuration or universal stop/target to quote.
Methodology
Evidence as of: the INVENTORY review dated November 5, 2026; this is the consolidation planning baseline, not a new observation window. Publication does not extend any study's sample.
Replay inputs differ from the live macro decision surface; historical proxies and degraded availability can change a category. Bucket N, overlapping forward returns, effective sample size, category/composite IC and equity are different statistics. The native dynamic verdict is result-specific prose; this article introduces no new significance gate.
Decisions remains protected while Sizing Lab is public. Training results, held-out rows, null checks and reweight searches remain distinct. The inventory has no dated mode/breadth/window/cost/result payload, benchmark return or frozen pass/fail; no replay or experiment was run for this post.
Results
| Study / variant | Inventory evidence | Window / basis | Strategy return / CAGR | Benchmark | Max drawdown | Closed trades / observation N | Significance / IC | Native verdict |
|---|---|---|---|---|---|---|---|---|
| Decision replay / buckets / IC / equity | D14 / F27 / H33 | not recorded | not recorded | not recorded | not recorded | not recorded | not recorded | Unknown — no frozen result |
| Five sizing variants / thresholds / train-test / null | D14 / F28 / H34 | not recorded | not recorded | not recorded | not recorded | not recorded | not recorded | Unknown — no frozen result |
The linked workspace retains the complete native output and notes when available. Missing evidence is not a zero return, a successful empty result or a failed strategy.
Verdict and limitations
Unknown — no result/verdict recorded in this inventory. A YES category is not a promise of profit, and a winning sizing variant is not independent validation. Preserve proxy, degraded-data and effective-N caveats and all adverse/null outcomes. No reweight Apply, signal or vote is added.
Sources and live backtest
INVENTORY D14, F27–F28/H33–H34. Sources to check: frontend/components/should-i-trade/ShouldITradeBacktestBoards.tsx, frontend/lib/strategy-pages.ts, backend/app/scoring/should_i_trade_backtest.py (_verdict and replay notes) and should_i_trade_experiments.py. Open: captured mode/breadth/window/cost/result and replay-versus-live macro differences. Inspect the separate Sizing Lab.