Research note
TA Strategy Optimizer: the best cell is unvalidated
Parameter grids preserve all cells and exact drill-in settings. No frozen optimum, performance statistic or pass/fail is inventoried.
Backtest verdict: Unknown — no frozen result recorded in the inventory. A top-ranked in-sample cell is an optimization result, not a passing strategy test.
Open this research tool in Backtesting
The rule
The optimizer sweeps one or two real strategy parameters while holding ticker, other parameters, history and execution settings explicit. The grid supports up to 256 cells and seven ranking metrics. Those are configuration limits, not independent trades or measured evidence.
Entry, exit, filters, warmup and signal timing remain the selected strategy's definition. Transaction cost retains round-trip units; slippage, stops/targets, short support, sizing and benchmark options remain exact per-cell settings. A cell link opens an individual full run with that cell's actual configuration; it is a new request, not retrieval of a historical ledger.
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.
Read the complete sweep request and all cells, including invalid, failed or unavailable cells and unscored holes. Ranking by a chosen metric selects on the same sample; it does not establish out-of-sample efficacy. Correlated parameter neighbors are not independent confirmations, and repeated searches increase selection bias.
The inventory has no frozen sweep payload, selected strategy/ticker, exact dates, price basis, benchmark result, cost record or validation artifact. No optimization was run to fill these gaps. The tool's cell count cannot stand in for a trade count.
Results
| Study / variant | Inventory evidence | Window / basis | Strategy return / CAGR | Benchmark | Max drawdown | Closed trades / observation N | Significance / IC | Native verdict |
|---|---|---|---|---|---|---|---|---|
| Full sweep / best cell / invalid cells | D11 / F14 / H15 | 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. The best grid score is not PASS. Missing or invalid cells remain unavailable, never zero-return successes. Any later claim must identify the full historical search and a genuinely separate validation result without hiding adverse cells.
Sources and live research
INVENTORY D11, F14/H15. Sources to check: frontend/components/BacktestOptimizer.tsx, frontend/lib/taBacktest.ts, the optimizer in backend/app/scoring/technical/service.py and backend/app/api/ta_schemas.py. Open: full historical sweep request/cells/selection plus separate validation. Individual full-results view remains distinct from the sweep.