← Back to public data research
Phase 2 · Features & backtest
Quant features and backtesting workbench
Fix data, time horizon and execution assumptions first, then review characteristics and benchmark strategies. This is a reproducible research experiment, not live market data, model signals or personalized advice.
Research status
Awaiting execution
No results loaded; all inputs and results remain on this local page.
Provenance bar
Current research scope
Data are taken from the VCS-stored FRED `SP500` official CSV fixture. Calendar gap dates are explicitly counted and excluded, not filled by interpolation, zeros or the prior price. If the fixture fails to load, the results area will be cleared and an error shown.
Research controls
Set research parameters
Fixed parameters are for reproducibility only; do not use held‑out results to retune parameters.
Research output
Results reproducible
—
Feature view
Latest features
| Characteristics | 值 | Formula / calculation basis |
|---|
Baseline metrics
Baseline strategy and benchmark
Equity curve
test‑period cumulative curve
Runs the study and displays a text summary.
Walk-forward report
Preserve test window
| Split | Train／Test | Strategy | Benchmark | Excess | MDD | Entry |
|---|
Limitations & compliance
Review limitations before reading results
This is a long‑only 0/1 moving‑average deterministic benchmark. Signals are computed at close t and executed on the next bar; costs are calculated as turnover × bps/10,000. No slippage, taxes, market impact, leverage, shorting, halts, delistings or corporate‑action modelled; adjusted vs unadjusted input prices will directly affect results.
The 'future volatility' label is used only for historical regime research and must not be interpreted as a buy/sell signal, price target, return or risk guarantee. Average excess return describes this fixture with fixed parameters; it is not a prediction of future performance.