STOCK RESEARCH STUDIO
Build your own indicator combinations
One workspace for multiple indicators. Choose conditions, combine signals, and measure what historically happened afterward.
Historical observations, not guaranteed future probabilities. No live data or investment recommendations. CSV calculations stay in your browser.
Historical results & strategy comparison
Input data quality report
Cost-aware historical trade simulation
Signal after a bar closes → buy at next bar open → exit at close after the chosen horizon. Only one trade can be open at a time. The earliest segment is training data; the later segment is reserved for separate validation.
Training segment
Held-out validation segment
Historical trade risk diagnostics
Trade-based measures are calculated from non-overlapping simulated fills, not overlapping signal counts. The exposure fraction includes bars held in a position. A past loss streak or worst trade cannot bound future losses.
| Risk measure | Training | Validation |
|---|
Trade-exit drawdown can miss losses inside a trade. The additional low-price stress estimate compares observed bar lows against the earlier closing-mark peak; it is not an executable stop loss or a full intraday path. Profit factor is undefined when no losses occurred, and the historical 10th percentile is hidden below 20 trades. No Sharpe ratio or future win probability is inferred.
Validation equity curve vs buy-and-hold (starting at 1.00)
Solid green: selected strategy. Dashed blue: buy-and-hold benchmark. Both include the entered entry/exit costs. Benchmarks buy at the first eligible entry and sell at the end of each segment.
This curve compounds modeled trade returns only; it ignores idle cash interest, market spread changes, partial fills, dividends and taxes. If validation has no trades, a curve cannot be drawn.
Stability across three chronological validation subperiods
The SAME frozen rule is applied separately in the first, middle, and last thirds of the validation history. No automatic tuning or strategy optimization is performed. If you edit your rules after viewing these slices, they are no longer independent unseen tests.
| Validation slice | Trades | Win rate | Strategy net return | Buy-and-hold |
|---|
Compare up to 5 combinations on the same dataset
| Combination | Samples | Win frequency | Baseline | Mean return |
|---|
The historical signal frequency above measures close-to-close changes without costs. The separate trade simulation below includes your commission and slippage assumptions; it is not a live execution model. Overlapping signal outcomes are not statistically independent; the Wilson interval ignores serial dependence. Changing the rules after seeing held-out results invalidates the claim of independent validation. No results guarantee future performance.