Grid Search
Optimization workflow · Equity ETFs · Workflow
The rules, in plain English
Grid search is the brute-force way to tune a rule: try every combination on a grid — here fast moving averages of 20, 50, and 80 days against slow averages of 100, 150, and 200, nine combinations in all — and score each with a real backtest's Sharpe ratio. It is a selection diagnostic. Picking the best past score is not the same as finding a rule that will work next.
The published backtest
| Total return | — |
|---|---|
| CAGR (annual growth rate) | — |
| Sharpe ratio | — |
| Maximum drawdown (worst peak-to-trough fall) | — |
| Win rate | — |
| Profit factor | — |
| Risk grade (our shorthand, from the published figures) | Not graded |
| Asset class | Equity ETFs |
| Style | Optimization workflow |
| Timeframe | Workflow |
| Type | Walk-forward |
— means no figure was published for this strategy. Ungraded and workflow entries publish diagnostics rather than a graded portfolio backtest, so some metrics are not published for them.
Honest verdict
Workflow example — no portfolio backtest published
This is a workflow example, not a portfolio strategy, so there is no return to judge. Its value is in how it tests other strategies — treat it as a method for avoiding self-deception, not as a way to make money by itself.
Walk-forward note
Exhaustive 3x3 grid over SMA fast in {20,50,80} and slow in {100,150,200} — 9 combinations, each scored by a real backtest's Sharpe.
Get all 50 in the workbook
Every strategy, its rules, metrics, and verdict — losers included — in one spreadsheet you can sort, filter, and study offline.
Attribution: strategy and published backtest results from the QuantJourney open-source library (Apache-2.0), via quants.md — published backtest results by QuantJourney, reproduced with attribution. Results are hypothetical backtest results based on historical data; they are not live trading results and do not guarantee future performance. Educational content only — not financial advice, and not a recommendation to buy, sell, or hold any security.