StrategyLens methodology

Every verdict is built from at least 100 regime-tested backtests. Here's how the numbers are produced and what they mean.

Regime classification

Every backtest window is classified by its benchmark return (buy-and-hold CAGR). Crash: below −20%. Decline: −20% to 0%. Modest: 0% to 20%. Bull: above 20%. A single ticker appears in multiple regimes across different time windows — tested in the conditions it would actually face.

Win rate and beating buy-and-hold

Two different figures. Win rate is the share of closed trades that made money, from each strategy's default settings over the full price history; a position that is still open and down is not in it. Beat buy-and-hold is the share of backtests whose CAGR exceeded buying and holding that ticker over the same window. A strategy can win most of its trades and still trail buy-and-hold, because it sits in cash while the market rises. Beating buy-and-hold in a crash regime is easy, since cash alone does it, so a high share there means less than the same share in a bull regime.

The Lens pentagon

Five axes scored 0–1. Beat holding: share of backtests that beat buy-and-hold. Excess: average CAGR above buy-and-hold (−30% to +30% scale). Sharpe: risk-adjusted return, capped at 1.5. Resilience: inverse drawdown severity. Evidence: number of runs, capped at 20. Larger shape = stronger evidence — but not a recommendation.

Risk levels

Stock and non-commodity ETF tiers use average drawdown and the share of backtests that beat buy-and-hold; commodity cards use the worst measured drawdown only. Conservative: drawdown no worse than −45% and beat buy-and-hold in at least 70%. Moderate: drawdown no worse than −65% and at least 40%. Aggressive: drawdown no worse than −80% and at least 15%. Informed gamble: leveraged ETPs, drawdowns worse than −80%, or under 15% — speculative.

What this is not

A research tool, not a recommendation engine. Past performance does not predict future returns. Backtests use historical data and cannot account for future regime changes, liquidity events, or structural shifts. Not FCA regulated. Not financial advice.

How to read these results

These are simulations. We run a strategy's rules over historical prices and record what would have happened. Nobody traded these results. Real trading has costs, delays and mistakes a simulation does not fully capture.

The past does not repeat. A strategy that beat buy-and-hold over one period can lose over the next.

Most strategies we test do not beat holding. We publish that. When many strategies are tested on many tickers, a good result on one ticker is more likely luck than skill.

What a figure depends on. The dates tested, the data source, the trading costs assumed (headline figures assume no trading costs; a fee you pay, foreign-currency charges and tax lower them), the rule settings, and how dividends and splits are handled. Change any of these and the figure changes.

Holding a fund in an ISA is a separate question. Many of the funds we test are listed in the US, and UK platforms differ on whether they carry them: see Can UK investors hold US-listed ETFs in an ISA?

What we do not know about you. Your goals, time horizon, tax position, other investments, or how much loss you can take. So nothing here can tell you whether an investment suits you.

You can lose money. Prices go down as well as up. You may get back less than you put in, or nothing.

Related research and how our test differs

AQR's "Hold the Dip" (2025) studies buying the dip on the S&P 500.

Bonini, Shohfi and Simaan (2024) is related research on the same question.

Our test differs in one way: each ticker is tested on its own price history with the same rules, so the table compares tickers rather than one index.