What a regime filter is, and why results change by regime

A market regime is a description of the conditions a market is in over a period: rising strongly, rising gently, falling, or crashing. A regime filter is a rule that changes what a strategy does depending on the regime, such as trading only when the market is rising, or stepping aside when it is falling.

Regimes matter for reading backtests too, because the same strategy can look excellent in one regime and poor in another.

How regimes are defined

There is no single standard. Common approaches use the price relative to a long moving average, the return over a past window, or measures of volatility. We classify each backtest window by the return of buying and holding over it: below minus 20% a year is a crash, minus 20% to 0% a decline, 0% to 20% modest, and above 20% a bull regime. Our methodology page sets this out.

What a regime filter does

A typical filter keeps a strategy in cash during regimes judged to be falling, and lets it trade otherwise. The aim is to avoid the deepest falls. The cost is that regimes are only clear in hindsight: a filter that waits for confirmation acts late, and a filter that acts early is often wrong.

Why beating holding is easier in a falling market

When a market falls, almost anything that spends time in cash beats buying and holding, because holding takes the whole fall. When a market rises strongly, almost anything that spends time in cash falls behind, because holding takes the whole rise. So a high share of wins in a crash means much less than the same share in a bull market.

Our own platform-wide figures show this clearly. In crash windows, 13,266 of 14,775 backtests beat holding (89.8%). In decline windows, 13,505 of 20,825 did (64.8%). In modest windows, 22,474 of 71,959 did (31.2%). In bull windows, 1,498 of 11,068 did (13.5%).

The regime figures are the platform-wide statistics printed on each StrategyLens published card, updated 29 October 2026: hypothetical backtests across every tested ticker, with no trading costs.

The pattern is what you would expect from strategies that are out of the market part of the time: they shine when holding does badly and lag when holding does well.

What this means for reading results

Ask which regimes a result came from. A strategy that beat holding mostly in crash windows has shown that it avoided falls, not that it has an edge in normal markets. A strategy that beat holding in bull windows has done something much harder.

Ask how often each regime occurs. Over long periods, modest and bull windows are far more common than crashes in broad stock markets, so a strategy that wins only in crashes spends most of its time behind.

Testing a regime filter fairly

Define the regime rule using only information available at the time. Test the strategy with and without the filter over the same dates. Compare return per year, maximum drawdown and time in the market, and compare both with holding. Check how many times the filter switched, because each switch is a trade with costs.

Where to see regimes on our pages

Each ticker page names its dominant regime, and the regime figures above appear on every published card. The methodology page explains how regimes are classified.

Why regimes are hard to call in real time

A regime is easy to name afterwards: looking back at a chart, the crash and the recovery are obvious. At the time, the first days of a crash look like an ordinary dip, and the first days of a recovery look like a bounce that might fail. Any filter has to decide using only what was known then, and it will be late, early or wrong some of the time.

That is why a regime filter's value shows up mostly in the deepest, longest falls, where acting late still avoids a large part of the damage. In short, sharp falls that reverse quickly, a filter tends to sell near the bottom and buy back higher.

Regimes and individual shares

A single share can be in its own regime that differs from the market's. A company can be falling on its own bad news while the index rises. Filters based on the index can keep a strategy in a share that is falling, and filters based on the share alone can miss a market-wide fall that has not reached it yet. Our regime classification is per backtest window for each ticker, based on that ticker's own buy-and-hold return.

In short, a regime filter is a trade-off rather than an improvement. It usually swaps some return in rising markets for smaller losses in falling ones. Whether that swap is worth making depends on how often each kind of market occurs over your time horizon, and on how much a deep fall would affect you. Backtests can show the size of the swap in the past; they cannot say how future markets will divide between regimes.

This is research, not financial advice. Past results, real or simulated, do not tell you what will happen next.

What is a market regime?

A description of market conditions over a period, such as rising strongly, rising gently, falling or crashing.

What does a regime filter do?

It changes a strategy's behaviour by regime, for example holding cash when the market is judged to be falling.

Why do strategies beat holding more often in crashes?

Because holding takes the whole fall, so anything that spends time in cash looks better. In rising markets the reverse is true.