Regime detection: simple volatility filters that help

Most systematic strategies bleed money during high-volatility breaks or stagnant ranges because they lack a basic regime filter. You don't need a fancy hidden Markov model to keep your account safe. Sometimes, looking at the raw environment is enough to tell you when to step back.

Start with an ATR percentile filter. If the current Average True Range is in the top decile of its rolling 250-day history, volatility is extreme. Scaling down position sizes or pausing new entries during these spikes prevents the common mistake of getting chopped up during a liquidity vacuum. Similarly, tracking the slope of a 200-day moving average acts as a blunt but effective trend filter. If the slope is negative, you might restrict your model to short-only signals or stay in cash.

Here is how a basic volatility regime filter might look on a daily chart:

These filters are easy to implement, but they come with a clear trade-off. By sitting out during messy sessions or high-volatility spikes, you will inevitably miss the occasional V-shaped recovery. Your total trade count drops, which can feel frustrating when you're used to high turnover. However, the goal isn't to catch every move; it's to avoid the moves that wipe out your edge.

If you are looking for ways to backtest these conditions, sites like QuantConnect offer the necessary data to build these filters into your existing pipeline. Have you found that filtering by volatility actually improves your Sharpe ratio, or does the reduced trade frequency just make your equity curve feel more erratic?