Regime Filter

Every trading strategy has an implicit or explicit assumption about market behavior. A breakout strategy assumes directional continuation — that price breaking above resistance will keep moving higher. A mean-reversion strategy assumes price will return to a central value after deviating. A gap-and-go strategy assumes the opening momentum will sustain through the morning. These assumptions are valid in some market environments and systematically wrong in others.

A regime filter is the meta-layer that asks: "Is the current market environment the type in which my strategy's core assumption holds?" Without this filter, strategies generate signals in unfavorable conditions, producing losing trades that would never have been taken by a disciplined human trader who intuitively recognized the wrong environment.

The Core Problem: Strategy-Regime Mismatch

Consider a momentum breakout strategy that works well in trending markets. Backtest results over a multi-year period show a 62% win rate and 2.1R average winner. But examine the regime-specific results:

• In trending regimes (defined by ADX > 25 or high efficiency ratio): 74% win rate, 2.6R average winner
• In choppy regimes (ADX < 15 or low efficiency ratio): 41% win rate, 1.2R average winner

The strategy isn't a '62% win rate strategy' — it's a 74% win rate strategy in trending regimes masquerading as a mediocre strategy because it's also being traded in conditions where it doesn't work. A simple regime filter that disables the strategy in choppy conditions would dramatically improve the real-world performance.

Common Regime Filter Types

Trend/momentum filters validate that a trend exists before taking trend-following signals:

• Moving average direction: only take long breakouts when the 50-day SMA is sloping upward (positive slope over 20 days)
• ADX threshold: require ADX > 25 for trend-following entries; avoid when ADX < 20 (no trend)
• Efficiency Ratio (ER): require ER > 0.35 for directional trades (price moving efficiently, not zigzagging)

Macro regime filters evaluate broad market conditions:

• Market breadth: only take long individual stock signals when >60% of S&P 500 stocks are above their 50-day MA
• VIX filter: avoid entries when VIX > 30 (extreme fear produces incoherent price action that breaks most strategy assumptions)
• SPY trend: only take long setups in individual stocks when SPY is above its 20-day SMA

Intraday regime filters apply within a single trading session:

• Opening range direction: only take longs when a stock is holding above its 15-minute opening range
• Market gap filter: if SPY opens down more than 0.75%, require a reclaim of the prior close before taking long momentum entries
• Session efficiency ratio: calculate ER on 5-minute bars using the past 14 periods; if below 0.30, apply stricter entry thresholds or skip momentum setups entirely

The Efficiency Ratio as a Regime Filter

The Kaufman Efficiency Ratio (also called the Fractal Efficiency Ratio) is one of the cleanest intraday regime indicators for day traders:

ER = |Net directional move over N bars| ÷ (Sum of all individual bar moves over N bars)

An ER of 1.0 means price moved in a perfectly straight line — maximum efficiency, pure trend. An ER of 0.0 means price zigzagged and ended exactly where it started — maximum inefficiency, pure chop. In practice:

• ER > 0.40: Strong trend, momentum and breakout strategies perform well
• ER 0.20–0.40: Mixed conditions, apply additional confirmation requirements
• ER < 0.20: Choppy/noisy, avoid momentum entries, consider mean-reversion setups instead

This filter is particularly powerful because it is non-parametric — it doesn't require choosing a moving average length or threshold and measures the actual quality of price movement rather than a derived oscillator.

Regime Filters in Algorithmic Systems

In systematic trading code, a regime filter is implemented as a gate that must pass before any signal evaluation occurs:

if regime_filter.is_favorable():
    candidates = screener.scan()
    for candidate in candidates:
        evaluate_entry(candidate)
else:
    skip_scan()  # no trades today, wrong conditions

The filter can also be continuous rather than binary — weighting the required conviction score higher in marginal regime conditions. This approach applies the filter with graduated intensity rather than a hard on/off switch.

Strategy-Specific Regime Requirements

Different strategy types require different regime filters:

Strategy	Favorable	Unfavorable
Momentum/breakout	Trending, high ER, ADX > 25	Choppy, low ER, low ADX
VWAP mean-reversion	Range-bound, moderate VIX	Strong trending, news-driven
Gap and go	Clear directional gap with volume	Narrow gap, ambiguous direction
Opening range breakout	Clear trend day signal	Whipsaw/reversal session
Pairs trading	Mean-reverting spread	Diverging, correlated market move
Short selling	Downtrending market, weak breadth	Broad market uptrend, strong breadth