Market Regime Coverage Deep Dive

The Market Regime Coverage modal opens from the Regimes card on the results dashboard. It answers two questions that the raw equity curve cannot: what market conditions existed during the test, and did the strategy participate equally across all of them?

A strategy that trades profitably in trending, low-volatility conditions but avoids ranging or high-volatility environments has a hidden dependency. Its returns only materialise when the market cooperates. Understanding that dependency is the difference between deploying a strategy with realistic expectations and being surprised when it stops working.

The modal is divided into four panels in a 2x2 grid. The top row shows how market time and strategy exposure are distributed across volatility (left) and trend (right) regimes. The bottom row measures selectivity — whether the strategy over- or under-weights certain conditions — and translates the pattern into plain-language implications.

The top-left panel classifies every candle in the backtest into one of three volatility regimes — High, Medium, or Low — and shows how the strategy's trade exposure compares to the overall market distribution.

How Volatility Is Classified

The backend uses a 14-period Average True Range (ATR) to measure price movement magnitude. ATR captures the full range of each candle including gaps, then smooths the values using Wilder's method. To make the measure comparable across different price levels, ATR is normalised as a percentage of price.

Rather than using fixed thresholds (which would break across assets and timeframes), the system uses rolling percentiles. For each candle, it looks back at a window of up to 100 prior ATR values and computes the 33rd and 67th percentiles:

• ATR at or below the 33rd percentile → Low volatility
• ATR at or above the 67th percentile → High volatility
• Everything between → Medium volatility

This relative classification means "high volatility" for a stable asset like a bond ETF represents a completely different absolute ATR than "high volatility" for a cryptocurrency. The system adapts to whatever instrument it analyses.

Market vs Exposure Bars

Each regime row shows two overlapping bars. The dimmer bar represents market availability — the percentage of the test period that spent in that volatility level. The brighter bar represents strategy exposure — the percentage of the strategy's total in-market time that fell within that regime.

The numbers to the right (e.g., "37% → 14%") show market percentage first, then exposure percentage. When these numbers diverge significantly, the strategy is being selective. In the screenshot, the strategy was exposed to High volatility only 14% of the time despite the market being in High volatility 37% of the time — it actively avoided volatile conditions.

Transition Count

At the bottom, "211 transitions" counts how many times the volatility regime changed over the test period. A high transition count means the market moved between regimes frequently — conditions were unstable. A low count means the market sat in one regime for extended periods. This number sets context for interpreting the distribution: a backtest with 10 transitions may have only experienced 2-3 distinct volatility environments, while one with 200+ transitions saw constant regime changes.

The top-right panel uses the same dual-bar layout but classifies candles by trend strength rather than volatility. The three regimes are Trending, Transition, and Ranging.

How Trend Is Classified

The backend computes a trend strength metric similar to the Average Directional Index (ADX). It measures the magnitude of directional movement (up vs down) relative to total range, smoothing over a 14-period window. The resulting value ranges from 0 (no directional tendency) to 100 (perfectly directional).

Classification uses fixed thresholds on this strength value:

• Strength below 20 → Ranging — the market is moving sideways with no clear direction
• Strength between 20 and 25 → Transition — the market is in an ambiguous zone, possibly starting or ending a trend
• Strength at or above 25 → Trending — the market has a clear directional bias

Unlike volatility (which uses relative percentiles), trend classification uses absolute thresholds. This is standard practice because ADX-family indicators have well-established interpretation ranges that apply across instruments.

Reading the Bars

In the screenshot, 55% of the market time was classified as Trending, but only 27% of the strategy's exposure fell in trending periods. Meanwhile, Ranging conditions represented 25% of the market but captured 66% of the strategy's exposure. This strategy heavily over-weights ranging markets and under-weights trends — a pattern typical of mean-reversion or counter-trend approaches.

The 220 trend transitions indicate frequent regime changes, which is common in crypto markets where trends can establish and break down rapidly on shorter timeframes.

The bottom-left panel quantifies what the bars already suggest visually: selectivity is the difference between exposure percentage and market percentage for each regime.

The Calculation

For each regime, selectivity = strategy exposure% - market availability%. A positive number means the strategy over-weights that regime (it trades there more than the market sits there). A negative number means it under-weights the regime (it avoids those conditions).

Colour Coding

Values are colour-coded to make the pattern immediately scannable:

• Blue (positive, above +5%) — the strategy over-weights this regime
• Orange (negative, below -5%) — the strategy under-weights this regime
• Grey (between -5% and +5%) — neutral, roughly proportional participation

Volatility Selectivity

The screenshot shows High at -23%, Medium at +4%, and Low at +41%. The strategy massively over-weights Low volatility (+41%) and heavily avoids High volatility (-23%). Medium is roughly neutral. This is a strategy that waits for calm conditions before committing capital.

Trend Selectivity

Trending at -28%, Transition at +9%, and Ranging at +41%. The strategy strongly prefers Ranging markets and avoids Trending ones. Combined with the volatility selectivity, the picture is clear: this strategy trades in quiet, sideways markets and exits (or never enters) during volatile trends.

Dominant Regime Summary

Below the selectivity numbers, two labels show the dominant regime across each dimension. "Dominant volatility: Low" and "Dominant trend: Trending" describe the market itself — what regime was most prevalent. This is distinct from what the strategy selected. The market was mostly trending and low-volatility, but the strategy chose to concentrate its exposure in ranging and low-volatility windows.

Selectivity Thresholds

The system classifies overall selectivity based on the maximum absolute selectivity value:

• Max selectivity above 15% → Selective — the strategy has clear regime preferences
• Max selectivity between 5-15% → Moderately selective — some preference but not extreme
• Max selectivity below 5% → Indiscriminate — the strategy trades proportionally across all regimes

The bottom-right panel translates the numbers into two narrative blocks: Regime Coverage and Implications.

Regime Coverage

This section describes the market environment during the test period. The text varies based on dominant volatility:

• If the dominant regime was High volatility: "The test period was dominated by high volatility conditions. Observed behaviour reflects the volatility profile of the tested period."
• If Low volatility dominated: "The test period was characterized by low volatility conditions. Observed behaviour reflects the volatility profile of the tested period."
• For mixed or medium dominance: "The strategy operated across a range of volatility and trend conditions, providing exposure to multiple market environments."

The phrasing "observed behaviour reflects the volatility profile" is a deliberate caveat. It signals that the strategy's results are shaped by the conditions it encountered — not necessarily by its own robustness. A strategy that only ran through calm markets hasn't proven it can handle a crisis.

Implications

The implications text uses the selectivity classification:

• If the strategy is selective (max selectivity above 15%): "Strategy exposure shows selectivity relative to market availability. Performance outcomes may be sensitive to regime prevalence." This is the key warning — if the regime mix changes in live trading, the strategy's behaviour may change too.
• If the strategy is not selective: "Strategy participation appears [description] across regimes. Periods of regime transition may contribute to variance in results." The description fills in "selective", "moderately selective", or "indiscriminate" based on the thresholds.

Regime analysis addresses the most common failure mode in backtesting: strategies that look good because they were tested in favourable conditions, not because they are genuinely robust.

The Survivorship Problem

A trend-following strategy backtested exclusively during a bull market will show excellent results. But those results describe the market's behaviour, not the strategy's edge. The Regime Coverage modal makes this visible — if 80% of market time was trending and 95% of the strategy's exposure was in trending conditions, the results are a product of a single environment.

Selectivity Is Not Automatically Bad

A strategy that avoids high-volatility regimes by design is making a deliberate choice. If the entry conditions naturally filter out volatile periods — perhaps requiring a low ATR before entering — that selectivity reflects the strategy's logic working correctly. The problem arises when the selectivity is accidental or unrecognised.

The Forward-Looking Question

The regime modal helps you ask: "If the market's regime mix changes, will this strategy still work?" A strategy with +41% selectivity toward Low volatility is betting that low-volatility conditions will continue to be available. If the market shifts into a sustained high-volatility phase (as crypto markets periodically do), the strategy may simply stop triggering — or worse, trigger in unfamiliar conditions where its assumptions break down.

Cross-Referencing with Exposure

The Exposure Timeline deep dive shows when the strategy traded. The Regime Coverage modal shows what conditions it traded in. Together, they answer whether the strategy's active periods coincided with a specific regime. If the exposure timeline shows a concentrated block of trades in one part of the backtest, and the regime modal shows that block fell entirely within trending conditions, you have a clear regime dependency.

Cross-Referencing with Performance

If the Performance Metrics card shows a high Sharpe ratio but the regime analysis reveals the strategy only operated in one regime, the Sharpe is conditional — it applies only when that regime is present. The true unconditional Sharpe (across all market environments) may be much lower.