Drawdown Analysis Deep Dive

The Drawdown Analysis modal opens from the Drawdown card on the results dashboard. It answers the question that equity curves are designed to obscure: how bad did it get, and how long did it take to recover?

An equity curve that ends 40% above its starting point looks encouraging. But if it spent 1,200 days below its previous peak at one point — over three years waiting to break even — most traders would have abandoned the strategy long before the recovery completed. The drawdown modal makes this experience visible.

The modal is divided into four panels in a 2x2 grid. The top row quantifies depth and duration. The bottom row visualises the underwater curve and classifies the overall drawdown pattern into plain-language descriptions.

The top-left panel shows the strategy's worst-case loss from peak equity, along with current status and recovery statistics.

Maximum Drawdown

The hero metric. Displayed in large rose-coloured text, this is the single worst peak-to-trough decline during the backtest, expressed as a percentage. If the max drawdown is 26.4%, it means at some point the strategy was down 26.4% from its highest equity value.

The calculation walks through the equity curve, tracking the running maximum. At each point, it computes the percentage decline from that running peak. The worst such decline across the entire curve becomes the maximum drawdown.

Current Status

"Currently 6.6% below peak" means the strategy ended the backtest in a drawdown — it never fully recovered from its latest decline. This is important context. A strategy that ends in drawdown has an incomplete recovery that hasn't been tested yet. The next question is always: will it recover, or has the edge deteriorated?

When the strategy ends at a new high, the display reads "Currently at equity high" instead — a more reassuring state, though not a guarantee of future performance.

Drawdown Count and Recovery Rate

Drawdowns observed: 9 counts every distinct drawdown period — each time the equity curve dropped below its previous peak and either recovered or remains unresolved. More drawdowns in a backtest can be reassuring (more recovery data) or concerning (frequent equity erosion), depending on their depth and duration.

Recovered: 8 / 9 means eight of nine drawdown periods ended with the equity reaching a new high. The ninth is ongoing — the strategy ended the test still underwater. A 100% recovery rate across many drawdowns strengthens confidence that the strategy has a genuine edge rather than a lucky run. An unresolved drawdown at the end of the test is common and not automatically concerning, but it does mean the last chapter hasn't finished.

The top-right panel separates two related but distinct time measurements: how long the pain lasted, and how long the recovery took.

Pain Length

Duration measures the total time from the start of a drawdown (when equity first drops below the previous peak) to when it makes a new high. This is the full pain window — the entire period where an investor would have been below their high-water mark.

• Longest: 1268d — the worst drawdown lasted nearly three and a half years from peak to new peak. This is the number that makes or breaks a strategy in practice. Most investors cannot tolerate being underwater for years, regardless of what the backtest eventually shows.
• Average: 258.25d — the typical drawdown lasted about 8.5 months. This is computed across completed drawdowns only (ongoing drawdowns are excluded because their final duration is unknown).

Recovery Tail

Recovery measures the time from the drawdown's trough (deepest point) to the next new high. This is a subset of the total duration — it excludes the descent and measures only the climb back.

• Median recovery: 18d — half of all drawdowns recovered within 18 days of hitting their trough. The median is used rather than mean because a single extended recovery would distort the average.
• Longest stagnation: 1323d — the longest gap between consecutive equity peaks, regardless of drawdown depth. Stagnation differs from recovery: it measures the total flat period where the account went nowhere, even if the drawdown itself was shallow. A 1,323-day stagnation means the strategy made no progress for over 3.5 years at its worst stretch.

Duration vs Recovery vs Stagnation

These three metrics each capture a different flavour of time risk:

• Duration = peak to new peak (the full drawdown experience)
• Recovery = trough to new peak (how long the climb back takes)
• Stagnation = peak to peak with no regard for intermediate drawdowns (the longest period of no progress)

A strategy with short recoveries but long stagnation is one that bounces back from dips quickly but doesn't generate meaningful new highs between them. The equity curve chops sideways. A strategy with long recoveries but short stagnation is one that takes big hits but eventually breaks through to new highs once it gets going.

The bottom-left panel shows the underwater curve — a continuous plot of how far below the running peak the equity sat at each point in time.

How It Works

At every point in the backtest, the system tracks the highest equity value seen so far (the running maximum). The underwater percentage at each point is:

underwater = ((current equity - running peak) / running peak) x 100

The result is always zero or negative. Zero means the strategy is at a new high. -26.4% means it is 26.4% below the best equity it ever achieved.

Reading the Chart

The y-axis runs from 0% at the top down to the worst drawdown. The x-axis spans the full backtest period with date labels. The area between the curve and the 0% line is filled with a rose-coloured gradient — deeper drawdowns produce wider, more visible pools of red.

The visual immediately communicates what the numbers alone cannot: the shape of drawdown behaviour. Key patterns to look for:

• V-shapes — sharp declines followed by quick recoveries. These indicate the strategy has a strong mean-reverting equity curve. Losses are real but temporary.
• U-shapes — declines that reach a trough and sit there for an extended period before recovering. The flat bottom is the dangerous part — the strategy isn't losing more, but it isn't recovering either.
• L-shapes — declines that never recover within the test period. These appear as a curve that dips and stays down. An L-shaped drawdown at the end of the backtest raises the question: is this a temporary drawdown or permanent capital impairment?
• Saw-tooth patterns — repeated shallow dips followed by new highs. This is the healthiest drawdown profile, indicating consistent edge with manageable risk.

What the Equity Chart Hides

A standard equity chart that grows from $10,000 to $14,000 looks like a success. The underwater curve for the same strategy might show it spent 70% of the test below its previous peak, with three separate drawdowns exceeding 15%. The equity chart emphasises the destination. The underwater curve emphasises the journey.

The bottom-right panel distils the raw drawdown data into four categorical assessments and a synthesis sentence.

Frequency

Two dimensions combined into one label:

• Depth: max drawdown above 30% = "Deep", 15-30% = "Moderate", below 15% = "Shallow"
• Count: more than 10 drawdown periods = "Frequent", 5-10 = "Periodic", fewer than 5 = "Infrequent"

"Periodic, moderate" in the screenshot means 5-10 drawdowns (periodic) with a max depth of 15-30% (moderate). This is a middle-ground profile — the strategy draws down meaningfully but not catastrophically, and it happens regularly but not constantly.

Recovery

Based on average drawdown duration and spacing behaviour:

• Average duration over 200 days → "Recovery tails dominate risk" — the primary risk is not the depth of loss but the time it takes to recover
• Average duration 60-200 days → "Extended recovery periods"
• Average duration under 60 days → "Quick resolution"

The suffix ", uneven pacing" is added when drawdowns are clustered together or the longest stagnation exceeds one year. This indicates that recovery times are inconsistent — some drawdowns resolve quickly while others drag on, making the experience psychologically difficult to plan around.

Spacing

When there are two or more drawdown periods, the system analyses the gaps between them using the coefficient of variation (standard deviation divided by mean):

• CV below 0.5 with minimum gap over 14 days → "Well-distributed" with the range shown (e.g., "43-164d")
• CV above 1.0 with minimum gap under 7 days → "Clustered" — drawdowns bunch together, often during adverse market conditions
• Otherwise → "Irregular" — no clear pattern to the timing

Resolution

• All drawdowns recovered → "Capital recovery achieved"
• More than 80% recovered → "Mostly recovered, patience required"
• Otherwise → "Some periods unresolved"

Synthesis

The bold sentence at the bottom combines depth and recovery into a single diagnostic. If all drawdowns recovered: "[Depth] but recoverable drawdowns with [duration description] recovery." If some remain unresolved: "[Depth] drawdowns with [duration description] recovery tails." This sentence is designed to be quotable — a single line that captures the strategy's drawdown character.

Below the four-panel grid, a collapsible section lists every drawdown period individually. Each row shows:

• Period — the start date of the drawdown
• Depth — peak-to-trough percentage decline (highlighted in rose when exceeding 10%)
• Duration — total days from peak to recovery (or current if ongoing)
• Recovery — days from trough to new high (or "-" if not yet recovered)
• Status — completed (reached a new peak) or ongoing (still underwater)

This table lets you inspect each drawdown in isolation. The pattern across rows often reveals more than any single metric. If the first five drawdowns all recovered in under 30 days but the sixth took 1,200 days, something changed — possibly the market regime shifted, or the strategy's edge weakened over time.

Sorting by depth shows whether the worst drawdowns cluster early or late in the backtest. Sorting by duration reveals whether recovery times are trending longer — a potential sign of edge decay.

Drawdown data becomes most diagnostic when combined with other dashboard cards.

Drawdown + Exposure

The Exposure Timeline shows when the strategy was in the market. Overlaying this with the underwater curve reveals whether drawdowns happen during active trading periods or bridge the gaps between positions. A strategy that goes underwater while flat is experiencing slippage from missed opportunities, not trading losses — a subtly different problem with different solutions.

Drawdown + Regimes

The Regime Coverage modal shows which market conditions the strategy experienced. If the deepest drawdown coincides with a shift from Low to High volatility (visible in the regime transition count), the strategy may have a regime vulnerability. This is actionable information — you could add a volatility filter to avoid entering during hostile conditions.

Drawdown + Performance Metrics

The Performance Metrics card shows the Calmar ratio (annualised return divided by maximum drawdown). A Calmar above 1.0 means the strategy returned more per year than its worst drawdown. Below 0.5 means negative risk-adjusted returns in the worst case. Cross-referencing the max drawdown from this modal with the Calmar ratio gives a complete picture of return-per-unit-of-pain.

Drawdown + Behaviour

The Behaviour card's streak analysis shows how sequences of wins and losses unfold. A max losing streak of 9 (as in this strategy) likely contributed to the deep and prolonged drawdown periods. Knowing that losses come in extended runs helps calibrate expectations — the drawdown profile is a consequence of the behavioural pattern, not an independent phenomenon.

The Practical Question

After studying the drawdown modal, the question is always the same: could you hold through this? A maximum drawdown of 26.4% lasting 1,268 days means your account would have been below its peak for nearly three and a half years. The median recovery of 18 days offers hope — most dips resolve quickly — but the 1,323-day stagnation period means there was a stretch where the strategy made no forward progress for over 3.5 years. That is the real test of conviction, and no backtest can tell you whether you'd pass it.