Skip to main content

Drawdowns & Losing Streaks

Updated Jun 24, 2026 at 2:35pm

Research Draft Medium 1,151 words

A drawdown is the peak-to-trough decline in account equity; a losing streak is a run of consecutive losing trades. For the swing trader they are not signs of a broken edge — they are the guaranteed background noise of any positive-expectancy system. The core psychological tension is that drawdowns are mathematically inevitable yet feel, in the moment, like proof the strategy has stopped working. The discipline this node covers is the ability to distinguish normal variance from a genuinely broken edge, size and behave correctly through the trough, and avoid the behavioral spiral (revenge trading, deviation, doubling down) that turns a survivable dip into account death.

How it's calculated / formed

Drawdown = (peak equity − current equity) / peak equity. Maximum drawdown (MaxDD) is the worst such decline over a period — the standard measure of downside pain.

Losing-streak probability follows the binomial. If the per-trade loss rate is q, the probability of n losses in a row at a given starting point is q^n. With a 40% loss rate, P(5 straight) = 0.40^5 ≈ 0.0102, roughly 1 in 98 (BacktestBase gives the same formula and notes a 3-loss streak at a 40% loss rate is 0.40³ = 6.4%). The key intuition is that over many trades, runs that are individually rare become near-certain somewhere in the sequence: BacktestBase's table indicates that at a 60% win rate you should expect a longest run of roughly 7–8 consecutive losses over 1,000 trades, and ~10–11 over 10,000. A standard order-of-magnitude estimate for the expected longest losing run is log(n)/log(1/q) for n trades, after Schilling's result on the predictability of long runs (Schilling 1990).

Recovery asymmetry is the cruel part, and it's a pure identity: required gain = 1/(1 − DD) − 1. A 10% drawdown needs an 11.1% gain; 25% needs 33.3%; 50% needs 100%; 75% needs 300% (Intellectual Finance, TradeZella). Deep holes scale geometrically, which is why prevention beats recovery.

How it's used in practice

A master swing trader treats drawdown as an engineering and behavioral problem, not an emotional one.

  • Size for the worst expected streak, not the average. Estimate the worst plausible losing streak from your win rate and trade count (Monte Carlo, or the log formula above), then set risk-per-trade so that streak stays inside your tolerable drawdown. As a rough illustration: if you expect a worst run of ~8 losses and cap drawdown at 20%, fixed-fractional risk would need to be at most ≈ 20% / 8 = 2.5% per trade (and lower once you account for runs longer than the expectation). This is the author's own worked example, not a figure from any cited source; it leans on the 1–2% risk-per-trade rule, which most practitioners treat as the safer default.
  • Pre-commit a circuit breaker. Define a hard line before trading — e.g. stop opening new positions at a fixed equity drawdown, or use equity-curve trading: pause new entries when account equity falls below its 20-day moving average and resume only when it crosses back above (Enlightened Stock Trading, MQL5). Caveat: equity-curve overlays can hurt strictly systematic strategies whose drawdowns are pure variance, so they suit discretionary/regime-sensitive swing books better.
  • Diagnose variance vs. broken edge. Variance: you're following the plan, setups still match conditions, R per trade is in historical range. Broken edge: market regime has structurally changed (e.g. trend strategy in a chop regime), or your behavior has drifted (rule-breaking, oversized entries, revenge trades). The first calls for reduced size and patience; the second calls for stopping. The honest tell is the journal — see Journaling & Self-Review.
  • Reduce, don't increase, exposure in the hole. Trimming risk during an active drawdown both protects capital from the recovery-asymmetry math and lowers the emotional stakes of each trade, which improves execution.
  • Key failure trigger to watch: raising size to "win it back faster." Recovery asymmetry guarantees this is the exact moment larger bets are most dangerous.

Adoption, debate & evidence

Loss aversion is among the most robust findings in behavioral economics: Tversky & Kahneman's 1992 cumulative prospect theory put the loss-aversion coefficient at λ ≈ 2.25 — losses feel a bit more than twice as painful as equivalent gains (Neuroprofiler, Decision Lab). (The exact value is debated; a 2024 meta-analysis finds the 2.25 figure may be too high and that loss aversion varies across people and contexts — so treat "roughly 2x" as a robust direction, not a fixed constant.) The related disposition effect — selling winners early and holding losers — is documented even among profitable professionals (Garvey & Murphy, "Are Professional Traders Too Slow to Realize Their Losses?", Financial Analysts Journal, 2004). The "revenge trading" reflex is empirically real: Coval & Shumway (Journal of Finance, 2005) found CBOT proprietary traders are ~16% more likely to take above-average afternoon risk after morning losses. So the behavioral hazards of drawdowns are evidence-backed, not folklore.

What is contested or commonly overstated: precise prescriptions. Claims like "size below 1% during drawdowns measurably reduces cortisol-driven errors" circulate in trading-education content but rest on weak or absent peer-reviewed support — treat them as plausible heuristics, not established fact. The math of streaks and recovery is exact and uncontroversial; the behavioral coaching layer is largely practitioner consensus. Van Tharp's framing — that strategy is a minor share of trading success versus psychology and position sizing — is influential but a teaching device, not a measured decomposition.

Strengths & limitations

The strength of treating drawdowns quantitatively is that it converts a panic trigger into a known parameter: if you've pre-computed your expected streak and sized for it, a six-loss run is expected output, not an emergency. The framework works best when you have a real sample (ideally 30+ trades, per the system-metrics node) so streak/drawdown estimates mean something.

It fails when (1) the historical sample is too small or drawn from one favorable regime, so the real worst streak dwarfs the modeled one; (2) the trader confuses a regime-driven broken edge with variance and keeps "trusting the process" into ruin; or (3) the math is known but ignored under emotional load. The #1 misuse is increasing position size during a drawdown to recover faster — directly fighting the recovery-asymmetry identity and the most common path from a recoverable 20% dip to a fatal one.

Sources

Disputes flagged: the math of streaks/recovery is exact; the behavioral-coaching layer (e.g. cortisol/sizing claims, "psychology = 90% of success") is practitioner consensus without strong peer-reviewed backing and is qualified accordingly. Loss aversion, disposition effect, and revenge-trading are academically supported.