Concept

Equity-curve-based Throttling

Equity-curve-based Throttling, also known as trading the equity curve, is a Risk, Sizing & Exits concept. A reference entry: the Library explains it rather than implements it.

What is Equity-Curve-Based Throttling?

Equity-curve-based throttling, often called trading the equity curve, treats a strategy's own profit-and-loss line as an input to risk decisions. Instead of running every signal at full size regardless of recent results, the trader monitors the equity curve and reduces size, pauses trading, or restores size based on rules applied to that curve: a moving average of equity, a drawdown threshold, or a losing-streak count. The strategy generates signals as usual; the throttle decides how much capital each signal receives.

The idea exists because most strategies are regime-dependent. A trend system bleeds in ranges, a mean-reversion system bleeds in trends, and the equity curve is one place where that mismatch shows up without needing to diagnose the market directly. If performance is persistent, meaning losing periods tend to continue for a while, then cutting size after the curve deteriorates shifts capital away from hostile conditions and back in when conditions improve.

The honest caveat is that the benefit depends entirely on that persistence. If trade outcomes are close to independent, throttling systematically skips trades after losses that were about to be winners, and backtests often show it lowering total return while smoothing the ride. Practitioners therefore test whether the strategy's returns show autocorrelation, for example with drawdown statistics and streak analysis, before trusting an equity filter. Throttling is a risk-shaping tool, not a source of edge.

How traders use it

  • As an on/off filter: trade full size while equity is above its own moving average (30 to 50 closed trades is a common window), and paper-trade or trade minimum size while below it, re-engaging when the curve recovers. This tends to cap deep drawdowns at the cost of some missed recoveries.
  • As graduated de-risking: cut the fixed-fractional risk per trade in steps, for example from 1% to 0.5% after a defined drawdown, and restore it only after equity makes a new high or recovers a set fraction of the loss.
  • As a circuit breaker within loss control rules: a hard equity floor for the day, week, or month that halts trading entirely, protecting the account from tilt and from strategy decay while the trader reviews.
  • As a portfolio allocator: when running several systems, capital is tilted toward the ones with the healthiest recent curves, which overlaps with strategy switching and rotation and inherits the same whipsaw risk.
  • As a validation exercise: before deploying a throttle, traders often backtest it against the untouched strategy and against random skipping, since a filter that only smooths randomness typically reduces expectancy rather than protecting it.

Equity-Curve-Based Throttling vs related concepts

Martingale / Anti-Martingale: Martingale-family rules scale the next trade off the last outcome, one trade deep. Equity-curve throttling reads a longer window of the curve (averages, drawdowns, streak counts) and usually reduces rather than increases risk after losses, making it an anti-martingale applied at the account level.

Drawdown Statistics: Drawdown statistics measure and describe the equity curve after the fact. Throttling is the acting layer built on those measurements: it converts a drawdown reading into a size or on/off decision.

Strategy Switching & Rotation: Rotation reallocates capital between different systems based on relative performance. Throttling manages exposure to a single system against its own history; the two are often combined in multi-system accounts.

Concept family

Risk, Sizing & Exits

37 concepts mapped · 37 in the Library

Equity-curve-based Throttling FAQ

Turn Equity-curve-based Throttling into a trading strategy.

Describe your Equity-curve-based Throttling idea to Quant. It builds the strategy with you and backtests it on real data.