An equity curve filter applies a simple technical rule, usually a moving average, to your strategy's own profit-and-loss history. When the curve falls below its average, you pause the system or cut size. When it recovers, you scale back up. This is the core of equity curve trading: you judge the system's health, not each individual trade.
It sounds like free protection. It isn't. Used well, it trims size during bad stretches. Used badly, it makes you miss the recovery and adds one more setting to overfit. Here is how to build it, test it and decide whether your scalping system should use it at all.
TL;DR
- An equity curve filter compares your strategy's equity to a moving average of that equity, then reduces or stops trading when equity sits below it.
- Two ways to run it: binary on/off, or dynamic position sizing. Sizing down, never to zero, is the gentler option.
- Keep a "shadow" equity curve running while live trading is paused. Without it you can't know when to restart.
- The filter adds parameters. Test it across many market regimes, or you are curve-fitting a second time.
- It is a defensive tool, not a fix for a strategy with no edge.
Table of Contents
- What an equity curve filter actually does
- Two ways to implement it
- The shadow equity curve: the step most traders skip
- Choosing the filter period
- Does it even work? The honest answer
- Avoiding curve-fitting the filter
- Perspective: where we land on equity curve filters
- Applying it to a scalping workflow
- Risk limits the filter doesn't replace
- Build the signals, then test the filter
- FAQ
- Sources
What an equity curve filter actually does
Your equity curve plots account value after each trade, or after each bar if open trades are included. A filter smooths that line, most often with a simple or exponential moving average, and uses the crossover as a switch.
The logic has three parts:
- The curve. Closed-trade P&L or mark-to-market equity of one strategy.
- The filter. A moving average (or similar) of that curve over N trades or N bars.
- The action. Trade normally when equity is above the filter. Reduce or pause when it is below.
The goal is to keep the strategy from bleeding during a regime that doesn't suit it, or during outright failure. Practitioners describe it as a pause trigger for a strategy that stops working. That is the honest scope. It is not a prediction tool.
Two ways to implement it
Practitioners use two main methods: switch the system fully off and on at the crossover, or scale position size up and down depending on where equity sits relative to its average.
| Method | How it works | Strength | Weakness |
|---|---|---|---|
| Binary switch | 100% size above the filter, 0% below | Simple, clear rule | Whipsaws near the average, can miss the first winners of a recovery |
| Dynamic sizing | e.g. 100% above, 25-50% below | Smoother risk, you stay in the game | Needs a floor and a ceiling rule |
| Tiered sizing | Size steps with distance from the average | Fine control | More parameters, more overfitting risk |
For most retail scalpers, a floor above zero is the practical choice. One common description of the full-stop version is that you stop trading in real time and keep plotting equity in simulation. Cutting to a reduced size does the same job while keeping you exposed to the recovery.
A worked example
Say your scalping system trades a fixed 1R risk per trade, with R defined as your risk per trade. You track cumulative R after each trade and apply a 20-trade simple moving average to that cumulative R line.
- Cumulative R is 14.0 and its 20-trade average is 11.5. Equity is above the filter. Risk 1R per trade.
- A losing run drops cumulative R to 10.2. The average is now 11.8. Equity is below the filter. Risk 0.5R per trade.
- Cumulative R climbs back to 12.4, above the 11.9 average. Back to 1R.
The numbers are illustrative, not a recommendation. They show the mechanic: the rule needs no opinion about why the system lost.
The shadow equity curve: the step most traders skip
If you stop recording results when you stop trading, the filter freezes. The curve can't rise above its average with no new data, so you never restart.
The fix is a continuous virtual curve. The simulation must keep running while the live version is paused, so you have data to decide when to resume.
In practice:
- Log every signal your system generates, taken or not, with its entry, stop and target.
- Compute hypothetical R for each one using the same rules.
- Build the filter on that hypothetical series, not on your live fills alone.
With non-repainting signals confirmed on candle close, this is far more reliable, because the signal you log is the signal you would have seen live. A repainting indicator would give you a shadow curve that never existed. We cover why that matters in our guide to backtestable breakeven stop rules.
Choosing the filter period
The period, N, is where most traders go wrong. The trade-off is simple:
- Short N reacts fast but tends to be overwhelmed by transaction costs and signal chopping. You flip between full and reduced size constantly.
- Long N is steadier but lags at turning points. You are still at full size when a slump begins and still reduced after it ends.
A workable approach for scalpers:
- Measure in trades, not days. A scalper may take dozens of trades per day, so a 20-trade filter can flip within a session.
- Test a range, such as 10, 20, 30, 50 and 100 trades. Judge the whole range, not the single best value.
- Prefer a setting where neighboring values behave similarly. If N=20 works and N=19 and N=21 don't, you found noise.
- Count how often the filter flips. Many flips mean many size changes, which means extra slippage and commissions.
Does it even work? The honest answer
Sometimes. Not always. Treat any claim of guaranteed improvement with suspicion.
A systematic-trading analysis using random data found the technique's value depends on the type of system. Trend-following systems tend to alternate between win and loss streaks in a way that makes a restrictive filter counterproductive. Mean-reversion systems, where streaks are less clustered, showed more promise.
A long-running trader discussion reaches a similar caution: applying simple indicators to an equity curve can bring more downside than upside, because it adds parameters and does nothing about a strategy's lack of robustness.
The same logic applies to any extra layer you bolt onto a model. Before trusting a signal or a feed, check it for noise and bias, as developers who work with market data warn when a headline accuracy figure hides a weak signal.
Filter or fix?
Drawdown is the cost of harvesting any edge. A filter is a defensive stance, not an elimination tool. The key question is whether your system is in a quiet "out of sync" period or has fundamentally failed from over-optimization.
- If the edge is real but the market regime is wrong, a filter can reduce exposure until conditions return.
- If the edge was never there, a filter only slows the losses. Go back to the strategy.
Avoiding curve-fitting the filter
You already risk overfitting your entry rules. The filter is a second layer. Traders often curve-fit filter settings to past data, then feel protected by something that was tuned to history. Experts advise testing across a broad range of market regimes, not one backtested window.
A testing checklist:
- Split your data. Choose N on one period. Evaluate it on a period you never touched.
- Cover regimes. Include trending days, choppy ranges, high-volatility news sessions and quiet ones.
- Compare to baseline. Same strategy, no filter. Compare max drawdown, recovery time and net R, not just the final profit.
- Include costs. Commissions, spread and slippage on every size change.
- Check the neighbors. Adjacent N values should give similar results.
- Run it forward. Paper trade the filter live before putting real size behind it.
If the filter only improves one metric in one window, discard it.
Perspective: where we land on equity curve filters
We build signal tools for 1m to 15m scalping, so we see this question from the execution side. Our view is practical.
First, fix the inputs before filtering the outputs. A filter on the curve of a repainting or loosely defined system measures fiction. That is why our indicators confirm signals on candle close and ship as open-source Pine Script v6: you can read the logic, log the signals and trust that a historical signal is the one that fired live.
Second, use volatility and confluence gates on the trade, then size on the curve. Gating entries by volatility and requiring multiple confirmations tends to reduce bad trades at the source. A curve filter then handles what remains: a stretch when the market simply doesn't suit your setup. Our 2 of 4 signal rule for equities shows the confluence idea in practice.
Third, prefer sizing down to switching off. A scalper who goes to zero after a loss streak often misses the sharp recovery days. Halving risk keeps you in rhythm and limits damage.
Fourth, keep it boring. One filter, one reasonable period, one floor. Every extra knob is another chance to fool yourself.
We don't claim a curve filter makes a weak system profitable. It is a risk valve, and you should test it like any other rule.
— Tran
Applying it to a scalping workflow
Here is a lean setup you can build in a spreadsheet or a TradingView strategy script.
- Define R. Fix your risk per trade as 1R. Use fixed stop distances or a projected stop from your indicator.
- Log every signal. Include skipped and paused ones. Record outcome in R.
- Plot cumulative R. This is your shadow curve.
- Add a moving average. Start with 20 or 30 trades and test around it.
- Set two states. Full risk above the average. Reduced risk, say half, below it.
- Add a re-entry rule. Return to full risk only after the curve closes above the average, not the instant it touches it.
- Review monthly. Check flips, costs and whether the filter beat the baseline.
Your exits shape the curve too. A consistent plan for partial profits, such as the approach in our guide to scaling out at the first target, produces a smoother curve, which makes any filter easier to evaluate. The same goes for fixed reward-to-risk targets, as covered in our piece on multi-timeframe scalping signals.
Risk limits the filter doesn't replace
An equity curve filter reacts after losses. It does not cap a single bad trade or a gap. Keep hard rules underneath it: a per-trade stop, a daily loss limit and a maximum position size. Our risk management checklist lists the basics.
Keep the odds in view as well. The CFTC states that two out of three retail forex traders lose money each quarter. A filter doesn't change that base rate. If you trade leveraged products, read our risk disclosure and size so a losing streak is survivable. Talk to a registered financial professional if you are unsure how much of your capital can sit in leveraged trading.
Build the signals, then test the filter
A curve filter is only as good as the signals under it. Scalping-Algo's TradingView suite gives you non-repainting buy and sell signals on 1m to 15m charts, volatility gating, multi-timeframe confluence, projected TP/SL, Discord webhook alerts and a backtesting dashboard. Pair that with your own shadow-curve log and you have clean data to test with.
Plans are $79 per month, $799 per year or a one-off $1999 for lifetime access. See the plans and full indicator suite, or start with Smart Scalping Signals.
FAQ
What is the best moving average period for an equity curve filter?
There is no universal best. Shorter periods react quickly but suffer from chopping and costs, while longer ones lag at turning points. Test a range, compare neighboring values and choose a setting that sits in a stable zone, not a single peak.
Should I stop trading completely when equity falls below the average?
Not necessarily. Many traders prefer cutting size to a floor, such as half, so they stay exposed to the recovery. Whichever you choose, keep recording a shadow curve so you know when to resume.
Does trading the equity curve work for every strategy?
No. Research on random data suggests it may be counterproductive for trend-following systems with alternating win and loss streaks, and more promising for mean-reversion systems. Test it on your own strategy against an unfiltered baseline before using it live.
Can I run an equity curve filter on TradingView?
Yes, in a strategy script you can compute a moving average of the strategy's equity and scale position size from it. A public example script uses fast and slow averages on the equity curve to cut size during drawdowns. Backtest it carefully, since results on one chart and period prove little.
Sources
- Equity Curve Trading: When to Pause a Strategy That Stops Working
- Random data: Evaluating "Trading the equity curve"
- Trading the equity curve doesn't seem to work (Elite Trader)
- 4 Ways to Trade Through Drawdowns
- CFTC: Forex Frauds
- Trading the Equity Curve Position Sizing Example (TradingView script listing)