I lost a funded account’s good week in one afternoon because I kept buying a falling market. It was a Thursday, US100 had been sliding for three sessions, and every small bounce looked like the turn. I bought the first bounce, stopped out. I bought the second bounce, stopped out again. By the third attempt I was not trading a setup at all; I was arguing with a downtrend, and the argument cost me $1,340 before I finally closed the platform. That evening I looked at the weekly chart properly for the first time that week, and the picture was embarrassingly clear: the long-term momentum line had been falling for a month, and every bounce I had bought was just noise inside a decline I had refused to see.
In this article
That week is why a slow indicator now sits on my charts. The Coppock Curve, created by economist Edwin Coppock and first published in Barron’s in 1962, was designed to spot when a long decline in a major index had finally turned. It is a momentum line built from two rates of change, smoothed by a weighted average, and it moves at the speed of a glacier compared with an RSI or a stochastic. On a funded account that slowness is not a weakness. It is the point: funded accounts are lost by overtrading against the bigger picture, not by missing one fast entry.
This article explains the Coppock Curve honestly for funded trading: what it measures, its formula worked by hand, my weekly-bias settings, the three signals I trust, a 20-trade journal test, the mistakes that turn it into a slow bleed, and how it fits the rules that end funded accounts.
What the Coppock Curve actually measures
The Coppock Curve does not measure price. It measures how fast price has changed over two long lookbacks, added together and smoothed. When the line is above zero, the market has risen over those lookbacks on balance; when it is below zero, it has fallen. When the line turns up from below zero, momentum is shifting from negative to positive after a decline, which is the classic Coppock buy signal. It was never designed to give sell signals, and I do not force it to: a falling Coppock line is a reason to stop buying, not an automatic reason to sell.
That design matters on a funded account because most of my expensive mistakes are not bad entries. They are entries taken in the wrong direction for the week. A fast oscillator can look oversold on an hourly chart while the weekly momentum is still falling, and buying that hourly oversold reading is exactly how I lost that Thursday. The Coppock line answers a different question from the entry indicator: not “is this bounce tradeable” but “is this market even rising over the last few months”. If the answer is no, I either skip the long or I demand a much better entry, a smaller size, and a faster exit.
One honest limitation first. Coppock was built for monthly index charts, and spot forex has no single index to apply it to. I apply it to the instrument I trade — US100, EURUSD, XAUUSD — on a weekly chart, and treat the result as a bias reading for that instrument, not a market-wide prophecy. It also lags by construction: by the time the line turns, part of the move has already happened. I accept that lag because I am not using it to pick bottoms, only to avoid fighting the trend.
The formula, worked by hand
The formula is short enough to do on paper, which is why I trust it more than indicators I cannot explain. First, take the rate of change over 11 periods and over 14 periods. Then add them. Then smooth the sum with a 10-period weighted moving average, which gives the most recent value a weight of 10, the one before it 9, and so on down to 1.
| Step | What you calculate | Formula |
|---|---|---|
| 1 | 11-period rate of change | ROC11 = (Price now − Price 11 periods ago) ÷ Price 11 periods ago × 100 |
| 2 | 14-period rate of change | ROC14 = (Price now − Price 14 periods ago) ÷ Price 14 periods ago × 100 |
| 3 | Add the two rates | Sum = ROC11 + ROC14 |
| 4 | 10-period weighted moving average | Coppock = (10×Sum now + 9×Sum last period + … + 1×Sum 9 periods ago) ÷ 55 |
Here is a worked example on a weekly US100 chart. Suppose the index closes this week at 21,500. Eleven weeks ago it closed at 20,600, and fourteen weeks ago at 20,100. The 11-week rate of change is (21,500 − 20,600) ÷ 20,600 × 100 = 4.37%. The 14-week rate is (21,500 − 20,100) ÷ 20,100 × 100 = 6.97%. Added together, this week’s sum is 11.34. That single sum is not the Coppock value yet; it still has to be smoothed with the nine previous weekly sums.
Suppose those ten weekly sums, newest first, are 11.34, 10.20, 9.10, 7.80, 6.40, 5.00, 3.60, 2.10, 0.50 and −1.20. Weighting them 10 down to 1 gives 113.4 + 91.8 + 72.8 + 54.6 + 38.4 + 25.0 + 14.4 + 6.3 + 1.0 − 1.2 = 416.5. Divide by 55 and the Coppock value is 7.57 — above zero and, if last week’s value was lower, rising too. That is a green bias for longs. Doing it once by hand is what stops me treating the line as magic.
My settings for funded trading
I do not change the classic 11, 14 and 10 numbers. Changing them to chase faster signals would destroy the one thing the indicator is good at, which is slowness. What I change is the chart and the job: the Coppock line lives on the weekly chart and decides bias, while entries come from price action on the daily and four-hour charts. That split — slow bias, fast entries — is the whole method.
| Job | Chart | Settings | Rule |
|---|---|---|---|
| Bias filter | Weekly | 11 / 14 / 10, classic | Above zero and rising: longs only; below zero and falling: shorts only or flat |
| Context check | Daily | Same 11 / 14 / 10, read only | Daily line must agree with the weekly line before I add size |
| Entry | H4 / Daily price action | No Coppock on this chart | Break-and-retest or pullback entry, stop beyond structure |
| Review | Weekly close, Sunday | Record the value in my journal | One number a week; if the bias flipped, next week’s plan flips with it |
The Sunday review is boring: I write down the weekly Coppock value, whether it is above or below zero, and whether it rose or fell that week. That single line decides the direction of my whole next week and takes about two minutes. A filter that only changes once a week cannot be overtraded, and on a funded account fewer decisions usually means fewer breaches.
One setting I will not use: Coppock on an intraday chart. On a 15-minute chart, 11 and 14 periods cover a few hours, and the “long-term momentum” the indicator was built to measure no longer exists. The line whipsaws, gives false turns all day, and invites exactly the overtrading it is supposed to prevent. If a setting makes a slow indicator fast, it is not an optimisation. It is a different indicator wearing the same name.
The three signals I actually use
I use only three readings, and I ignore everything else the line does. The first is the zero cross from below: the line turns up and crosses above zero after a decline. On my weekly chart that is permission to look for longs, not an order to buy at market. The second is the rising-above-zero continuation: the line is already positive and still rising, my normal trading state. The third is the falling-below-zero warning: the line crosses down or keeps falling while negative. That is not a sell signal for me; it is a stop-buying signal, and it is the one that would have saved that Thursday.
| Signal | What the line does | What I do |
|---|---|---|
| Turn up from below zero | Crosses above zero after a decline | Start looking for long entries on H4/Daily; normal size |
| Rising above zero | Positive and still climbing | Trade longs freely within my daily stop; this is the green state |
| Falling below zero | Crosses down or keeps falling while negative | No new longs; shorts only with a proven setup, or flat for the week |
Notice what is missing: there is no “sell when the line falls” rule, no divergence trade, and no overbought level, because a strong trend can keep the line high for months while price keeps rising. I once shorted a merely strong line on US100 and paid for it. Now a high line means the trend is healthy, not tired.
I tested it: 20 trades with and without the filter
I tested the filter the only way that means anything to me: I replayed 20 of my own US100 pullback trades from my journal, first with the filter applied and then without it, keeping the same entries, stops and targets in both counts. The account in the test is a $100,000 funded account with a 5% daily loss limit of $5,000. My personal daily stop sits at half that limit, $2,500, and my risk per trade is 0.5% of the account, which is $500. Nothing in the test changes those numbers; the only change is whether a trade against the weekly Coppock bias is allowed in.
| Version | Trades taken | Wins / Losses | Net result | Value at $500 risk |
|---|---|---|---|---|
| Without the filter (all 20 trades) | 20 | 9 wins, 11 losses | 9 × 1.8R − 11 × 1R = +5.2R | +$2,600 |
| With the Coppock filter (bias trades only) | 14 | 9 wins, 5 losses | 9 × 1.8R − 5 × 1R = +11.2R | +$5,600 |
| Trades the filter removed | 6 | 0 wins, 6 losses | −6.0R | −$3,000 avoided |
The result surprised me by how clean it was: all six trades the filter removed were losers, all bought against a falling weekly line. The filter did not find me better entries. It removed my worst habit, which was buying bounces in a decline. The filtered version also traded less — 14 trades instead of 20 — and that matters under prop firm rules, because fewer trades means fewer chances to touch the daily limit and a cleaner consistency record. I checked the sizing maths on the max daily drawdown calculator before writing this up, and the worst filtered day in the test was three losses totalling $1,500, inside my $2,500 personal stop. The worst unfiltered day was five losses totalling $2,500, exactly at the stop, with no room left for anything else.
Position sizing in the test follows one rule: the lot size fits the stop, never the other way round. At $500 risk on EURUSD, where one standard lot moves about $10 per pip, the grid is:
| Stop distance | Lots at $500 risk | What it means |
|---|---|---|
| 20 pips | 2.50 | Tight H4 pullback stop; only with a clean structure level |
| 30 pips | 1.67 | My usual H4 pullback stop |
| 40 pips | 1.25 | Daily-chart pullback stop |
| 50 pips | 1.00 | Wide daily stop; fewer trades, same dollars at risk |
| 60 pips | 0.83 | Weekly-structure stop; patience trade, not a scalp |
I am not claiming the filter doubles every trader’s results. Twenty trades is a small sample from my own journal, not a blind backtest, and a different 20 could look worse. What the test proves to me is narrower: on my own trading, the expensive trades were the ones taken against the weekly bias, and a one-number weekly check removed them.
Common mistakes that turn a slow filter into a slow bleed
The first mistake is using Coppock as an entry trigger. The line turns weeks after a bottom, so buying the turn itself means buying late, often into the first pullback of the new trend. I use the turn as permission, then wait for a pullback entry with its own stop. The second mistake is running it on a fast chart, which I covered above: a 15-minute Coppock is noise with a famous name. The third mistake is trading both directions off one line. Coppock was built to find buying opportunities after declines; it has no symmetrical sell logic, so I do not invent one. When the line is falling, my default is flat, not short, unless a separate short setup I already trust is present.
The fourth mistake is the quietest: ignoring the line when it disagrees with a trade I already want. If I override the filter whenever it is inconvenient, I have not got a filter at all; I have got decoration. My journal rule is literal: a trade against the weekly bias needs a written reason, a half-size position, and a same-day exit plan. Writing that sentence out usually talks me out of the trade, which is exactly what it is for. Related reading on the same discipline: my notes on losing streaks on a funded account and on setting a personal daily stop below the firm limit.
How it fits prop firm rules
Prop firm rules reward exactly what a slow filter produces: fewer trades, one direction per week, and no chasing. The daily loss limit is protected because the filter removes the revenge-style bounce trades that cluster on bad days; my test’s worst filtered day was $1,500, leaving $1,000 of room inside my personal stop, compared with the unfiltered version. The consistency rule is protected because a week of bias-aligned trades produces steadier daily results than a week of long-short flip-flopping, and the best-day share stays under control when no single counter-trend gamble dominates the week. The minimum-trading-days style rules are unaffected, since the filter changes direction, not frequency enough to matter: 14 trades across a month still covers the required days comfortably.
There is one rule interaction to watch. Because a weekly line may stay in one state for months, a trader who waits only for Coppock turns will trade too rarely and may overtrade out of boredom in the gaps. My answer is that the filter does not decide whether I trade; it decides which direction I trade when my normal setups appear, as in my trend pullback plan and my Klinger volume notes. The Coppock line only vetoes the ones pointing the wrong way, and I check the veto’s maths on the strategy expectancy calculator before changing any filter rule.
For the indicator’s history, the Coppock curve reference gives the cleanest summary of the 11, 14 and 10 construction and the mourning-period story behind those numbers: 11 and 14 were not computer-optimised, they came from how long people take to recover from a loss — a reminder that the line measures sentiment recovering, not a law of physics.
Frequently asked questions
Can I use the Coppock Curve for day trading a funded account?
Not as a bias filter, no. On intraday charts the 11 and 14 periods cover hours, not months, and the line becomes noise. Use the weekly line for bias and take intraday entries from price action that agrees with it.
Does Coppock give sell signals?
By design, no. It was built to find buying opportunities after declines. A falling line tells me to stop buying; it does not by itself tell me to sell.
What if the weekly Coppock line and the daily line disagree?
The weekly line wins, and I trade half size or not at all until they agree. Disagreement is information: it usually means the market is in transition, and transition weeks are when funded accounts get hurt.
Should I change the 11, 14 and 10 settings for forex?
I do not. Those numbers are the indicator’s identity, and changing them to make it faster destroys its purpose. Change the chart — weekly instead of monthly — not the numbers.
How often does the weekly Coppock bias actually change?
Rarely, and that is the point. On a weekly chart the bias may hold for months. A filter that changes once a quarter cannot be overtraded, which is precisely why it protects a funded account.
Can Coppock alone pass a prop firm challenge?
No. It is a filter, not a strategy: it has no entry, stop or target of its own. It only improves a strategy that already has those rules, by removing the trades taken against the bigger trend.
Risk disclaimer: this article describes my own journal test and trading routine for education only. It is not financial advice, and past results — mine or anyone else’s — do not guarantee future results. Funded trading involves the risk of losing the account, including through rule breaches that have nothing to do with market direction.





1 comment
[…] copies of that article, because the smoothing and the weighting change what the line means. And my Coppock Curve article also blends two ROCs, but Coppock is a slow monthly bias filter built from ROC 11 and ROC 14 with a […]