I found the Centre of Gravity indicator on a forum thread in the last week of October, and the pitch hooked me in one line: an oscillator with almost no lag. I had spent a month watching lagging indicators confirm moves after the money was gone. I loaded it on GBPUSD, took every crossover it printed for three days, and lost four trades in a row for minus $480. The indicator was not broken. I was using it in the one condition where it cannot work: a trend.
The Centre of Gravity, usually shortened to COG, was developed by John Ehlers, the engineer behind the cycle indicators I already use. It treats recent prices like weights on a see-saw and finds the balance point; when the balance point moves, price is supposed to follow. On a funded account that matters more than anywhere else, because a lagging entry can eat your daily loss limit. Everything below is built around that limit: a $100,000 account, a $5,000 daily loss limit, and my own $500 maximum risk per trade.
What the Centre of Gravity actually measures
The COG answers one question: where is the centre of mass of the last ten closes? It weights each close by its position in the window, the oldest close carrying the most weight and the newest the least. The resulting line is plotted against a signal line, a short moving average of the COG itself. A cross of the two lines is the signal.
The honest version of the zero-lag claim is narrower than the forum pitch. The COG does not predict price. It reacts to the shape of the last ten closes faster than a moving average, because of its weighting. In a range that speed is gold. In a trend that speed is a trap: the line turns every time price pauses, and each turn looks like a reversal that never comes.
Two things separate the COG from the oscillators I have already covered here. Against Ehlers Sinewave, also John Ehlers and also cycle-based, the COG is simpler: no cycle-length estimation, just a weighted average of closes. Against Stochastic, which measures where the close sits inside the recent high-low range, the COG measures the balance of the closes themselves.
The formula, worked by hand
The published formula is short, and the arithmetic is the whole lesson. Over the last n closes, with the oldest close weighted n and the newest close weighted 1:
| Step | Formula | My numbers |
|---|---|---|
| Weighted sum | Close × weight, weights 10 to 1 | 60.2763 (10 closes) |
| Plain sum | Sum of the same ten closes | 10.9643 |
| Raw COG | Minus weighted sum ÷ plain sum | Minus 5.4975 |
| Centred COG | Raw COG + 5.5 (mean of 1–10) | Plus 0.0025 |
| Signal line | 3-period average of the centred COG | Plus 0.0006 |
| Position size | $500 ÷ (stop pips × $10 per pip per lot) | 25-pip stop → 2.00 lots |
The centring step is the part most guides skip, and skipping it is why the formula looks wrong the first time you work it. The weights 1 to 10 average 5.5, so the raw ratio always sits near minus 5.5. Charting platforms add the 5.5 back, which centres the line on zero. Here is the hand-worked window behind the bullish cross I traded, ten hourly EURUSD closes from a V-shaped dip to 1.0945 and back up to 1.0990:
| Bar | Close | Weight | Close × weight |
|---|---|---|---|
| 1 (oldest) | 1.0970 | 10 | 10.9700 |
| 2 | 1.0960 | 9 | 9.8640 |
| 3 | 1.0950 | 8 | 8.7600 |
| 4 | 1.0945 | 7 | 7.6615 |
| 5 | 1.0948 | 6 | 6.5688 |
| 6 | 1.0955 | 5 | 5.4775 |
| 7 | 1.0965 | 4 | 4.3860 |
| 8 | 1.0975 | 3 | 3.2925 |
| 9 | 1.0985 | 2 | 2.1970 |
| 10 (newest) | 1.0990 | 1 | 1.0990 |
| Total | 10.9643 | 55 | 60.2763 |
Sixty point two seven six three divided by ten point nine six four three is 5.4975, so the raw COG is minus 5.4975 and the centred COG is plus 0.0025. The two previous centred readings were minus 0.0014 and plus 0.0006, so the signal line sits at plus 0.0006. The COG at plus 0.0025 is above its signal, and the line has crossed up through zero in the same move. The reading matters only against its own signal line. The formula is published in the forexbee Centre of Gravity guide.
Best settings for prop firm trading
I trade the COG at its default settings, because the defaults are the point: Ehlers designed the weighting around a 10-period window, and every change I tested made the line either too jumpy or too slow. The only setting I changed is the timeframe. On M5 the COG crosses so often that the spread alone eats the edge, and on H4 the signals are so rare that a 30-day challenge ends first. H1 for the signal and M15 for the entry is the pair that survived.
| Setting | My value | Why |
|---|---|---|
| COG period | 10 | Ehlers default |
| Signal line | 3-period average of COG | Crosses, filters noise |
| Signal timeframe | H1 | Ranges last on H1 |
| Entry timeframe | M15 | Sharper entry |
| Range filter | ADX below 20 | Blocks trend crosses |
| Range check | Price inside the last 5-day range | Confirms the range |
| News filter | No trades 30 minutes either side of red-folder news | News breaks ranges |
One practical note: the COG is a custom indicator on MetaTrader and a built-in on TradingView, so check your platform before you plan around it. On MT5 I run the free community version with the same 10 and 3 inputs, and the readings match the hand-worked values above.
How to read its signals
The COG gives four readings, and only two of them are tradeable on a funded account. The crossover is the entry signal. The extreme reading is the warning. Divergence is the confirmation. The zero cross is the context.
| Signal | What it looks like | What I do |
|---|---|---|
| Bull cross | COG crosses above signal near range low | Long at range low, stop below it |
| Bear cross | COG crosses below signal near range high | Short at range high, stop above it |
| Extreme reading | COG far from its recent range | No trade; wait for the cross back |
| Divergence | Price extremes, COG does not | Confirmation only |
| Centre cross | COG crosses zero | Context only, not an entry |
The rule that saves the account is the one I broke in October: a cross is only a trade when the range filter passes. ADX below 20 and price inside the last five-day range, both true, or I do nothing. A cross in a trend is not a cheaper entry; it is a donation to the traders already positioned with the trend. When price prints a new low and the COG refuses to follow, the sellers are tired — but I still wait for the cross.
My worked trade: a range long on EURUSD
The trade below is the one the hand-worked calculation pointed at. EURUSD had spent five days between 1.0900 and 1.0980, an 80-pip range. The COG printed its bullish cross at the dip to 1.0945, the M15 chart printed a bullish engulfing candle off the dip, and ADX sat at 14. I bought at 1.0920 on the engulfing close, with the stop at 1.0895, five pips below the range low. The risk was $500 — 10% of my $5,000 daily loss limit. My position sizing guide and maximum daily drawdown calculator run the same maths.
| Step | Level | Maths |
|---|---|---|
| Entry (long) | 1.0920 | COG cross + engulfing |
| Stop loss | 1.0895 | 25 pips, 5 under range low |
| Risk | $500 | 25 pips × 2.00 lots × $10 per pip |
| Size | 2.00 lots | $500 ÷ (25 × $10) = 2.00 |
| Take profit 1 | 1.0950 | +30 pips, half, +$300 |
| Take profit 2 | 1.0975 | +55 pips, rest, +$550 |
| Total | Plus $850 | 1.7R on $500 risk |
The same $500 risk at other stop distances:
| Stop distance | Size | Risk |
|---|---|---|
| 15 pips | 3.33 lots | $499.50 |
| 20 pips | 2.50 lots | $500.00 |
| 25 pips | 2.00 lots | $500.00 |
| 30 pips | 1.66 lots | $498.00 |
| 40 pips | 1.25 lots | $500.00 |
| 50 pips | 1.00 lot | $500.00 |
The trade did what a range trade is supposed to do: it paid for the range, not for a breakout. TP1 filled at 1.0950 for plus $300 and I moved the stop to breakeven. TP2 filled at 1.0975 for plus $550, five pips short of the range high, because I never target the exact edge of a range. The whole trade cost $34 in spread and commission: a 1-pip spread on 2 lots is $20, commission about $7 per lot round trip, and $34 is 6.8% of the $500 risk. That cost is why the COG never trades M5 on my account.
The range filter: my 16-trade test
After the four October losses I ran the test properly. Sixteen COG crossover trades on EURUSD and GBPUSD, H1 signal, M15 entry, $500 risk each, every cross taken with no filter. Then the same sixteen trades re-run with the filter applied: ADX below 20 and price inside the last five-day range. The filter removes six trades, and all six were losers.
| Trade | Pair | COG signal | Filter | Result |
|---|---|---|---|---|
| 1 | EURUSD | Bull cross | Passed | Plus $750 |
| 2 | GBPUSD | Bear cross | Passed | Plus $500 |
| 3 | EURUSD | Bull cross | Failed (ADX 31) | Minus $500 |
| 4 | GBPUSD | Bull cross | Passed | Plus $650 |
| 5 | EURUSD | Bear cross | Failed (trend) | Minus $500 |
| 6 | GBPUSD | Bear cross | Passed | Minus $500 |
| 7 | EURUSD | Bull cross | Passed | Plus $750 |
| 8 | GBPUSD | Bull cross | Failed (ADX 27) | Minus $500 |
| 9 | EURUSD | Bear cross | Passed | Plus $500 |
| 10 | GBPUSD | Bear cross | Passed | Minus $500 |
| 11 | EURUSD | Bull cross | Failed (breakout) | Minus $500 |
| 12 | GBPUSD | Bull cross | Passed | Plus $650 |
| 13 | EURUSD | Bear cross | Failed (ADX 24) | Minus $500 |
| 14 | GBPUSD | Bear cross | Passed | Plus $500 |
| 15 | EURUSD | Bull cross | Passed | Minus $500 |
| 16 | GBPUSD | Bull cross | Failed (trend) | Minus $500 |
| Run | Trades | Win rate | Net | Expectancy |
|---|---|---|---|---|
| COG only, no filter | 16 (7W / 9L) | 43.8% | Minus $200 (minus 0.4R) | Minus 0.03R per trade |
| COG plus range filter | 10 (7W / 3L) | 70.0% | Plus $2,800 (plus 5.6R) | Plus 0.56R per trade |
| Filter worth | 6 trades removed, all losers | — | Plus $3,000 (plus 6.0R) | — |
Read the two runs together and the story is clean. Unfiltered, seven wins totalling $4,300 against nine losses totalling $4,500 leaves minus $200 — the indicator on its own is a losing system on my pairs. Filtered, the same seven wins stand against only three losses totalling $1,500, for plus $2,800. The filter is worth plus $3,000, and the filtered run’s best day, $750, is 26.8% of the net — inside the 30% consistency ceiling my firm watches. Sixteen trades is a small sample, but the filter is the difference between a losing indicator and a tradeable one. The discipline that protects the test is my personal daily stop: the filter never works if two early losses put me on tilt.
Centre of Gravity versus the oscillators I already use
The COG does not replace anything on my charts. It fills the one gap my existing oscillators leave: a fast turning-point read inside a range. The neighbours table is how I decide which tool to reach for — and the honest answer to the overlap question, since the COG is not a re-skinned version of anything already on this site.
| Indicator | What it measures | Best condition | COG difference |
|---|---|---|---|
| Centre of Gravity | Balance point of the last 10 closes | Ranging market | The subject of this guide |
| Ehlers Sinewave | Market cycle phase | Cyclical turns | Needs cycle estimate |
| Stochastic | Close inside the high-low range | Range extremes | Uses wicks; COG uses closes |
| On-Balance Volume | Cumulative signed volume | Volume confirmation | Volume-based |
| Chaikin Money Flow | Volume-backed money flow | Trend participation | Needs volume data |
The practical split is simple. When the market is ranging, the COG is my first read and Stochastic is my second opinion at the extremes. When the market is trending, the COG goes away entirely and Chaikin Money Flow tells me whether the trend has participation behind it. The COG earns its place by being the fastest of the five inside a range, and it loses its place the moment the range ends.
Common mistakes
Every mistake below is one I made, most of them in the same October week. They share one root cause: treating a fast line as a certain line. Trading every cross is the first and most expensive — a cross in a trend is the market pausing, not reversing, and the ADX filter exists to catch exactly that. Taking crosses on M5 is the second: the line crosses three times as often and the spread takes three times the toll. Skipping the reversal candle is the third: the cross alone is not an entry, because the COG can cross and cross back inside a single hour. Sizing from the range instead of the stop is the fourth: a narrow range tempts a tight stop with a big size, which is how a normal 15-pip breath becomes a full $500 loss. The fifth mistake ends funded accounts: trading the COG through red-folder news. My rule is no trades 30 minutes either side of red-folder news, no exceptions.
My daily routine with the Centre of Gravity
The routine is deliberately boring. The COG is a waiting indicator: most days it gives me nothing, and the days it gives me something are the days the range has already been drawn for me. I check it twice a day and I do not watch it in between.
| Time (PKT) | Step | Check |
|---|---|---|
| 07:00 | Mark the ranges | 5-day high/low, EURUSD + GBPUSD |
| 07:10 | Read the filter | ADX below 20, or no trading |
| 07:15 | Check the news | Block 30 min around red-folder news |
| 08:00–16:00 | Wait for the cross | COG cross + M15 candle at range edge |
| On entry | Size from the grid | $500 risk, stop outside range |
| On TP1 | Bank half, stop to breakeven | Worst case is a scratch |
| 16:00 | Close the book | Max 2 trades; log both |
Two trades a day is the cap, not the target. Most days I take one or none, and a day with no trade is a good day for a funded account. The log makes the routine work: every trade goes in with its COG reading, its ADX, and whether the filter passed.
Frequently Asked Questions
Does the Centre of Gravity indicator really have zero lag?
What are the best settings for the Centre of Gravity?
Can I use the Centre of Gravity on a funded account?
Why did the Centre of Gravity keep losing in a trend?
Is the Centre of Gravity better than Stochastic?
What is the minimum account size for trading the Centre of Gravity this way?
Trading involves substantial risk of loss and is not suitable for every investor. The Centre of Gravity results in this guide come from my own sixteen-trade test on a demo and funded account, and past performance does not guarantee future results. Never risk money you cannot afford to lose, and always check your prop firm’s rules before trading any indicator on a live funded account.




