Rate of Change: The Simplest Momentum Measure
Rate of change, often written ROC, measures how much price has changed over a set number of periods, expressed as a percentage. It is the plainest way to quantify momentum.
The calculation is direct: today's close minus the close N periods ago, divided by the close N periods ago, times one hundred. A 12-period ROC on a daily chart tells you the percentage change over the last 12 days.
Positive values mean price is higher than it was N periods ago, and negative values mean lower. The zero line separates the two. The magnitude shows how strong the move has been.
Uses include trend confirmation, since sustained positive ROC supports an uptrend. Another use is comparing momentum across pairs, as the percentage figure is comparable between them, which helps rank which currencies are moving most. A third is spotting momentum loss, when price makes a new high but ROC does not, a form of divergence.
Because ROC is a lagging measure that depends on a single past price, it can jump when an old high leaves the window. That quirk is worth knowing, and smoothing with a moving average helps.
Momentum strategies in currencies have, in academic studies, shown modest returns over longer periods such as one to twelve months, though results vary by sample.
Practical step: compute the 20-day ROC for four major pairs, and rank them by strength.
Rate of Change: The Simplest Momentum Measure
Rate of Change: The Simplest Momentum Measure
It’s Fairman 
Re: Rate of Change: The Simplest Momentum Measure
Ranking pairs by ROC every weekend is a quick way to see which currencies are actually moving. I like it for picking which pair to focus on rather than for entries. A 20 day ROC on the main pairs takes a couple of minutes to check. If GBP shows strong positive readings against several currencies and JPY shows strong negative ones, GBPJPY or GBPUSD might be worth focusing on for trend setups that week. Pairs with ROC near zero are probably ranging, which suits a different approach. It's not precise, but it stops you spending the week on a pair that isn't going anywhere while something else trends cleanly.
It’s Fairman 
Re: Rate of Change: The Simplest Momentum Measure
The drop off effect is worth remembering. A sudden ROC jump can come from an old big candle leaving the window, not anything happening today. Say there was a large down day 12 days ago. When that day drops out of a 12 period window, ROC jumps up even if today's price barely moved. That can look like a momentum signal when it's just arithmetic. One way to reduce the effect is to smooth ROC with a short moving average, or to compare two lookback periods. If both show the same direction, the signal is more trustworthy. If only one moves, check what left the window.
It’s Fairman 
-
FXS Prop Desk
- Posts: 1689
- Joined: Wed Oct 07, 2026 9:52 am
Re: Rate of Change: The Simplest Momentum Measure
A useful refinement is to normalise ROC by volatility, dividing the percentage change by ATR as a share of price. Otherwise, a 2% ROC means very different things on EUR/USD and on GBP/JPY. The normalised version allows comparing momentum across pairs, which is helpful for ranking which instruments currently have the strongest trend.
Plotting ROC with a zero line and a band of one standard deviation of its own history also makes extremes easier to spot without guessing at thresholds.
Plotting ROC with a zero line and a band of one standard deviation of its own history also makes extremes easier to spot without guessing at thresholds.
FXS Official Team | The Idea Lab & Strategy
-
Shadow Trader
- Posts: 5594
- Joined: Wed Sep 09, 2026 5:19 pm
Re: Rate of Change: The Simplest Momentum Measure
One detail that matters for the weekly ranking is quote direction. If EURUSD shows a 20 day ROC of plus 2 percent, the dollar measured against the euro is down about 1.96 percent, not 2, because 1 divided by 1.02 is 0.9804.
On a couple of percent the gap is tiny, but on a 10 percent move in USDJPY it becomes 10 against 9.1. Using the log of the price ratio fixes it, since a log change of 0.0198 one way is exactly minus 0.0198 the other way, and log changes add up cleanly across days.
The second catch is the candle count. A broker whose server runs on GMT prints a short Sunday candle each week, so 20 bars there covers a little over three weeks, while a server set to the New York close gives five bars a week and 20 bars is four full weeks.
That alone can make the same pair rank differently on two platforms. If you compare readings with anyone else, agree on the close time first.
Do people here rank on the raw percentage or on the volatility adjusted version mentioned above? Raw figures tend to favour the high volatility crosses like GBPJPY whether or not their trend is any cleaner.
On a couple of percent the gap is tiny, but on a 10 percent move in USDJPY it becomes 10 against 9.1. Using the log of the price ratio fixes it, since a log change of 0.0198 one way is exactly minus 0.0198 the other way, and log changes add up cleanly across days.
The second catch is the candle count. A broker whose server runs on GMT prints a short Sunday candle each week, so 20 bars there covers a little over three weeks, while a server set to the New York close gives five bars a week and 20 bars is four full weeks.
That alone can make the same pair rank differently on two platforms. If you compare readings with anyone else, agree on the close time first.
Do people here rank on the raw percentage or on the volatility adjusted version mentioned above? Raw figures tend to favour the high volatility crosses like GBPJPY whether or not their trend is any cleaner.
-
Shadow Trader
- Posts: 5594
- Joined: Wed Sep 09, 2026 5:19 pm
Re: Rate of Change: The Simplest Momentum Measure
For intraday use the lookback is easier to judge in clock time. A 12 period ROC on M5 covers exactly one hour, so at the end of the first London hour it measures the whole opening move.
Say EURUSD sits at 1.1000 and rises 15 pips in that hour, which gives a reading of about 0.14 percent. On its own that looks like momentum, though most mornings produce something similar.
The chart (illustration) shows ten made up mornings with first hour moves between 0.08 and 0.16 percent, averaging 0.12, so today's 0.14 is ordinary. A reading of 0.30 on the same measure would stand out as a genuinely unusual open.
Comparing the same hour across days strips out the session effect without needing ATR or any other adjustment. It is a percentile view built from a short list of numbers.
Is anyone using ROC on the lower timeframes at all, or only for the weekly pick of which pair to watch? The drop off quirk is much harsher on five minute bars, since the opening candle leaves the window exactly an hour later.
Say EURUSD sits at 1.1000 and rises 15 pips in that hour, which gives a reading of about 0.14 percent. On its own that looks like momentum, though most mornings produce something similar.
The chart (illustration) shows ten made up mornings with first hour moves between 0.08 and 0.16 percent, averaging 0.12, so today's 0.14 is ordinary. A reading of 0.30 on the same measure would stand out as a genuinely unusual open.
Comparing the same hour across days strips out the session effect without needing ATR or any other adjustment. It is a percentile view built from a short list of numbers.
Is anyone using ROC on the lower timeframes at all, or only for the weekly pick of which pair to watch? The drop off quirk is much harsher on five minute bars, since the opening candle leaves the window exactly an hour later.
-
Shadow Trader
- Posts: 5594
- Joined: Wed Sep 09, 2026 5:19 pm
Re: Rate of Change: The Simplest Momentum Measure
Autocorrelation measures how strongly a return relates to previous returns. At very short horizons in forex, the sign of autocorrelation can change with the scale and the conditions. Positive autocorrelation implies momentum, and negative implies reversal.
Over tick or second-level intervals, bid-ask bounce creates negative autocorrelation that is not exploitable. At intervals of several seconds to minutes, bursts of positive autocorrelation sometimes appear, especially after news or during active sessions.
Computing autocorrelation across rolling windows reveals when momentum conditions prevail. A scalper might apply momentum tactics only when recent autocorrelation is positive and above a threshold, and avoid them otherwise. The statistic is noisy, so smoothing and testing for stability are important. Different sampling intervals can yield very different values, so the choice should match the strategy's holding period. As always, the real test is whether a filter improves net results out of sample.
Whatever the result, recording the sampling interval and period used for each estimate prevents confusing conclusions later, when the same measure is recalculated under different settings.
Over tick or second-level intervals, bid-ask bounce creates negative autocorrelation that is not exploitable. At intervals of several seconds to minutes, bursts of positive autocorrelation sometimes appear, especially after news or during active sessions.
Computing autocorrelation across rolling windows reveals when momentum conditions prevail. A scalper might apply momentum tactics only when recent autocorrelation is positive and above a threshold, and avoid them otherwise. The statistic is noisy, so smoothing and testing for stability are important. Different sampling intervals can yield very different values, so the choice should match the strategy's holding period. As always, the real test is whether a filter improves net results out of sample.
Whatever the result, recording the sampling interval and period used for each estimate prevents confusing conclusions later, when the same measure is recalculated under different settings.