Moving Averages: How SMA and EMA Differ in Responsiveness
A simple moving average treats every bar equally. An exponential moving average weights recent bars more. See how that math changes signal timing and lag.
By Imperial Analytics
A moving average is one of the few indicators a trader will see on almost every chart, and one of the few where the choice between the two common forms changes the signal in a measurable way. A simple moving average and an exponential moving average of the same length do not return the same number, do not turn at the same time, and do not produce the same crossover dates. This post defines both, walks through the arithmetic that makes them differ, and names where the trade-off between responsiveness and stability matters for a futures trader.
By Imperial Analytics
What a moving average actually is
A moving average is a running summary of the most recent N closing prices, expressed as a single number that moves with each new bar. Its purpose is to smooth the price series so the trader sees the trend underneath the bar-to-bar noise. The choice of N controls how much smoothing happens. A short N follows price closely and changes direction often. A long N follows price loosely and changes direction rarely.
The point of summarizing the recent closes is that the raw closing-price series carries two kinds of information at the same time. One is the direction the instrument has been moving over a meaningful window. The other is the random tick-to-tick noise that lives on top of that direction. A trader looking at the raw closes sees both at once and has no clean way to separate the signal from the noise by eye.
A moving average smooths the noise by averaging it against the signal. If the underlying direction is slowly upward and the bar-to-bar noise averages to zero around the direction, the moving average tracks the direction and the noise cancels. The trader sees a line that updates at the cadence of the chart but moves at the cadence of the trend.
The cost of that smoothing is lag. The moving average's value at the current bar is a function of bars that have already happened, so by construction it responds to a change in the underlying direction one or more bars after the price has begun to turn. Shorter windows lag less and react more. Longer windows lag more and react less. The trader's choice of N is therefore a choice between two failure modes: too short and the line whips around on noise, too long and the line confirms the turn after the move has already happened.
How a simple moving average is computed
A simple moving average over N bars is the arithmetic mean of the last N closing prices. Every bar inside the window contributes equally to the result. When a new bar prints, the oldest bar in the window drops out and the new bar joins, so the average shifts forward by exactly one bar. The math is elementary; the assumption underneath it is that every bar in the window is equally informative about the present.
The formula is the textbook one. For an N-bar simple moving average on a price series, the value at bar t is the sum of the closes from bar t minus N plus one through bar t, divided by N. A twenty-bar simple moving average updates every bar by dropping the close from twenty bars ago and adding the new close, then dividing by twenty.
The equal weighting is the structural feature that matters. A close from twenty bars ago has exactly the same effect on the current twenty-bar simple moving average as the close from this bar does. That property makes the simple moving average stable, because no single bar can dominate the result, and it also makes the simple moving average slow to respond to a directional change, because the new bar's information is diluted across the full window.
The simple moving average also has a known property at the moment an old, extreme bar drops out of the window. If the bar leaving the window had an unusually high close, the moving average can shift noticeably downward on the next bar even when the new close is not unusual. That shift is sometimes called the drop-off effect, and it is a feature of equal weighting rather than a bug. A trader who reads a turn from the simple moving average without checking whether the drop-off bar moved it should treat that turn with caution.
How an exponential moving average is computed
An exponential moving average weights recent bars more than older bars and never fully forgets the older bars. The current value is a weighted combination of the current close and the prior exponential moving average value, with the weight on the current close governed by a smoothing factor that depends on the length parameter. A twenty-bar exponential moving average reacts faster to a new close than a twenty-bar simple moving average and decays the influence of old bars geometrically rather than dropping them all at once.
The formula is the recursive one. The exponential moving average at bar t equals a smoothing constant times the close at bar t, plus one minus that constant times the exponential moving average at bar t minus one. The smoothing constant for an N-bar exponential moving average is most commonly two divided by N plus one. For N equal to twenty, the smoothing constant is roughly 0.0952. That means the current close contributes about ten percent to the new value and the prior exponential moving average contributes about ninety percent.
The structural feature here is geometric decay. Yesterday's close still influences the exponential moving average today, but at a smaller weight than today's close. The close from twenty bars ago influences it at a much smaller weight, and the close from a hundred bars ago at a tiny but nonzero weight. The series of weights running backward in time looks like a geometric series, which is why the indicator carries the word exponential.
The practical effect is that an exponential moving average turns earlier than a simple moving average of the same length when price actually changes direction. It is also more sensitive to a single sharp bar, which means a single outlier can push the exponential moving average around more than it would the simple moving average. The trader gets faster response in exchange for higher sensitivity to individual bars.
Data note
Numerical examples in this post use round smoothing constants and simplified closes for clarity. Real charts will produce values to several decimal places and may use slightly different smoothing-constant conventions across charting platforms. The trader should rely on the actual values produced by their own platform with the platform's documented formula, not on the rounded numbers used here.
Where the two moving averages differ in practice
The two moving averages of the same length differ on three observable behaviors. They differ on the bar at which they first reflect a directional change. They differ on how an outlier bar moves the line. And they differ on how a trader who builds entries off a moving-average crossover sees the entry timing. The choice between them is therefore not a stylistic one. It changes the dates on which entries fire.
The first behavior is the response lag. A twenty-bar simple moving average and a twenty-bar exponential moving average given the same close series will produce different values, and the exponential will turn earlier when the underlying direction shifts. A trader using a moving average as a trend filter, where price above the line means uptrend and below means downtrend, will switch sides on the exponential before the simple. That timing difference is sometimes the difference between catching the second half of a move and entering on confirmation that lasts long enough to count as confirmation.
The second behavior is sensitivity to outliers. A single sharp bar moves the exponential more than it moves the simple, because the recent weighting puts more of that bar's influence on the current value. The simple moving average dilutes the same sharp bar across N equal weights. A trader running a fast exponential moving average on a noisy instrument will see the line whip more than a trader running a simple moving average of the same length on the same instrument.
The third behavior is the crossover date when two moving averages of different lengths are used together. A common configuration runs a short moving average and a long moving average and treats the bar on which they cross as a signal. The crossover bar for two simple moving averages will not be the same as the crossover bar for two exponential moving averages of the same lengths. A trader who switches from simple to exponential without changing the length parameters will see different entry dates on the same chart on the same data.
The fourth behavior, related to the third, is how often the two cross under each other in the absence of a real trend. The exponential pair will produce more crossovers in a chop regime than the simple pair, because they each track price more closely. A trader who relies on crossover signals should expect the exponential configuration to produce more false signals in chop and the simple configuration to produce slower signals in trend. Neither is universally better.
↳ Note
A simple moving average treats every bar in the window the same. An exponential moving average never fully forgets an old bar but never fully trusts it either. Both are honest. Neither is universally faster.
What this means for a trader's chart and journal
The practical implication is that a trader should pick the moving-average form that matches the question they are asking the line, and tag their journal with which form was used on which entry. If the question is "what is the recent trend, smoothed against bar-to-bar noise," a simple moving average is the cleaner read. If the question is "has the trend turned in the last several bars," an exponential moving average updates earlier. The journal carries the tag so the same setup is testable across forms.
The trader who runs a moving-average-based setup for any length of time should record, on every trade, which form of moving average produced the signal, the length parameter used, and the value of the line at the bar of entry. With those three columns in the trade record, the trader can later ask the journal whether the simple or the exponential produced better outcomes on the same setup over a sample large enough to clear the strategy-claim sample-size minimum. That comparison is the only honest way to decide which form to keep.
A common mistake is to switch between simple and exponential mid-strategy on the same chart without tagging the change. The journal then mixes outcomes from two distinct signal regimes and produces a blended distribution that does not represent either one. The cure is the same as the cure for almost every behavioral-edge problem: tag the input, store the value, and let the data decide the question after the sample is big enough to answer it.
The other practical note is that moving averages are inputs to decisions, not decisions themselves. A clean trend filter does not, by itself, produce edge. It removes a class of low-quality entries from the strategy's universe. The edge still has to come from the rest of the setup. A trader who treats the moving-average configuration as the strategy will eventually rediscover that the form of moving average mattered less than the discipline applied to the entries it permitted.
Frequently asked questions
Frequently asked questions
- q: Which is faster to react, SMA or EMA? a: An exponential moving average of the same length reacts faster to a recent change in closing price because it weights the most recent close more heavily. A simple moving average treats every bar in the window equally, so a recent change is diluted across the full window.
- q: Why do some traders prefer SMA despite the lag? a: A simple moving average is more stable because it does not over-respond to any single bar. Traders who want a clean read of the underlying trend and who tolerate later signals often prefer the simple form. The choice depends on the question the trader is asking the line and on the noise profile of the instrument.
- q: Are crossover signals different between SMA and EMA pairs? a: Yes. Two simple moving averages of given lengths will produce different crossover bars than two exponential moving averages of the same lengths on the same price series. A trader who switches between forms will see different entry dates on the same chart, even with the same length parameters.
- q: What length should a trader use? a: There is no universal length. The length controls how much smoothing the line applies; a shorter length is more responsive and noisier, a longer length is smoother and laggier. The honest approach is to pick a length that matches the time horizon of the setup, record it in the journal, and test it on a sample that clears the sample-size minimum before drawing conclusions.