Indicators · Trend and moving averages
EMA vs. SMA: which moving average to use, tested on SPY
How exponential and simple moving averages weight prices, what two years of SPY show about their speed and signals, and when each one fits.
On this page
A simple moving average (SMA) adds up the last n closing prices and divides by n: every day in the window counts equally, and a day drops out completely when it is n days old. An exponential moving average (EMA) gives the newest close the most weight and lets older closes fade away gradually, so it never drops anything all at once. Both smooth out daily noise to show the trend; they differ in how they treat recent prices.
The usual summary, "the EMA is faster", is only partly true. Two years of daily data on SPY, the largest S&P 500 fund, show where the difference is real and where it vanishes. In the spring 2025 sell-off, the 50-day EMA stopped falling eight trading days before the 50-day SMA did. Yet the price climbed back above both on the same day, and over the two years the SMA's buy signals worked slightly more often than the EMA's. This page compares the two on that data and ends with a plain rule for choosing.
The two averages side by side
| Simple moving average (SMA) | Exponential moving average (EMA) | |
|---|---|---|
| Weighting | Equal on each of the last n closes | Highest on the newest, decaying geometrically |
| Weight on today’s close (n = 50) | 2.0% | 3.9% |
| Weight on the last 10 closes (n = 50) | 20% | 33% |
| What happens to old data | Drops out on day n + 1 | Fades, never fully gone |
| Average age of the data (n = 50) | 24.5 days | 24.5 days |
| Reaction to a sudden turn | Slower | Faster |
| "Drop-off" jumps | Yes, when a big day leaves the window | No |
| Typical uses | 50- and 200-day trend levels, golden and death crosses | Short-term trend, MACD, ribbons |
| Full lesson | Simple moving average (SMA): formula, crossovers and limits | Exponential moving average (EMA): formula and how to use it |
The EMA's recursion means today's EMA holds a little of every past close: α of today's, α × (1 − α) of yesterday's, α × (1 − α)² of the day before, and so on. Finabase and most charting platforms follow TA-Lib and start the EMA from a simple average of the first n closes.
Same average age, different shape
A useful fact that is rarely stated: for the same n, the two averages have the same average age of data, (n − 1) ÷ 2 days. For n = 50 that is 24.5 days for both. So on a steady trend, both trail the price by about the same amount. What differs is the shape of the weights: the EMA leans harder on the last few days and spreads a thin tail far into the past, while the SMA is flat and then stops. That is why the EMA turns sooner after a sharp reversal but does not hug a steady trend more closely.
Two years of SPY, both 50-day averages
So the EMA turned up eight sessions earlier, which matters to anyone using the average's slope as a trend filter. But for the simplest signal, "close crosses above the average", the two gave the same answer on the same day, because the price had risen so far so fast that it cleared both at once.
Over the whole two years:
| Close crossing above the average | SMA (50) | EMA (50) |
|---|---|---|
| Crosses up / crosses down | 17 / 16 | 18 / 17 |
| Up-crosses with 20 sessions of data after them | 15 | 16 |
| Of those, higher 20 sessions later | 10 (67%) | 9 (56%) |
| Days the close was above the average | 340 of 452 | 351 of 452 |
The EMA produced one more round trip and slightly worse follow-through. On this market and period the extra sensitivity bought nothing. That is one stock-market fund over two years, mostly rising, and about 15 signals each: far too few to say one average is better in general. It does show that "faster" is not the same as "better".
Where each one goes wrong
SMA drop-off. When a very large move leaves the SMA's window, the average can jump on a quiet day for no current reason. A 50-day SMA "remembers" the April 2025 crash at full weight until mid-June, then forgets it at once.
EMA whipsaws. Because recent closes weigh more, a sideways market crosses the EMA more often. Each extra cross is a signal to act on and, usually, a cost.
Both lag. Neither average predicts. Both confirm a move after it has started, and in a choppy, trendless market both produce a string of false crossings. The Moving averages and trend: filters, the 50/200 and golden crosses lesson covers using them as trend filters rather than triggers.
Which to use when
| Situation | Better choice | Why |
|---|---|---|
| Long-term trend and widely watched levels | SMA (50, 200) | The levels most traders and the press quote |
| Golden cross and death cross | SMA | They are defined on the 50- and 200-day SMAs |
| Short-term trading, fast trend changes | EMA (for example 8, 13, 20, 21) | Turns sooner after sharp reversals |
| Building other indicators | EMA | MACD, TRIX and the Elder tools are built on EMAs |
| Avoiding drop-off jumps | EMA | No sudden exit of old data |
| Fewer false signals in choppy markets | SMA (or a longer period of either) | Less reactive to the latest bar |
In practice the period matters more than the type. A 20-day EMA and a 20-day SMA are much closer to each other than a 20-day and a 50-day of the same kind. Pick the period for your holding time first, then the type.
How Finabase does it
Both averages are studies on every Finabase chart, computed with TA-Lib on closing prices; a new EMA defaults to 10 periods and either can be set to any period. The link above opens a two-year SPY chart with the 50-day SMA and 50-day EMA drawn together, so the spring 2025 turn can be checked bar by bar. The technical gauge on each symbol page tests the close against a fixed set of SMAs and EMAs (from 5 to 200 periods) and counts each as bullish or bearish, and the Scanner lists stocks whose close has just crossed one of them. The numbers on this page came from the Learn indicator code, which is held to the chart's TA-Lib output by an automated test.
Frequently asked questions
Is EMA better than SMA?
Neither is better in general. The EMA reacts sooner to sharp turns and produces more signals; the SMA is steadier and defines the standard 50- and 200-day levels. On SPY over two years, the 50-day SMA's buy signals worked slightly more often.
Why do traders use the 200-day SMA rather than the 200-day EMA?
Convention. The 200-day SMA is the level the press, many funds and most trend studies quote, so it is the one other traders watch. Over 200 days the two are close anyway.
Does the EMA have less lag than the SMA?
Only after a sudden change. On a steady trend both have the same average lag, (n − 1) ÷ 2 days. The EMA turns sooner after a sharp reversal because it weights the newest closes more.
Which moving average is best for day trading?
Short EMAs, such as the 9- or 20-period on intraday bars, are the common choice because they follow fast price changes. Any average still lags and whipsaws in a choppy session.
Summary
The SMA weights the last n closes equally; the EMA weights the newest most and fades the rest. For the same period both have the same average data age, so they lag equally on a steady trend, but the EMA turns sooner after a sharp reversal and crosses the price more often. On two years of SPY data the 50-day EMA bottomed eight sessions before the 50-day SMA after the April 2025 low, yet both were regained on the same day, and the SMA's signals held up slightly better. Choose the period for your time frame first, use the SMA for the standard long-term levels, and use the EMA where speed matters and for indicators built on it.
Check your understanding
4 questions on this lesson. Each answer comes with the reason, and nothing is scored.
What share of a 50-day EMA’s weight sits on the newest close?
- 2.0%, the same as a 50-day SMA
- About 3.9%, because α = 2 ÷ (50 + 1)
- 50%
Show the answer
B. About 3.9%, because α = 2 ÷ (50 + 1) The EMA’s smoothing factor is 2 ÷ (n + 1) = 0.0392 for n = 50. A 50-day SMA gives each close 1 ÷ 50 = 2%.
On a steady, straight-line uptrend, how do a 50-day EMA and a 50-day SMA compare in lag?
- The EMA has about half the lag
- The SMA has no lag
- They lag by about the same amount, because both have an average data age of 24.5 days
Show the answer
C. They lag by about the same amount, because both have an average data age of 24.5 days For the same n, both averages have an average data age of (n − 1) ÷ 2. The EMA’s advantage appears only after a sharp change, when its heavier recent weights turn it sooner.
After SPY’s April 8, 2025 low, the 50-day EMA bottomed 15 sessions later and the 50-day SMA 23 sessions later. When did the close move back above each?
- Above the EMA eight sessions before the SMA
- Above both on the same day, May 1, 2025
- Above the SMA first
Show the answer
B. Above both on the same day, May 1, 2025 The price rebounded so fast that it cleared both averages on the same close. The EMA’s earlier turn showed up in its slope, not in the crossover signal.
You want to follow the level most investors and news reports call "the long-term trend". Which average fits?
- The 200-day SMA
- A 9-day EMA
- A 20-day EMA on hourly bars
Show the answer
A. The 200-day SMA The 200-day SMA is the convention for the long-term trend and the golden and death crosses, so it is the level others watch.
Sources and further reading
- TA-Lib — SMA and EMA function definitions (EMA seeded with a simple average)
- Finabase chart data — SPY daily bars, 2024-10-30 to 2026-10-29; averages computed with the Learn indicator code, which test/learn-labs.test.ts holds to the chart’s TA-Lib output
