EMA: The Exponential Moving Average Explained

Key Takeaways

• EMA exponential moving average uses a smoothing multiplier so each new price overwrites the prior average by a fixed fraction.
• A 10-period EMA applies about 18 percent weight to the most recent price; a 20-period EMA applies about 10 percent.
• The most common mistake is comparing a 10-day SMA and a 10-day EMA as if they smooth the same amount; they do not.
• EMAs power most popular indicators: MACD, TRIX, KAMA, DEMA, TEMA, T3, and many adaptive systems.

What It Is

An exponential moving average is a recursive average. Each new value is a weighted blend of today's price and yesterday's EMA. The weight on today's price is fixed by the chosen period, and the weight on the prior EMA absorbs everything before it.

Older prices never fully drop out, but their influence decays geometrically. That is the "exponential" part of the name. The further back a price sits, the smaller its effective weight on today's line.

The Intuition

A simple moving average treats a price from N days ago and yesterday's price as equally informative. That is fine for slow horizons but feels wrong on fast ones. If a stock just broke out, yesterday matters more than a quiet print three weeks ago.

The EMA encodes that intuition mechanically. The smoothing multiplier k controls how aggressively the line bends to new prices. A bigger k (shorter period) leans on recent data and reacts quickly. A smaller k (longer period) damps the response and produces a smoother line.

How It Works

The standard recursive form, popularized by StockCharts and Wilder-style smoothing, is:

EMA_today = (Price_today * k) + (EMA_yesterday * (1 - k))

where k = 2 / (N + 1)

For a 10-period EMA, k equals 2 / 11, about 0.1818. So today's price gets 18.18 percent weight and the prior EMA gets 81.82 percent. StockCharts notes that a 20-period EMA applies a 9.52 percent weighting to the most recent price.

The first EMA value is usually seeded with the SMA of the first N prices, then the recursive formula takes over. Some platforms seed with the first close instead, which produces a small startup difference that washes out within roughly 3N bars.

A useful rule of thumb: an EMA of length N has the same lag profile as an SMA of length (N + 1) / 2 in steady-state. A 20-period EMA reacts roughly like a 10 or 11 period SMA, not like a 20-period SMA. This is why direct comparisons of "the 10 SMA versus the 10 EMA" can be misleading.

Worked Example

Start with five daily closes: 100, 102, 104, 103, 110. Seed the EMA with the 5-period SMA, which is 103.8. Use k = 2 / (5 + 1) = 0.333.

Imagine the next bar closes at 112. Apply the formula:

EMA = (112 * 0.333) + (103.8 * 0.667)
    = 37.30 + 69.23
    = 106.53

Now the bar after closes at 114:

EMA = (114 * 0.333) + (106.53 * 0.667)
    = 37.96 + 71.06
    = 109.02

The EMA climbed from 103.8 to 109.02 across two bars. Compare to the SMA from the SMA article on the same kind of move and you would see the EMA reach a similar value sooner. That is the responsiveness gain. The cost is more sensitivity to single-bar spikes.

Common Mistakes

1. Comparing same-length SMA and EMA directly. A 10-day EMA reacts roughly like a 5 or 6 day SMA. If you want similar smoothing, use roughly 2N - 1 EMA length or (N + 1) / 2 SMA length.

2. Forgetting the seed sensitivity. Different platforms seed the first EMA differently (SMA of N, first close, or zero). Early in a backtest the values diverge. Always run the warmup period before evaluating signals.

3. Treating EMA crosses as automatic signals. Like SMA crosses, EMA crosses whipsaw in sideways markets. The faster reaction helps in trends and hurts in chop. Pair with a volatility or trend filter.

4. Stacking too many EMAs without purpose. Ribbons of five or six EMAs look impressive on a chart and add little new information. Two well-chosen lengths usually beat six redundant ones.

5. Using EMA on a non-stationary input. EMAs of cumulative volume, raw open interest, or other unbounded series can mislead. Smooth percent changes or normalized values instead.

Frequently Asked Questions

What is EMA exponential moving average in simple terms? The EMA exponential moving average is a price average that gives more weight to recent bars. Older prices fade out gradually rather than dropping off.

How does EMA affect investment decisions? Active traders use short EMAs (9, 21) for entry timing because they react faster than equivalent SMAs. Longer EMAs (50, 200) serve as trend filters in systematic and discretionary strategies alike.

What is a real-world example of EMA use? The MACD indicator, one of the most popular momentum tools on Wall Street, is built entirely from EMAs. It subtracts a 26-period EMA from a 12-period EMA and smooths the result with a 9-period EMA.

How can investors use EMAs effectively? Pick a length that matches your horizon, then use the EMA as one input among several. Confirm crosses with volume or a longer trend, and respect that the EMA still lags price, just less than the SMA.

How is EMA different from WMA? EMA uses a recursive multiplier and decays old prices smoothly forever. WMA uses linear weights inside a fixed window and drops the oldest price entirely each bar.

Sources

1. StockCharts ChartSchool. "Moving Averages - Simple and Exponential." https://chartschool.stockcharts.com/table-of-contents/technical-indicators-and-overlays/technical-overlays/moving-averages-simple-and-exponential
2. Fidelity Learning Center. "Exponential Moving Average (EMA)." https://www.fidelity.com/learning-center/trading-investing/technical-analysis/technical-indicator-guide/ema
3. Charles Schwab. "A Moving Average: Simple vs. Exponential." https://www.schwab.com/learn/story/simple-vs-exponential-moving-averages
4. Britannica Money. "Simple vs. Exponential Moving Averages." https://www.britannica.com/money/simple-vs-exponential-moving-averages

Disclaimer

This article is educational content only and is not financial advice. Nothing here is a recommendation to buy, sell, or hold any security. Consult a licensed advisor before making investment decisions.