How Much More Does an Exponential Moving Average Respond to Recent Prices?
Understand exponential moving average weights and recursive calculations, and compare EMA with SMA. Check differences caused by initialization, historical data length, and intraday changes.
An exponential moving average gives greater weight to recent prices and carries the previous average forward in its calculation. It is more sensitive to recent changes than a simple moving average of the same period, but it can also fluctuate more in response to temporary price spikes or drops. A faster response alone does not establish better predictive power.
The period does not impose a strict memory cutoff
A typical EMA multiplies today’s price by α and yesterday’s EMA by 1−α, then adds the two. Conventionally, α is set to 2÷(period+1). For a 9-period EMA, α is 0.2. Unlike an SMA, which excludes observations beyond a fixed count, an EMA retains a progressively smaller influence from earlier values. It is therefore inaccurate to describe a 9-period EMA as “a weighted average using only the last nine prices.” Even with the same price today, the path taken by earlier prices still has some influence on the current average.
Different initialization and input histories can produce different charts
The calculation needs an initial EMA; one approach uses the SMA of the initial period. If only a short data history is loaded, the starting value can retain an influence, causing small differences between services. When comparing them, match more than just the latest close: obtain enough historical data and check the initialization method. During the session, the closing-price input may effectively track the current price, so the EMA can keep changing. A difference between the line saved after the close and the line seen intraday does not by itself establish a calculation error.
The costs of choosing a faster line
A short EMA turns down quickly in a sharp decline but also changes direction readily in a rebound. If a rule trades whenever price crosses the line, greater sensitivity translates into more trades and execution costs. First decide whether the aim is to recognize direction quickly or reduce frequent changes in judgment. Placing an SMA and an EMA on the same chart and comparing which reacted more to the latest one or two bars can clarify the difference. Include sideways and rapidly changing periods rather than selecting only periods with favorable results and declaring one indicator better. In a comparison table, select dates when the two averages with the same period diverged and explain the effects of recent prices and initialization separately to make the difference clearer.
Worked example
Hypothetical example: If yesterday’s 9-period EMA was 100 won and today’s close is 110 won, today’s EMA is 110×0.2+100×0.8=102 won. If the next day’s close returns to 100 won, the EMA becomes 100×0.2+102×0.8=101.6 won. Although price has returned to its original level, the average still reflects the previous day’s spike. When comparing charts, record both the price and EMA for both dates so this lingering effect is not missed.
What to check
- Check the relationship between the α used and the number of periods.
- Check the initial calculation value and the length of historical data.
- Distinguish intraday values from values based on the final close.
- When increasing sensitivity, compare the number of crossovers as well.
Sources and primary materials
This is educational material about research methods. For an actual company decision, verify the filings and primary sources for the relevant period.