EMA vs Parabolic SAR
Michael Whitman
·
Trading since 2017, full-time since 2020
·
Published 2026-09-08
·
Updated
·
Reviewed
The exponential moving average smooths closes into a line and takes no position. Wilder's parabolic stop and reverse is a stop that moves toward price faster as a trend extends and flips sides when touched, so as a system it is always long or short. One is a reference line; the other is a trailing exit.
They get filed together as trend tools. One is a smoothed line that takes no view. The other is a trailing stop with a built-in reversal, published by Welles Wilder in 1978, which is always holding one side or the other. On SPY their behavior differs in a way that follows directly from that design.
What each one is
The exponential moving average smooths closes into a line, weighting each new close by 2/(N+1). Price sits above or below it; the line itself holds no position. The exponential moving average page has the formula.
Wilder’s parabolic stop and reverse (SAR) is a stop that accelerates. Each session it moves a fraction of the way toward the most extreme price of the current side. That fraction, the acceleration factor, starts at 0.02 and rises by 0.02 with each new extreme, up to 0.20, the StockCharts defaults. When price touches it, it jumps to the other side. Parabolic SAR covers it.
So one describes and the other acts. The line says where price is relative to its recent average; the SAR says where a stop would be and which side it would leave you on.
Where they differ
Whether time matters. The exponential average does not care how long a trend has run. The SAR tightens with every new extreme, so an older move is held on a shorter leash.
Where each line sits. On SPY the SAR was a median 2.385% from the close and the 50-day EMA 2.262%, similar distances measured very differently: one by acceleration, one by averaging.
How each changes side. Price crossing an average can reverse the next day. A SAR flip resets the stop to the far extreme of the old side, so an immediate flip back takes a larger move.
How often each reverses. Price crossed the 50-day exponential average 19.55 times a year on SPY, and most of those crosses were undone within a week; SAR flips were undone that fast far less often.
Both are built from the same bars, whereas only the SAR uses highs and lows rather than closes alone. Neither is a forecast: both move only after a bar closes, while an order can rest at either level in advance.
Which one to use: EMA or Parabolic SAR?
Use the exponential average to define the trend you will trade, for example only taking longs while price is above it.
Use the SAR as the exit for a position you already hold, because it tightens as the move ages, which is what a trailing stop is for.
Use neither as an always-in system when the market is ranging. The SAR still has to be long or short, and a range produces reversals at both edges.
And when the two disagree, the average says trend and the SAR says time. On SPY they pointed the same way on 72.44% of sessions.
How the stop behaved on SPY
The test. SPY daily highs, lows and closes from 3 January 1994 to 2 October 2026. The SAR uses Wilder’s rules with the default 0.02 step and 0.20 cap; the comparison line is the 50-day exponential average of closes. A flip is a session on which the SAR changes side; a cross is a session closing on the opposite side of the average to the session before.
The SAR changed side between 19 times a year (1994, 2002 and 2023) and 30 (2014). The Ichimoku vs parabolic SAR page gives the full count with the same settings. It was long on 60.17% of sessions, the median side lasted 9 sessions, and the longest lasted 39, from 6 January to 3 March 2017.
Price crossed the average 640 times, 19.55 a year, but the yearly count ran from 1 to 45. 382 of 639 crosses (59.78%) were followed by another within 5 sessions, against 247 of the 800 gaps between SAR flips (30.88%).
Why the stop tightens with age
The acceleration factor only rises. Each new high on a long side adds 0.02 to the share of the gap the SAR closes every session, so the stop’s pace picks up just as a move is extending.
On SPY the gap in average true range shrank after the first week. On long sides that lasted at least that long, the close sat a median 2.388 average true ranges above the SAR on the first session, 2.525 on the fifth, 2.156 on the tenth and 1.553 on the twentieth (73 sides reached day 20).
A one-day side. On 16 July 2025 SPY’s low of $618.05 met the SAR, and it flipped short to $627.86, the highest price of the long side just ended. The next day the high reached $628.40, above that stop, and it flipped long again at $618.05. Through both sessions the close stayed above the 50-day exponential average, near $601, so the average never changed its reading.
The original data
The sample. SPY’s daily high, low and close from Yahoo Finance as traded, first bar 29 January 1993, with every count taken from 1994 onward to 2 October 2026. SAR start: long if the second close is above the first. Each session the SAR may not sit above the prior two lows on a long side (below the prior two highs on a short side). Average true range is Wilder’s 14-session version.
The headline figure. 59.78% of average crosses and 30.88% of SAR flips were reversed again within 5 sessions.
The file. Every session’s SAR value and side, each flip, the 50-day exponential average and each close across it are in the SPY SAR and EMA file.
What is left out. No flip or cross is scored for what price did next, and no trading rule is tested. Different step and cap settings give different counts; these are the defaults only.
When it fails
The failure is the SAR used as an entry system in a range. Every flip opens a new position, so a sideways month produces a long at the top of the range and a short at the bottom, each stopped out when price returns to the other edge. The SAR behaved as built; the build assumes a trend exists. Filtering entries with a trend line, such as the exponential average, and keeping the SAR for exits removes that assumption from the entry.
The second failure is reading a cross of the average as a reversal. On SPY most were undone within a week.
A third is widening the SAR settings after a whipsaw. It then lags on the next trend.
A fourth is holding through a flip because the average still agrees. The SAR is the stop.
And a fifth is comparing the two on one chosen year. Yearly counts swing widely.
Related
The exponential moving average covers the reference line. Parabolic SAR covers the stop and reverse. And ATR trailing stop covers a trailing exit sized from measured range.
Use the parabolic stop and reverse to decide when to get out of a position you already chose, not when to get in. Its reversals come about every two weeks on SPY, so as an entry rule it puts you into a new position about that often, whether or not anything is trending.
— Michael Whitman
This page is educational, not financial advice. Test every idea on your own charts before risking money.