Parabolic SAR Tested: The Stop-and-Reverse Rule Lost 1,094 Pips in 1,182 Trades

Parabolic SAR is the most requested indicator in my inbox and the easiest one to misread. It draws a dot under price in an uptrend and above price in a downtrend, and when price touches the dot the whole system flips: close the long, open a short, same bar. Wilder called it stop and reverse, and that name is the entire strategy. I ran it over 16,926 real EURUSD 5-minute bars from June 3 to August 26, 2026, with no discretion, no optimisation on the outcome and no cherry-picked window. The result is not close.

How was Parabolic SAR tested?

Same dataset and the same measurement rules as my ADX above 25 test and my moving average crossover test, so every number here is directly comparable to those.

• Indicator: classic Wilder Parabolic SAR, step 0.02, maximum 0.2 — the MT4 and MT5 default. The acceleration factor rises by the step each time a new extreme is made and is capped at the maximum.
• Signal: the bar on which the dot switches sides. Dot moves from above to below price, that is a long. The mirror is a short.
• Two exit rules, measured separately: a fixed hold in bars, and the real SAR rule of holding until the next flip.
• Measurement: the move in pips after the exit, in the signal direction, before spread. No stop, no target — the number describes the signal, not a money-management scheme layered on top of it.
• Overlap rule for the fixed-hold tests: after a signal the clock skips forward by the hold length, so one move is never counted twice.

What did the real stop-and-reverse rule pay?

This is SAR as Wilder wrote it and as every tutorial demonstrates it: you are always in the market, and each flip closes one trade and opens the opposite one.

A 28.5% win rate means roughly seven trades in ten close red. That is survivable on paper if the winners are enormous, and they are not: the best trade in nearly three months was 61.1 pips against a worst of -41.4. And this is before spread. At a typical EURUSD cost of around one pip, as measured in my spread and swap breakdown, 1,182 trades hand another 1,182 pips to the broker. The strategy as taught does not lose slowly. It loses twice.

Is the entry bad, or the exit?

Here is where the test earns its keep. I took exactly the same flip signals and, instead of holding to the next dot, held each one for a fixed number of bars.

The four-hour row is the only positive one, and it is not marginal by the standards of this series — 49.7% wins is the highest win rate any single entry rule has produced on this dataset, ahead of the Ichimoku cloud breakout‘s 44.4%. So the dot switch does contain information. It just needs four hours to express it.

Now put that next to how long SAR actually lets you stay in a trade.

How long does a Parabolic SAR trend last?

This is the whole article in one number. The signal pays at four hours. The indicator ends the trade after 65 minutes at the median, and in 73.1% of cases inside 100 minutes. Only a handful of trends in three months lasted long enough to reach the horizon where the edge lives — the single longest run in the entire sample was 5 hours 35 minutes.

Parabolic SAR is not a bad entry attached to a good exit. It is a usable entry welded to an exit that fires roughly three hours too early, and because the exit also reverses you, the early close is not neutral — it puts you on the wrong side of the move you just correctly identified. That is how a signal worth +0.65 pips becomes a system worth -0.93.

Does a trend filter fix it?

The obvious repair is to only take SAR flips when a trend is genuinely running. I filtered the two-hour test by ADX, using the same 25 threshold from my ADX study.

The filter runs backwards. A trend-following indicator did its best work when ADX said there was no trend, and its worst when ADX said there was one. That reads as a contradiction until you remember what a SAR flip is: the dot switching sides means price has just reversed hard enough to cross a trailing level. In a strong directional market that is usually a pullback inside a continuing move, so the flip fights the trend and loses. In a quiet range it is the turn of the range, and the turn is exactly where mean reversion pays. Measured honestly, SAR flips on 5-minute EURUSD behave like a reversal signal, not a trend signal — the opposite of what the indicator is sold as.

How close does the dot actually sit?

People use SAR as a trailing stop more often than as an entry, so the distance between the dot and price is a practical number.

Half the time your stop is inside 5 pips of price, and a quarter of the time it is inside 3. On a pair whose spread is around a pip, a 1.4-pip stop is not risk management, it is noise. My stop distance test found that stops under 10 pips on this timeframe were hit by ordinary intraday wobble far more often than by anything directional, and my fixed against ATR stop comparison reached the same conclusion from the other side. The SAR dot lives in exactly that zone for half the chart.

What I would actually do with this

• Do not trade stop and reverse. 28.5% wins and -1,094 pips before costs, on the rule as published.
• If you use the flip at all, use it as an entry and time your own exit. Four hours was the only horizon that paid, and SAR will never hold you that long by itself.
• Do not add a trend filter. ADX above 25 cut the result to -1.11 pips. The signal is a reversal signal wearing trend-following clothes.
• Be careful using the dot as a stop. It is inside 3 pips of price on a quarter of all bars.
• Count your trades before you count your pips. 1,182 round turns is over a thousand pips of spread in three months on one pair.

Frequently asked questions

Is the Parabolic SAR strategy profitable?

Not as published. On 16,926 EURUSD 5-minute bars the stop-and-reverse rule made 1,182 trades at a 28.5% win rate for -1,094.3 pips before spread. The flip signal on its own, held four hours instead, was mildly positive at +0.65 pips per trade — so the entry has some value and the built-in exit destroys it.

What are the best Parabolic SAR settings?

On this data, lowering the maximum acceleration from 0.2 to 0.1 while leaving the step at 0.02 moved the two-hour result from -0.04 to +0.41 pips per trade. The full settings table, including what happens when you speed the indicator up instead, is in my Parabolic SAR settings guide for MT4.

Why did SAR do better when ADX said there was no trend?

Because a dot switch is a reversal event. In a strong trend it usually marks a pullback, so taking it means trading against the dominant move; in a range it marks the turn, which is where the money is. The measured behaviour is mean-reverting even though the indicator is classified as trend-following.

Is Parabolic SAR better than Supertrend?

They point the same way 78.7% of the time and diverge on the rest, and which one wins depends entirely on the hold. At two hours SAR was less bad; at eight hours Supertrend was far less bad. The side-by-side is in my Parabolic SAR against Supertrend comparison.

Does Parabolic SAR repaint?

No. The dot for the current bar is computed from prior bars and the current high and low, and once the bar closes it never moves. That is genuinely to its credit and it is one of the reasons it survived fifty years. Not repainting and being profitable are different questions, and this test only answers the second one.

What data was this run on?

16,926 EURUSD 5-minute bars, June 3 to August 26, 2026 — the same set behind my ADX against Supertrend comparison, my Bollinger squeeze test and my Fibonacci retracement test.