Parabolic SAR: step, maximum, whipsaws and a stop-and-reverse NQ test

Contents

1. How does Parabolic SAR work, and is stop-and-reverse a sound idea?
2. How should Parabolic SAR be set? The formula, step and maximum explained
3. Why does Parabolic SAR whipsaw in a range?
4. What is the difference between SAR and an ATR trailing stop?
5. Is Parabolic SAR profitable on futures? What one NQ daily-bar test shows
6. How do you test a SAR rule before trusting it?
7. Frequently asked questions
8. References

How does Parabolic SAR work, and is stop-and-reverse a sound idea?

Parabolic SAR is a trailing stop that speeds up toward price as a trend extends, and as a stop-and-reverse system it is sound only when the market trends most of the time. Stop-and-reverse closes the trade and opens the opposite one when price touches the stop, so the system never sits flat and every range costs it small losses.

SAR stands for "stop and reverse". The indicator is drawn as a dot on each bar: below price while you are long, above price while you are short. In an uptrend the dot rises every bar, and it rises faster each time price makes a new high. The construction and the chart reading are on the Parabolic SAR indicator page; this article is about how to set it and how to test it. Among the technical analysis tools and indicators you can test, SAR is unusual: it is not a signal you add to a strategy, it is a whole strategy. It gives the direction, the entry and the exit, and the exit of one trade is the entry of the next.

That completeness is also the problem. A normal indicator can say "no trade" and wait. A stop-and-reverse system cannot wait. It holds a long or a short on every bar, so in a market that moves sideways for three months it is wrong, or almost wrong, for three months. Wilder (1978), who introduced the method, presented it as a tool for trending markets, and the whole question of whether SAR is a sound system comes down to one number: what share of the time your market trends cleanly enough to pay for the time it does not.

How should Parabolic SAR be set? The formula, step and maximum explained

The step and the maximum decide how fast the dot closes the gap to price, and a larger step means earlier exits and more flips while a smaller step gives back more of each move but survives longer trends. There is one formula to know:

SAR(next) = SAR(now) + AF × (EP − SAR(now))

Here SAR(now) is the dot on the current bar and SAR(next) is the dot for the following bar. EP is the extreme point: the extreme point is the highest high reached since the current long trade began, or the lowest low since the current short trade began. AF is the acceleration factor: the acceleration factor is a fraction that starts at the step value when a new trade begins, grows by the step each time price sets a new extreme point, and stops growing when it reaches the maximum. So each bar the dot moves a fraction AF of the remaining gap toward the extreme point. Wilder (1978) used a step of 0.02 and a maximum of 0.20, and most charting platforms keep those as defaults.

Three details of the original method matter for testing. First, when price touches the dot, the trade reverses and the new dot starts at the extreme point of the trade that just ended, with AF reset to the step. Second, in a long trade the dot is never allowed above the low of the previous two bars, which stops it from touching price by accident in a calm stretch. Third, the dot never moves away from price: like any trailing stop it only ratchets in one direction until the flip.

Setting	Smaller value	Larger value
Step (how fast AF grows)	Dot stays far from price for longer; fewer flips; more of each move given back at the exit	Dot closes in quickly; earlier exits; more flips in choppy stretches
Maximum (where AF stops)	Dot never gets very tight; long trends are held through pullbacks	Dot can get very tight late in a trend; exits near the extreme but also on small pullbacks
What it changes	The balance between exit timing and flip count. Neither setting can turn a sideways market into a trending one.

When people search for "parabolic sar settings" they usually want a pair of numbers that avoids whipsaws. No pair does that. The settings only move the trade-off: a slower SAR flips less often but gives back more of each trend, and a faster SAR keeps more of each trend but flips more often in between. The next section explains why the second half of that trade-off is the expensive one.

Why does Parabolic SAR whipsaw in a range?

SAR whipsaws in a range because it must hold a position on every bar and the dot keeps accelerating toward price even when price is going nowhere, so a sideways market produces a chain of small losing flips, and each flip also pays commission and slippage. A whipsaw is a trade that is entered on a move and exited at a loss when the move reverses right away. In a range this happens over and over: price drifts up, the dot climbs under it, price falls back and touches the dot, the system goes short, price drifts up again and touches the new dot from above, the system goes long again.

Each of those flips loses a little in price and pays the full round-trip cost: commission on the exit, commission on the new entry, and slippage on both. A trend following system that is allowed to sit flat pays nothing while it waits. SAR pays for every wobble. AlphaPrime (Valynox) charges commission and slippage on every entry and exit inside its backtesting, with the fill on the next bar, so a SAR candidate's flip bill is already in the result you read.

This is why the claim that runs through this article matters: the range stretches, not the trend stretches, decide a SAR system's result after costs. Two traders can run the same SAR settings on two markets with similar trends and get very different results if one market spends more months going sideways. The common fix is to add a filter that keeps SAR out of ranges, for example a trend-strength reading or a range oscillator such as the one explained in the stochastic oscillator settings and signals article. A filter changes the system from always-in to sometimes-flat, which is a different strategy and needs its own test.

What is the difference between SAR and an ATR trailing stop?

SAR tightens with time and new extremes no matter how volatile the market is, while an ATR trailing stop sets its distance from current volatility and does not reverse the position, so the two behave differently in a quiet trend, a volatile trend and a range. Average true range (ATR) is the average size of a bar's full range, including any gap from the prior close, over a chosen number of bars; the ATR indicator page covers its calculation. An ATR trailing stop is an exit placed a chosen multiple of ATR away from the best price reached since entry, which moves toward price as the trade goes your way and, when hit, takes the position to flat.

Property	Parabolic SAR	ATR trailing stop
What sets the distance to price	Time in the trade and the number of new extremes, through AF	Current volatility, through the ATR multiple
Does it reverse the position	Yes: the exit is the opposite entry	No: the exit goes to flat; a separate rule must re-enter
Quiet, steady trend	Dot closes in fast and may exit on a small pullback late in the trend	Stop sits close because ATR is small; a small pullback can also hit it
Volatile trend	Dot closes in at the same pace as in a quiet trend, so wide swings touch it sooner	Stop sits farther away because ATR is large; wide swings are given more room
Range	Chain of reversing trades, each paying a round-trip cost	One exit to flat; no new trade until the entry rule fires again

Neither one is better across the board. Kaminski and Lo (2014) show that whether a stop rule adds or removes value depends on how returns behave after a loss, which is a property of the market, not of the stop. A market that trends with steady, low-volatility moves can suit SAR; a market whose trends come with large swings often needs the volatility-aware distance of an ATR stop. The only way to know which fits your market and bar size is to run both on the same data with the same costs. The ATR trailing stop rule and futures backtest article shows the ATR side of that comparison.

Is Parabolic SAR profitable on futures? What one NQ daily-bar test shows

One NQ daily-bar run of a pure SAR stop-and-reverse rule shows a drawdown of the same order as its net profit and a position held almost all of the time, and because the candidate shown was picked after the fact its figures flatter the rule. The run tested a Parabolic SAR rule as the only entry and exit, long and short, one contract, no stop, no target and no filter, on NQ daily bars from 2019-01-02 to 2025-06-30, with $20 commission and $10 slippage charged on every trade. The search tried 1,024 step and maximum combinations and kept the one with the best return against its drawdown. The box below is that one kept candidate.

Three readings stand out. First, the max drawdown of $128,220 is about three quarters of the $174,335 net profit. A trader holding one NQ contract through this period would have watched most of the gain disappear at some point before it came back. Second, time in market of 94.54% is what a stop-and-reverse rule must show: it is nearly always holding a position. Third, 53 trades over about six and a half years means roughly eight flips a year on daily bars, so the $30 of commission and slippage per trade is a small share of the $3,289.34 average trade. On intraday bars the same rule would flip many times more often and the cost share would grow with it.

These numbers do not say that Parabolic SAR earned this money or that a reader can expect anything like them. The candidate was chosen out of 1,024 for exactly the ratio that makes it look good, and Bailey, Borwein, López de Prado and Zhu (2014) show that the best of many tries carries a selection bonus that does not repeat on new data. The figures are useful for one thing: they show the shape of a SAR system on this market, a win rate near 70% from a rule that holds winners through long trends, paired with a drawdown that comes from the stretches where NQ went sideways and the rule kept flipping.

How do you test a SAR rule before trusting it?

Test a SAR rule by charging the full cost of every flip, checking that neighbouring step and maximum values behave alike, judging it unchanged on data the settings never saw, and counting the losing flips in the sideways stretches, then accept or reject by a rule you wrote before you looked. The order matters, because each step removes candidates cheaply before the next, more expensive step.

1. Add commission and slippage per flip from the first run. A SAR system pays twice at every reversal. Use the costs you would actually pay on your market and bar size, and rerun once with double slippage to see how much of the result depends on cheap fills.
2. Run the neighbouring step and maximum values. If 0.02 and 0.20 pass but 0.015, 0.025, 0.15 and 0.25 all fail, the setting is a lucky spike, not a plateau. A sound rule has a broad region of settings that behave alike.
3. Split the history by date and judge the later part unchanged. Choose the settings on the earlier part and read the result on the later part without touching anything. The in-sample vs out-of-sample split and how much is enough article gives the proportions and the trade counts to aim for.
4. Count the flips and losses in the sideways stretches. Mark the months where the market stayed inside a band and add up what SAR lost there, costs included. Compare that bill with what the trends made. If one or two trends carry the whole result, the system depends on their return.
5. Accept or reject by a rule written in advance. Write down the drawdown you will accept, the out-of-sample share of the in-sample result you require and the plateau width you expect, then apply the rule. If the candidate fails, change it, treat the change as a new candidate, and start again at step one.

This is the same discipline used to test any packaged signal, whether the WaveTrend indicator settings and cross signals or the LuxAlgo indicator signals and their repainting: the indicator supplies the rule, the test supplies the judgment, and the two are kept apart. A PSAR indicator on a chart will always look convincing in the trends, because that is where the dots are drawn far from price and the eye follows them. The ranges are where you should look.

A concrete next step: open your own NQ or TAIEX futures daily history, find the longest sideways stretch of the last three years, and count how many times the default SAR would have flipped inside it. Multiply that count by your round-trip cost. That one number tells you more about whether a parabolic sar strategy suits your market than any trend it caught.

Frequently asked questions

What are the best Parabolic SAR settings?

There is no pair of settings that works on every market. The common defaults are a step of 0.02 and a maximum of 0.20. A larger step exits earlier and flips more often; a smaller step gives back more of each trend but survives pullbacks. Test a range of values around the one you choose and prefer a setting whose neighbours behave alike.

Does Parabolic SAR repaint?

No. The dot for a bar is calculated from the previous bar's dot, the extreme point and the acceleration factor, all of which are known before the bar opens. Once a bar closes, its dot does not change. What can mislead is reading the dot on the current open bar as a signal before that bar has finished.

Can I use Parabolic SAR as a stop without reversing?

Yes. Many traders use the SAR dot as a trailing exit for a long trade and ignore the reverse signal. The system then goes flat at the exit and waits for a separate entry rule. This removes the chain of losing flips in a range, but it is a different strategy and needs its own backtest with its own entry rule.

Does Parabolic SAR work on intraday charts?

The formula works on any bar size, but the number of flips grows as the bars get shorter, and so does the share of each trade eaten by commission and slippage. A rule that flips eight times a year on daily bars may flip several times a day on 5-minute bars. Run the intraday version with full costs before judging it.

References

• Wilder, J. W. (1978). New Concepts in Technical Trading Systems. Trend Research.
• Bailey, D. H., Borwein, J., López de Prado, M., & Zhu, Q. J. (2014). Pseudo-Mathematics and Financial Charlatanism: The Effects of Backtest Overfitting on Out-of-Sample Performance. Notices of the American Mathematical Society, 61(5), 458–471.
• Kaminski, K. M., & Lo, A. W. (2014). When Do Stop-Loss Rules Stop Losses? Journal of Financial Markets, 18, 234–254.