Parabolic SAR is a trend-following stop rule: it trails price and flips sides when price reaches the stop. That makes it easy to turn into a mechanical system, but not a forecast—and in sideways price action, repeated flips can turn a tidy chart into a costly sequence.
What Parabolic SAR measures
Parabolic SAR (Stop and Reverse) is a price-and-time indicator that places a trailing stop below rising prices or above falling prices. When price crosses the SAR, the indicator reverses direction and begins calculating on the other side. StockCharts’ calculation guide and TradingView’s algorithm notes describe the same core mechanism.
“Stop and reverse” names the rule, not a promise that a real order would fill at the plotted value. A chart indicator does not model spread, gaps, slippage, or whether the next executable price is still near its stop. Treat the plotted line as a signal definition until a test models those details.
The moving stop has a memory
The calculation uses three pieces: the previous SAR, the trend’s extreme point (EP), and an acceleration factor (AF). In an uptrend, EP is the highest high reached so far; in a downtrend, it is the lowest low. The next SAR moves from its prior value toward EP by a fraction set by AF. Both StockCharts and TradingView document this recurrence and the guard that keeps SAR outside the previous bars’ range.
Common default settings start AF at 0.02, add 0.02 when the trend makes a new extreme, and cap it at 0.20. As AF grows, SAR catches up faster. A faster stop can respond sooner when a trend turns, but it also leaves less room for ordinary movement before a reversal. Those are implementation defaults, not universal constants; check the platform’s settings before comparing charts.
Uptrend: SAR below price → new high raises EP and AF → SAR moves up
Downtrend: SAR above price → new low lowers EP and AF → SAR moves down
Price crosses SAR → direction flips; the rule resetsThe two-bar constraint matters: in an uptrend, SAR is not allowed above the prior two lows; in a downtrend it is not allowed below the prior two highs. On a reversal, the next calculation also resets around the prior trend’s extreme point. Small differences in initialization or bar data can therefore change early signals; TradingView and StockCharts each describe the reversal and range constraints in their calculation notes.
Where the method gets awkward
Parabolic SAR is designed to trail a directional move. StockCharts says it works best with trending securities and warns that settings that make it too sensitive can produce whipsaws; its guide also shows that changing the step changes the number of reversals. TradingView’s description likewise treats AF as the sensitivity control. The practical trade-off is not “fast versus accurate”: it is earlier stop movement versus more exposure to ordinary back-and-forth.
That caveat is about the indicator’s mechanics, not proof of a profitable strategy. A review by Park and Irwin of modern technical-analysis studies found positive, negative, and mixed results, while flagging data snooping, post-hoc rule selection, risk estimation, and transaction-cost treatment as limitations (University of Illinois record). In a separate study of technical rules on the Dow Jones Industrial Average, Bajgrowicz and Scaillet report that the apparent in-sample performance they examined was offset by transaction costs and that future winning rules could not be selected in advance (authors’ abstract). Neither study tests every Parabolic SAR implementation or settles its value across markets. They do explain why a chart pattern or attractive historical parameter is not enough evidence.
How you would test a SAR rule
A useful test begins by writing the rule before looking at results. Fix the instrument, bar interval, SAR initialization, AF step and cap, reversal fill convention, and whether the system holds long, short, or cash. Otherwise two implementations bearing the same indicator name may not be the same strategy.
Then compare the SAR system with a simple control over the same dates and data. Include executable spread, commission, slippage, and gaps; report both the count and timing of reversals, net results, drawdown, and performance by trending and choppy periods. Reserve unseen data for evaluation, and report every parameter set tried rather than only the best one. The research cautions above make search discipline and costs part of the method, not footnotes.
For a practical companion, see how parameter sensitivity can expose a fragile backtest. The realbacktesting methodology page explains how its published cBot tests are made reproducible; that is a standard for evidence, not evidence that Parabolic SAR itself works.
Frequently asked
Is Parabolic SAR an entry signal?
It can be used to define a reversal rule, but a plotted flip alone does not establish an edge or guarantee an executable fill. The full entry, exit, and cost rules need testing.
Does Parabolic SAR work in a ranging market?
Its trailing design can generate repeated reversals when price moves back and forth. Whether a filter improves results is an empirical question; test the filter as a separate rule, including its costs and missed moves.
Are the default SAR settings optimal?
No setting is established as universally optimal by the indicator’s formula. Defaults are conventions; compare a predeclared range of settings on unseen data and account for the number of trials.
The stubborn takeaway: a stop that moves by formula is still only a hypothesis until its fills and failed reversals survive a fair test.