Parabolic SAR: formula, settings and limits

Parabolic SAR: definition, what it plots and what it does not say

The SAR answers a question about management more than analysis: if I follow the current move, at what price do I get out? TradingView's help page on the Parabolic SAR defines it as a time and price tool primarily used to identify points of potential stops and reverses. Its calculations create a parabola located below price during a bullish trend and above price during a bearish trend.

StockCharts' ChartSchool recalls that Wilder spoke of a system, the 'Parabolic Time/Price System', in which SAR is the indicator. It files it among trend-following indicators and describes it as a trailing stop: it rises and never decreases for as long as the uptrend lasts, and falls and never rises for as long as the downtrend lasts. MetaTrader 5's help page on the Parabolic SAR compares it to a moving average that would accelerate and that may change sides relative to price.

Wilder introduced it in New Concepts in Technical Trading Systems, the book that also contains the RSI, the ATR and directional movement, StockCharts and TradingView write. The Google Books record of the book gives the publisher, Trend Research, the year, 1978, 141 pages, and a chapter titled 'The Parabolic System'. What the SAR does not say fits in one sentence from TradingView: being time and price based, it shows the direction and duration of a trend, not its strength. Pages consulted in October 2026.

Calculating the Parabolic SAR: extreme point, acceleration factor, limit

Three values to track. The SAR of the previous candle. The extreme point, EP: the highest high reached since the current rise began, or the lowest low reached since the current fall began. And the acceleration factor, AF, a multiplier that grows as the trend goes on.

The formula. In a rise, StockCharts writes it: SAR = prior SAR + prior AF × (prior EP − prior SAR). In a fall, the same distance is subtracted: SAR = prior SAR − prior AF × (prior SAR − prior EP). TradingView sums up both cases in a single line, identical to the first. On each candle, the SAR therefore covers a fraction of the gap between itself and the extreme point.

The acceleration factor. It starts at 0.02 and increases by 0.02 each time the extreme point makes a new high in a rise, or a new low in a fall, without ever exceeding 0.20. With no new extreme, it keeps its last value. This is what produces the parabola: the SAR starts slowly, then closes in on price more and more tightly as the trend extends.

The limit. In a rise, the SAR can never be above the lows of the prior two candles: if the calculation puts it higher, the lower of the two lows is used. In a fall, it can never be below the highs of the prior two candles. StockCharts and TradingView write the same rule, and the Parabolic SAR indicator code published by MetaQuotes, the developer of MetaTrader, applies it the same way.

The reversal. As soon as the price of a candle crosses the SAR, the trend being followed changes direction. According to TradingView, the SAR of that candle then takes the value of the previous extreme point: the highest high of the rise that just ended becomes the first SAR of the fall. The new extreme point is the extreme of the reversal candle, and the acceleration factor restarts at 0.02. MetaQuotes' code does exactly that.

The start of the history. TradingView writes that the book gives no clear instructions for the first value. Its convention: no SAR on the first candle; on the second, the direction comes from comparing the two closes, and the SAR equals the high of the first candle if the direction is down, its low if it is up. MetaQuotes' code, for its part, always starts in a fall. The first dots on a chart can therefore differ from one platform to another, until the first reversal.

The Parabolic SAR worked by hand on twelve candles

The prices are invented, the settings are the original ones: 0.02 to start, a step of 0.02, a maximum of 0.20. These twelve candles are candles 18 to 29 of the diagram above. Before them the market was falling, and its lowest low was 100.00. Values are rounded to two decimals.

Candle 1 (high 104.00, low 101.00). Price crosses the SAR of the fall: reversal. The SAR takes the previous extreme point, 100.00. New extreme point: 104.00. Acceleration factor: 0.02.

Candle 2 (106.00 and 103.00). SAR = 100.00 + 0.02 × (104.00 − 100.00) = 100.08. New high: the extreme point moves to 106.00 and the factor to 0.04. Candle 3 (109.00 and 105.00). SAR = 100.08 + 0.04 × (106.00 − 100.08) = 100.32. Extreme point 109.00, factor 0.06.

Candle 4 (108.00 and 105.50). SAR = 100.32 + 0.06 × (109.00 − 100.32) = 100.84. No new high: the extreme point stays at 109.00 and the factor stays at 0.06. Candle 5 (112.00 and 107.00). SAR = 100.84 + 0.06 × (109.00 − 100.84) = 101.33. Extreme point 112.00, factor 0.08.

Candle 6 (116.00 and 103.00), a wide candle. SAR = 101.33 + 0.08 × (112.00 − 101.33) = 102.18. Its low, 103.00, stays above it: no reversal. Extreme point 116.00, factor 0.10.

Candle 7 (118.00 and 112.00). The calculation gives 102.18 + 0.10 × (116.00 − 102.18) = 103.56. That is above the low of candle 6, 103.00: the limit applies, and the SAR is 103.00. Extreme point 118.00, factor 0.12. Candle 8 (121.00 and 115.00). The calculation gives 103.00 + 0.12 × (118.00 − 103.00) = 104.80. The prior two candles have lows of 112.00 and 103.00: the limit applies again, SAR = 103.00. Extreme point 121.00, factor 0.14.

Candle 9 (120.00 and 114.00). SAR = 103.00 + 0.14 × (121.00 − 103.00) = 105.52. No new high. Candle 10 (117.00 and 108.00). SAR = 105.52 + 0.14 × (121.00 − 105.52) = 107.69. The low, 108.00, holds by 0.31.

Candle 11 (111.00 and 105.00). The calculation gives 107.69 + 0.14 × (121.00 − 107.69) = 109.55, brought back by the limit to 108.00, the low of candle 10. Price drops to 105.00 and crosses it: reversal. The plotted SAR moves above price, to the previous extreme point, 121.00. New extreme point: 105.00. Factor: 0.02. Candle 12 (108.00 and 102.00). SAR = 121.00 + 0.02 × (105.00 − 121.00) = 120.68, and the fall is trailed in turn.

What the example shows. The SAR rose from 100.00 to 107.69 while the high went from 104.00 to 121.00: it starts far behind and only catches up with price at the end. Someone who had bought when price crossed the falling SAR, at 103.33 in the diagram, and exited at the 108.00 stop, would have kept 4.67 points of a 21-point rise from the low to the high. That is the cost of a rule that only exits after the top.

Parabolic SAR settings, platform by platform

The SAR is set with two or three numbers: the starting value of the acceleration factor, its step, and its maximum. The defaults are the same everywhere, 0.02 and 0.20, but the number of fields changes.

TradingView: three fields, 'Start' at 0.02, 'Increment' at 0.02 and 'Max Value' at 0.2, plotted as crosses by default. Its help page states that it adopted the original algorithm described in Wilder's book. StockCharts: 0.02 for the step and 0.20 for the maximum step, with an optional field to give the start a value different from the step. MetaTrader 5: the code published by MetaQuotes has only two settings, 'Step' at 0.02 and 'Maximum' at 0.2; the starting value there always equals the step, and the indicator is drawn as dots.

What the step changes. StockCharts explains it: a lower step moves the SAR further from price and makes a reversal less likely; a higher step moves it closer and multiplies reversals, to the point of failing to capture the trend if it is set too high. The maximum works the same way, with less weight. Its examples compare a step of 0.01 and of 0.03 on the same stock, then a maximum of 0.10 and of 0.20. Its conclusion: there is no golden rule or one-size-fits-all setting.

What MetaTrader 5's help writes differently. Its formula uses the maximal price of the previous candle where StockCharts, TradingView and MetaQuotes' own code use the extreme point of the trend, and it says the acceleration factor doubles, whereas the code adds one step to it. The two readings do not give the same number as soon as a candle fails to make a new high: on candle 5 of the example, the help formula taken literally would give 101.27 instead of 101.33. The code is what draws the dots on screen.

Reading the Parabolic SAR: reversal and trailing stop

The side of the dots. Dots under the candles: the trend being followed is up. Dots above: it is down. The switch from one side to the other is the only event the indicator produces. TradingView describes it as the sign of a potential price reversal, and notes that many traders use it to place stop orders, or to enter at the beginning of a new trend.

The trailing stop. This is the use all three sources put first. MetaTrader 5's help presents the SAR as an indicator for exit points: a long position is closed when price sinks below the line, a short position when it rises above, and it often serves as a trailing stop line. StockCharts adds a discipline argument: since the SAR never moves back against the position, it guards against the propensity to move a stop further away. The logic of the stop is covered in the article on where to place a stop loss, and the order itself, with what becomes of it on a price gap, in the trailing stop section of the article on order types.

The full system. The name says 'stop and reverse': the rules TradingView reproduces describe a long trade that follows the short trade at the very point where the short one stops, and vice versa. The same page nevertheless advises against using it as a stand-alone to generate signals, and suggests pairing it with an indicator that measures trend strength, for example the same author's Directional Movement. StockCharts names the ADX for the same role: that is the lesson on the DMI and the ADX.

None of these readings is an order to buy or sell. They are conventions, to be tested on your market, your timeframe and your settings, with the method in the article on how to backtest a trading strategy. Tradoshi does not record the SAR at the time of your trades: the indicator is read on your chart. The journal is for the next step: you tag every trade exited on this rule, for example 'SAR exit', and the dashboard filters by tag to show what it produced for you.

The limits of the Parabolic SAR

With no trend, it keeps flipping. StockCharts writes that the SAR works best with trending securities, which, according to Wilder's estimates as it reports them, occur only roughly 30 % of the time, and that it is prone to whipsaws when a security is not trending. In the diagram, the last twenty candles swing between 89.80 and 100.80: the SAR flips four times there. Assuming each order is filled at the exact SAR level, with no costs and no slippage, the three round trips completed there lose 9.05, 9.71 and 9.14 points, in a zone only 11 points wide.

It gives back part of the move. The stop is behind price by construction. In the example, the exit comes at 108.00 after a high at 121.00. A higher step would exit sooner, and flip more often: no setting removes that trade-off.

It does not measure trend strength. TradingView says so: direction and duration only. Nor does it measure volatility: the distance between the SAR and price comes from the acceleration factor and the time elapsed, not from the range of the candles, which is the question the ATR answers.

The plotted level is not a guaranteed price. The SAR is a calculated number. A stop order placed at that level can be filled further away if the market opens beyond it. And the dot changes on every candle: an order placed once does not follow it on its own.

No study cited here measures it. Published tests cover other rules. Sullivan, Timmermann and White, in Data-Snooping, Technical Trading Rule Performance, and the Bootstrap (Journal of Finance, 1999), examine 7,846 rules from five families, filters, moving averages, support and resistance, channel breakouts and volume, on the Dow Jones index: the best one survives their correction from 1897 to 1986, and is not repeated from 1987 to 1996. The SAR is not among them, and that result says nothing about it. It only says that a technical rule is judged on data it has not already seen. The SAR also remains a trend-following tool, like moving averages: the page on technical analysis explains why two tools from the same family do not make two confirmations.

Frequently asked questions