The default Parabolic SAR settings are 0.02 start and 0.20 maximum. Here is what changing them does and when deviation is justified.
Parabolic SAR settings consist of two parameters: the starting acceleration factor and the maximum acceleration factor. Wilder's defaults are 0.02 and 0.20. Most traders never change them. The defaults work well enough across most markets that changing them without a systematic, tested reason typically produces no improvement and sometimes makes things worse.
The starting acceleration factor (AF start) determines how quickly SAR begins tracking price at the onset of a new trend. A lower AF start means the dots begin farther from price and take longer to accelerate toward it. A higher AF start means the dots begin closer to price from the first bar of the new trend.
The maximum acceleration factor (AF max) determines the tightest the trailing stop can get, regardless of how long the trend runs. Once AF reaches the maximum, it stops increasing. The dots continue moving — they track the extreme price level — but the rate of acceleration no longer increases. A lower AF max produces a looser trailing stop throughout the trend. A higher AF max allows the stop to become very tight in a mature trend.
The increment at which AF increases with each new extreme point is fixed at 0.02 by Wilder's design — the step from AF start toward AF max. On most platforms this is not a separately configurable parameter. With the default settings, a trend must set 10 new extreme points before AF reaches maximum (10 increments of 0.02 from 0.02 to 0.20).
Wilder selected these parameters through empirical observation on commodity daily charts in the 1970s. The 0.02 start was chosen to give new trends room to develop without premature stop-outs from a tight initial stop. The 0.20 maximum ensured that very mature trends would not allow excessive retracement before the signal flipped.
The 10-step progression from start to maximum meant a trend needed to set 10 new extreme points before the AF reached its ceiling. In Wilder's commodity markets, this corresponded to a well-established, mature trend with clear directional momentum. The stop would be tight precisely when a trend was statistically most likely to be near exhaustion.
These parameters have generalized reasonably well to other markets for the same reason his ADX settings have: the underlying logic (give early trends room, tighten stops on mature trends) is sound and market-agnostic, even if the specific calibration was developed for a different context. This is why the defaults remain the best starting point for most systematic applications.
Lower AF start (0.01): Dots begin farther from price. The initial stop is wider. Early-trend whipsaws are less likely because the stop has more room. Trade-off: if a trend fails immediately, the initial loss is larger. Appropriate for markets where early-trend volatility is high and strategies with longer intended holding periods.
Higher AF start (0.03 to 0.05): Dots begin closer to price. The initial stop is tighter. Earlier exits on failed trends but more false flips at the start of genuine trends before they establish direction. Appropriate for markets where trends establish quickly and strategies that prioritize early exit on failed trends.
Lower AF max (0.10 to 0.15): The stop never becomes as tight, even in very mature trends. More retracement allowed before exit. More of a long-running trend is captured but more gains may be given back before the SAR flip triggers. Appropriate for longer-duration trend strategies and assets with naturally wider swings.
Higher AF max (0.25 to 0.30): The stop becomes very tight in mature trends. Exits occur earlier in trend deterioration. Less retracement allowed, more frequent exits but potentially at better prices during reversals. Appropriate for faster-moving assets and shorter holding period strategies where preserving gains matters more than capturing the full trend extension.
The default 0.02/0.20 settings work across most markets without modification. Before deviating, have a clear reason derived from systematic testing on the specific asset and timeframe.
Daily charts: 0.02/0.20 default. Wilder's original calibration. No strong reason to deviate unless the specific asset consistently shows early-trend whipsaws (consider lowering AF start to 0.01) or consistently gives back too much on reversals (consider raising AF max to 0.25).
4-hour crypto charts: 0.02/0.20 default works well for most assets. Some traders use 0.02/0.15 for a slightly looser trailing stop that reduces early exits in choppy 4H conditions. Crypto's higher volatility can cause the standard 0.20 maximum to trigger exits on normal volatility rather than genuine reversals.
1-hour charts: Consider 0.02/0.15 or 0.01/0.20 to reduce false flips where price noise is proportionally higher relative to the trend's amplitude. The shorter the timeframe, the more likely normal price noise will breach a tight SAR stop that was calibrated for a slower-moving market.
The most important rule: the same parameter change that improves results on one asset may worsen results on another. Always test on the specific asset and timeframe being traded. Never transfer settings from one market to another without independent validation.
If the defaults consistently produce a specific problem, test alternative settings with a defined methodology before committing.
Define the problem precisely. "Too many whipsaws" means: how many trades exit within the first 3 bars at a loss before the trend continues? "Too much retracement" means: what percentage of average peak unrealized profit is given back before the SAR exit triggers? Precise problem definition leads to targeted parameter adjustment rather than random exploration.
Test a range, not a single alternative. For a whipsaw problem: test AF starts of 0.01, 0.015, 0.02, 0.025. For a retracement problem: test AF max values of 0.15, 0.18, 0.20, 0.25. Run each setting on the same historical dataset and measure the specific metric being targeted.
Apply a time-ordered split. Develop the test on the first half of available data. Validate on the second half without adjustment. If the improved setting holds on the out-of-sample half, it may reflect genuine market characteristics. If it degrades, it was fitting noise in the in-sample period.
Reject marginal improvements. A parameter that improves the target metric by 3% in-sample is not a meaningful finding. Only changes that produce clear, consistent improvement across multiple test periods justify deviation from the defaults. Marginal gains in backtesting typically disappear in live trading.
Overfitting. The defaults are less "optimal" on any specific historical dataset than custom-tuned settings will appear to be. This is precisely why the defaults are the better starting point. A custom setting showing 5% better performance on 2020-2023 data is likely fitting the characteristics of that specific period, not discovering a better general parameter. Out-of-sample validation is the only check against this.
Asset-specific calibration does not transfer. SAR settings that work well on BTC/USD daily charts may not work on ETH/USD 4H charts. Different assets have different trend characteristics, volatility profiles, and cycle lengths. Settings calibrated on one asset should never be applied to others without independent testing on that specific asset.
Platform inconsistencies. Some platforms implement SAR slightly differently, particularly in how the initial SAR value is calculated at the start of a new trend. A parameter setting that produces specific behavior on one platform may produce different behavior on another. Verify the platform's implementation matches expected behavior when applying custom settings, especially if migrating between platforms.
Changing market conditions. Even correctly validated settings can stop working when an asset's underlying characteristics change. Crypto market dynamics in 2021 differed significantly from 2022 or 2023. Settings calibrated for one regime may need review as market conditions evolve. Monitoring live performance against backtested expectations provides the ongoing signal that recalibration may be needed.
Wilder's default Parabolic SAR settings are a starting acceleration factor (AF) of 0.02 and a maximum acceleration factor of 0.20. The AF increases by 0.02 each time price sets a new extreme point in the trend direction, reaching the maximum of 0.20 after 10 new extreme points. These defaults work across most markets and timeframes without modification and remain the recommended starting point for systematic applications.
Start with the defaults: 0.02 start, 0.20 maximum. Crypto's higher volatility sometimes causes the 0.20 maximum to produce stops that are too tight on mature trends, getting hit by normal daily volatility rather than genuine reversals. Testing a lower AF maximum (0.15 to 0.18) on 4-hour crypto charts is a reasonable first experiment. Always validate any change on the specific asset using a time-ordered data split before applying to live trading.
Changing the starting AF affects how tight the stop is at the beginning of a new trend. A lower starting AF (0.01) places the initial stop farther from price, reducing early-trend whipsaws at the cost of a wider initial loss if the trend fails. Changing the maximum AF affects how tight the stop gets in mature trends. A lower maximum (0.15) allows more retracement before exit. A higher maximum (0.25) tightens the stop more aggressively as the trend ages, producing earlier exits during trend deterioration.
Only if you have identified a specific, measurable problem with the defaults on a specific asset and timeframe, and you have validated the improvement on out-of-sample data. Changing settings because a different number looks better or because someone else recommends it is not a systematic reason. The defaults are less optimal than custom-tuned settings on any specific historical dataset — which is exactly why the defaults generalize better to data they have not been fitted to.
The acceleration factor (AF) is the multiplier in the SAR calculation that determines how quickly the trailing stop moves toward price as the trend develops. It starts at the AF start value (default 0.02) and increases by 0.02 each time price sets a new extreme point in the trend direction, up to the AF maximum (default 0.20). A low AF produces dots far from price (wide trailing stop). A high AF produces dots close to price (tight trailing stop). The acceleration is what gives Parabolic SAR its characteristic tightening behavior in mature trends.
For day trading on 1-hour charts, consider testing AF start 0.02 with AF maximum 0.15 to reduce false flips from short-term price noise. A lower maximum gives the intraday trend more room before the stop tightens to the point where normal volatility triggers it. For 15-minute charts, consider starting even lower at 0.01 to give the trend room to establish before the acceleration begins. Test any deviation from defaults on the specific asset using out-of-sample validation before applying live.
Define the specific problem first (early whipsaws or excessive retracement), then test a range of values targeting that parameter. Split historical data in half chronologically. Test candidate settings on the first half only. Apply the best-performing setting unchanged to the second half. If the improvement holds out-of-sample, the setting may reflect genuine market characteristics. If it degrades, it was fitting past noise. Only changes showing consistent improvement across both halves justify deviation from the defaults.