The trend is NOT your friend!S&P 500SP:SPXEdgeTools32 Million Tests Exposed What Supertrend Actually Does "The trend is your friend." It is the first thing retail traders learn and the last thing they question. The sentence is intuitive, almost comforting. It promises that you do not need to predict the future, you just need to recognize the present direction and ride along. The Supertrend indicator, one of the most downloaded on TradingView, is the purest expression of this idea: a single line that tells you which way the market is going. Green means up. Red means down. Follow along and the trend will take care of you. But does it? Academic finance has studied momentum effects for decades, and the findings are more nuanced than the retail version of the story. The momentum effect is real, one of the most robust anomalies in financial economics. The distance between "momentum exists as a documented factor" and "this indicator on your chart tells you when to buy" is what this study measures. We tested 32,751,398 Supertrend configurations across 14 assets, five strategies, and 35 holding horizons. The goal was not to confirm or dismiss the indicator, but to understand exactly where it contains information and where it does not. The answer turned out to be more interesting than a simple yes or no. The Supertrend does contain real, statistically robust structure, but the structure is not where most traders look for it. It lives in the distance between price and the Supertrend line, not in the direction the line points. And that finding connects to a pattern this series has been documenting since the VWAP study: the consistent edge in technical analysis is not in following the trend. It is in recognizing when the trend has been stretched too far. Abstract We test five common Supertrend trading strategies across 14 liquid ETFs spanning five asset categories. From 32,751,398 parameter configurations covering ATR periods from 3 to 150, multipliers from 0.5 to 8.0, and holding periods from 1 to 252 trading days, we find 1,410,697 results surviving Bonferroni correction at alpha equal to 1.53 times ten to the negative ninth power. The aggregate long edge is positive 0.26 percentage points and the aggregate short edge is negative 0.63 percentage points. These averages are dominated by one strategy. Distance entry, where positions are taken when price is far from the Supertrend line, produces long edge of positive 0.67 percentage points with 877,891 Bonferroni-significant long results and negative short edge of minus 1.29 percentage points with 484,670 significant short results. Together, distance-based results account for 1,362,561 of the 1,410,697 total Bonferroni results, a concentration of 96.6 percent. The direction flip signal, the strategy retail traders most commonly associate with Supertrend, produces zero Bonferroni-significant results from 2,248,014 tests. The confirmation strategy produces zero from 11,240,600 tests. The Supertrend indicator contains genuine structure, but that structure is mean reversion from trend overextension, not trend following. This result extends the pattern documented across the previous five studies in this series and provides a concrete framework for building strategies around overextension from dynamic reference levels. 1. Introduction The Supertrend indicator was developed by Olivier Seban and popularized through trading platforms in the mid-2000s. It places a single adaptive line above or below price, calculated from the Average True Range, that flips direction when price crosses it. When price is above the Supertrend line, the indicator is bullish and the line sits below price as dynamic support. When price is below, the indicator is bearish and the line sits above as resistance. The visual output is clean and unambiguous: a green line below price means buy, a red line above means sell. This visual clarity made Supertrend one of the most popular indicators on TradingView. It appears in countless strategy scripts, tutorial videos, and trading courses. The appeal is the appeal of all trend-following tools: it promises to keep you on the right side of the market and tell you exactly when the trend has changed. The indicator builds on a sound mechanical foundation. The Average True Range, introduced by Wilder (1978), measures the actual trading range of an asset, incorporating gaps between sessions. By using ATR rather than a simple moving average of price, the Supertrend adapts to volatility. In calm markets, the line sits close to price and flips frequently. In volatile markets, it gives price more room and flips less often. This adaptive behavior is a genuine improvement over fixed-threshold trend indicators. What the indicator does not guarantee is that following its signals generates above-average returns. A Supertrend that flips to bullish at the start of a rally has identified the trend. Whether buying at that flip point produces returns above what you would have earned by holding the asset regardless is a separate, empirical question. The academic momentum literature, which we examine in section 8, provides context: momentum as documented in peer-reviewed research and momentum as implemented through a Supertrend on a single chart are different claims. Testing which aspects of the indicator contain real information requires the kind of exhaustive parameter search we conduct here. 2. What Supertrend measures The Supertrend calculation begins with the Average True Range: True Range = max(High - Low, |High - Previous Close|, |Low - Previous Close|) ATR is an exponential moving average of True Range over n periods: ATR(n) = (1/n) * TR + (1 - 1/n) * ATR(n-1) The Supertrend then computes two bands around the midpoint of the current bar: Upper Band = (High + Low) / 2 + factor * ATR(n) Lower Band = (High + Low) / 2 - factor * ATR(n) The Supertrend line itself is determined iteratively. In an uptrend, the line equals the lower band but never decreases: it ratchets upward as long as price stays above it. In a downtrend, the line equals the upper band but never increases: it ratchets downward as long as price stays below it. When price crosses the Supertrend line, the direction flips and the line jumps to the opposite band. The standard parameterization uses an ATR period of 10 and a factor of 3.0. Unlike Bollinger Bands, where the standard 20/2 setup was specified by the creator, there is no canonical Supertrend parameterization. Different platforms default to different values, and the trading community uses a wide range. This ambiguity is itself worth testing: if the indicator works, it should work across a broad parameter space, not only at one specific setting. The mathematical structure of the Supertrend is worth examining relative to other indicators in this series. RSI and MACD transform closing prices. Bollinger Bands use closing prices but access a second-order statistic through standard deviation. Supertrend uses high, low, and close, and accesses volatility through ATR. ATR is a first-order statistic of the trading range, not a second-order statistic of return dispersion. It measures how much the asset moved, not how dispersed the returns were around their mean. This is a subtler distinction than it appears. Bollinger Band width reflects return variance. ATR reflects absolute price range. Both capture some aspect of volatility, but through different lenses. Whether this matters for predictive power is part of what the data reveals. 3. Common Supertrend strategies We tested five strategies representing how retail traders and systematic strategy builders use the Supertrend indicator. The trend following strategy generates a long signal whenever the Supertrend direction is bullish and a short signal when it is bearish. This is the simplest interpretation: be long when the line is green, be short when the line is red. It tests the fundamental claim of the indicator, that the direction classification contains information about future returns. The direction flip strategy generates signals only at the moment of transition. A long signal fires when the Supertrend changes from bearish to bullish. A short signal fires at the opposite transition. This is the strategy that produces the green and red arrows on TradingView charts. It tests whether the timing of the flip, rather than the ongoing state, contains predictive power. The band bounce strategy identifies moments when price is very close to the Supertrend line in the direction of the trend. In an uptrend, price occasionally pulls back to nearly touch the rising Supertrend line before resuming higher. Traders interpret this as the trend line acting as dynamic support. A long signal fires when price is within 0.5 percent of the Supertrend line during a bullish regime. The short equivalent fires during bearish regimes. The distance entry strategy measures how far price has moved from the Supertrend line and generates signals when the distance exceeds a threshold. In a bullish regime, large distance means price has rallied well above the Supertrend. In a bearish regime, it means price has fallen well below. Seven distance thresholds from 1 to 10 percent are tested. This strategy tests whether overextension from the trend line contains information about subsequent returns. The confirmation strategy requires the Supertrend to have maintained the same direction for a specified number of consecutive bars before generating a signal. The idea is that new trends are unreliable and only established trends are worth following. Five confirmation thresholds from 2 to 10 bars are tested. Of these five, the first three represent how retail traders actually use the indicator. The direction flip is the default signal. The trend filter is used in conjunction with other indicators. The bounce is the "buy the dip to the trendline" approach taught in courses. The distance and confirmation strategies are less common in retail practice but represent systematic extensions of the indicator's logic. 4. Data and methodology 4.1 Asset universe We tested the same 14 liquid ETFs used in the Bollinger Band study: SPY, QQQ, IWM, and DIA for US equities; EFA, EEM, and VWO for international equities; GLD and SLV for commodities; TLT for bonds; XLV, XLE, XLF, and XLK for sectors. All data is daily, sourced from TwelveData with Tiingo as fallback, covering approximately 5,000 trading days per asset. 4.2 Parameter grid ATR periods range from 3 to 150 in steps of 1, giving 148 values. Multipliers range from 0.5 to 8.0 in steps of 0.25, giving 31 values. This produces 4,588 unique Supertrend configurations per asset. Holding periods span 35 values from 1 to 252 trading days: 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 18, 20, 22, 25, 28, 30, 35, 40, 45, 50, 60, 75, 90, 100, 120, 150, 180, 200, and 252. Distance thresholds use 7 values from 1 to 10 percent. Confirmation bars use 5 values from 2 to 10 consecutive bars. The total configuration count across 14 assets is 33,721,800 target tests. After filtering for sufficient data length and minimum signal counts, 32,751,398 valid tests remain. 4.3 Forward return measurement Edge is measured as the difference between mean forward returns following a signal and mean forward returns across all bars in the same asset sample. This baseline adjustment ensures that strategies in rising markets do not receive credit for capturing beta. A Supertrend long signal in SPY that produces the same return as holding SPY has zero edge. The indicator must beat the baseline, not merely be positive. 4.4 Statistical framework Significance is assessed using Welch's t-test for unequal variances. Given 32,751,398 tests, the Bonferroni-corrected significance threshold is 1.53 times ten to the negative ninth power. This is the strictest correction applied in the series. A result surviving this threshold would occur by chance fewer than once in 650 million tries under the null hypothesis. 5. Results 5.1 Overview Figure 1 presents the aggregate view. The distance strategy produces a wide positive distribution on the long side and a wide negative distribution on the short side, dominating the chart. Trend, flip, bounce, and confirmation strategies cluster tightly around zero on both sides. Across all 32,751,398 tests, mean long edge is positive 0.26 percentage points and mean short edge is negative 0.63 percentage points. The positive aggregate long edge and the negative aggregate short edge both come from the same source: the distance strategy, which captures a long-biased mean reversion effect. Remove the distance strategy, and the remaining 17,974,824 tests produce aggregate long edge of negative 0.07 percentage points and short edge of negative 0.06 percentage points. Indistinguishable from zero. 5.2 Results by strategy The direction flip is the strategy retail traders most associate with the Supertrend. It is the green arrow. From 2,248,014 tests, mean long edge is negative 0.24 percentage points and mean short edge is negative 0.17 percentage points. Bonferroni-significant results: zero. The flip signal does not contain predictive information about future returns beyond what holding the asset already provides. If someone asks "does buying on the Supertrend flip beat holding?", the answer across 2.2 million configurations is no. This does not mean the indicator is broken. It means the flip event itself is not where the information lives. The data points somewhere else, and we get there in the distance results below. The confirmation strategy extends the flip logic by requiring the Supertrend to maintain direction for N bars before entering. From 11,240,600 tests, mean long edge is negative 0.06 percentage points and mean short edge is negative 0.16 percentage points. Bonferroni-significant results: zero. Waiting for the trend to establish itself delays the entry without adding signal. Section 6 explains the mechanical reason for this. The trend following strategy, staying positioned with the Supertrend direction, produces a nuanced result. From 2,248,120 tests, mean long edge is negative 0.003 percentage points and short edge is negative 0.049 percentage points. Both are economically zero. However, 15,845 long results and 32,170 short results survive Bonferroni correction. The explanation is sample size: the trend strategy holds positions for extended periods, generating thousands of signal bars per configuration. The t-test can detect tiny deviations from zero with enough observations. The finding is real but too small to trade. It does confirm that the Supertrend direction classification is not random, it captures something, just not enough to build a strategy on by itself. The bounce strategy tests whether the Supertrend line acts as dynamic support and resistance. From 2,238,090 tests, mean long edge is negative 0.05 percentage points and mean short edge is positive 0.31 percentage points. 121 total Bonferroni-significant results. The line shows trace evidence of a support/resistance function, but at a scale that is not practically useful. The Supertrend line is a better reference level for measuring distance than for identifying touch points. The distance entry strategy produces the results. From 14,776,574 tests, mean long edge is positive 0.67 percentage points with 877,891 Bonferroni-significant results. On the short side, mean edge is negative 1.29 percentage points with 484,670 significant results. Long edge of positive 0.67 percentage points means: when the Supertrend is bullish and price is far above the line, buying produces returns that exceed baseline by 0.67 percentage points on average. Strong momentum tends to persist at medium to long horizons. Short edge of negative 1.29 percentage points means: when the Supertrend is bearish and price is far below the line, the market tends to bounce. Downside overextensions correct. Both findings point in the same direction: the market gives back extreme moves. Long-side distance captures momentum persistence combined with the equity risk premium. Short-side distance captures mean reversion from panic-driven overshoot. The Supertrend line, because it adapts to volatility through ATR, provides a useful reference for quantifying how far "too far" is. Together, distance-based results account for 1,362,561 of the 1,410,697 total Bonferroni results: 96.6 percent. This concentration matters for strategy design: the Supertrend contains real information, but it is concentrated in one specific usage pattern that differs from the way most tutorials teach the indicator. 5.3 Statistical significance The p-value distribution departs from uniformity, with 18.2 percent of long signals and 20.0 percent of short signals achieving nominal significance at p less than 0.05. These rates are roughly 3.5 to 4 times the chance level of 5 percent. After Bonferroni correction, 893,804 long and 516,893 short results survive, the highest absolute count in the series. But the count is misleading without context. Almost all significant results come from the distance strategy, not from the strategies that traders actually use. 5.4 Results by asset category International equities show the strongest effects: long edge positive 0.53 and short edge negative 0.89 percentage points. US equities follow with long edge positive 0.40 and short edge negative 0.67 percentage points. Sector ETFs show long edge positive 0.16 and short edge negative 0.70 percentage points. Commodities are nearly flat: long edge positive 0.005 percentage points and short edge negative 0.32 percentage points. The Supertrend distance effect is an equity phenomenon. Commodities, driven by supply shocks and mean-reverting inventory cycles, do not exhibit the same long-biased overextension pattern. Bonds show the only reversal: long edge negative 0.22 percentage points and short edge near zero. Being positioned with a bullish Supertrend in TLT generates returns below baseline. Bond dynamics are dominated by central bank policy and duration risk, neither of which an ATR-based trend indicator captures. 5.5 Parameter sensitivity The parameter landscape reveals a pattern consistent with the previous studies. Long edge increases with holding period, particularly for ATR periods between 10 and 60. Short edge is negative across nearly the entire grid, deepening at longer holding periods. The strongest long effects appear at moderate ATR periods combined with holding periods of 60 to 252 days. Figure 7 isolates the holding period dimension. Long edge is near zero for holding periods under 10 days and increases steadily to approximately 0.7 percentage points at the 252-day horizon. Short edge starts slightly negative and deteriorates continuously, reaching approximately negative 1.5 percentage points at longer horizons. The message is consistent with the distance strategy interpretation: short-term Supertrend signals carry no edge, while long-horizon positions capture the long bias filtered through the indicator's trend classification. 6. Why the flip signal fails and what that teaches us Understanding why the flip produces no edge is more useful than simply knowing that it does not. The mechanical reason is whipsawing. The Supertrend flips when price crosses the line, but the line itself is a function of recent ATR. In choppy, range-bound markets, price oscillates around the Supertrend repeatedly. Each crossing triggers a flip. Each flip fires a signal into a market that is going nowhere. The Supertrend has no filter for market regime. In a trending market, flips are infrequent and directionally meaningful. In a choppy market, they are frequent and random. An ATR-based line cannot distinguish between the two states: trend volatility and range volatility look the same to it. Huang, Li, Wang, and Zhou (2020) showed that trend-following profits are concentrated in high-uncertainty states. An indicator that fires signals regardless of regime averages across states where it has edge and states where it has negative edge, and the average is approximately zero. This explains why the confirmation strategy also fails: by the time the Supertrend has been bullish for 10 consecutive bars, the move has already happened. The entry is later, the remaining edge smaller, and the noise reduction does not compensate for the timing cost. Zakamulin (2014) found the same pattern for moving average strategies: after accounting for data snooping, most crossing rules lose significance. The Supertrend adds ATR-based adaptation, which is a genuine sophistication, but adaptation to volatility does not solve the fundamental problem. What does solve it is asking a different question: not "which direction is the trend?" but "how far has price moved from the trend?" That question leads to the distance strategy, where the edge lives. 7. Why the distance effect exists On the long side, positive distance in a bullish regime means price has rallied well above the rising Supertrend line. This configuration appears during strong momentum periods: sharp rallies following corrections, earnings-driven gaps, and macro-driven sector rotations. The data shows that these momentum periods tend to persist at medium to long horizons, consistent with Moskowitz, Ooi, and Pedersen (2012). On the short side, negative distance in a bearish regime means price has fallen well below the declining Supertrend line. This appears during panic selloffs, credit events, and cascading liquidations. Returns following deep bearish overextension are significantly above baseline, meaning shorts lose money. This is the same mean reversion from extremes documented in the Bollinger Band and VWAP studies. The asymmetry between long and short distance effects is notable: positive 0.67 versus negative 1.29 percentage points. Short-side mean reversion is roughly twice as strong as long-side momentum persistence. Drawdowns are faster and sharper than rallies, creating more pronounced overextension on the downside and stronger reversion when selling pressure exhausts. The Supertrend line serves a genuine purpose here, just not the one most traders expect. The ATR-based adaptation scales the reference level to current volatility, which makes it a good instrument for measuring "how far is too far." That property is valuable for strategy design. The finding applies to any volatility-adaptive reference level, but the Supertrend's clean visual output makes it one of the most practical implementations on TradingView. 8. What the academic literature tells us about making this work The academic momentum literature provides a roadmap, not for dismissing the Supertrend, but for understanding how to extract value from trend-based analysis. Jegadeesh and Titman (1993) documented cross-sectional momentum: stocks that outperformed over 3 to 12 months tend to continue outperforming. Moskowitz, Ooi, and Pedersen (2012) extended this to time-series momentum across 58 futures contracts. Both findings are robust and widely replicated. Momentum is real. The gap between the academic evidence and the retail Supertrend experience comes from three specific differences. First, the academic version trades dozens of instruments simultaneously. Diversification across uncorrelated markets is itself a source of risk reduction that a single-chart application forfeits. Second, the academic version uses simple past returns as the signal, not an ATR-based indicator with specific parameters. The measurement is simpler and less susceptible to overfitting. Third, Baltas and Kosowski (2013) showed that momentum profits depend on the rebalancing window: long lookbacks with infrequent rebalancing capture the effect, while short lookbacks with frequent rebalancing generate mostly transaction costs. These are not reasons to abandon trend analysis. They are design specifications. The distance strategy result in this study is consistent with the academic findings: it captures momentum at longer horizons (60 to 252 days), it works across multiple assets, and it measures overextension rather than directional flips. A trader who wants to use the Supertrend profitably can use the academic literature as a checklist: diversify across assets, extend the holding period, and focus on the distance from the line rather than the direction of the line. 9. Where Supertrend fits in the series Six indicators. Ninety-nine million tests. One framework. The comparison across studies reveals a pattern that is becoming increasingly useful for strategy design: RSI: zero Bonferroni-significant results from 26 million tests. Turn of the Month: 21 significant results from 385 tests. A real calendar anomaly. VWAP: 150,546 significant results. Distance-from-mean edge of 0.89 percentage points (short). MACD: 3,235 significant results. Histogram divergence long edge of 0.32 percentage points. Bollinger Bands: 320,256 significant results. Band penetration long edge of 1.22 percentage points. Supertrend: 1,410,697 significant results. Distance-based long edge of 0.67 percentage points. The three strongest findings, VWAP, Bollinger, and Supertrend, all share the same structure: they measure how far price has deviated from a dynamic reference level. VWAP uses a volume-weighted mean. Bollinger uses a standard deviation envelope. Supertrend uses an ATR-adjusted trend line. Different reference levels, same principle: when price moves far from where the data says it typically sits, it tends to correct. This is not a negative finding. It is a blueprint. The data across 99 million tests consistently points to overextension as the exploitable structure in technical indicators. That gives traders something specific to look for and build on. 10. How to use this The data across 32.8 million tests gives specific, actionable guidance for anyone working with the Supertrend or building trend-based strategies. Use the Supertrend as a reference level, not a signal generator. The ATR-based line adapts to volatility, which makes it a strong yardstick for measuring how stretched price is relative to the current regime. The direction flip does not beat holding. But the distance between price and the line identifies conditions with genuine statistical edge. That reframing, from "follow the arrow" to "measure the distance," is the practical takeaway. Focus on longer holding periods. The distance effect is near zero at short horizons and grows steadily toward 60 to 252 day holdings. This aligns with the academic momentum literature: the effect operates at medium to long horizons, not at the daily level where most retail strategies live. Diversify the application. The distance effect is strongest in equities, weaker in commodities, and absent in bonds. A strategy built on Supertrend distance across multiple equity ETFs captures diversification benefits that a single-chart approach cannot access. Combine with a regime filter. A Supertrend distance signal that activates only during elevated VIX or widening credit spreads targets the market state where the effect is strongest. The broader principle from this series is now well-established: the overextension from a dynamic reference level, whether that reference is a volume-weighted mean, a standard deviation band, or an ATR-adjusted trend line, is where exploitable structure consistently appears in technical indicators. That gives traders a clear direction for strategy development, grounded in 99 million tests across six indicators. 11. Limitations The analysis uses daily data only. Supertrend is frequently applied to intraday timeframes, particularly the 15-minute and 1-hour charts. The indicator may behave differently at higher frequencies where intraday momentum dynamics differ from daily patterns. The study tests each strategy in isolation. Combining Supertrend direction with other indicators, volume filters, or volatility regime detection could alter results. The distance strategy in particular might benefit from conditional filters that distinguish between momentum-driven and mean-reverting market states. Execution is assumed at the close of the signal bar. In practice, Supertrend flips are often visible only after the bar closes, meaning the realistic entry is the next day's open. Overnight gaps could reduce or augment the observed effects. Transaction costs were not deducted from the edge figures. For the distance strategy's long edge of 0.67 percentage points, round-trip costs of 0.10 to 0.15 percentage points leave a net edge of approximately 0.52 to 0.57 percentage points. This is positive but not large, and it deteriorates further for less liquid instruments or higher rebalancing frequencies. 12. Conclusion 32,751,398 configurations. Five strategies. Fourteen assets. The Supertrend indicator contains real, statistically robust information. It is just not in the signal that most traders use. The direction flip and confirmation strategies produce zero Bonferroni-significant results. The bounce strategy produces negligible results. These are the strategies taught in tutorials and coded into default scripts. They do not beat holding the asset. The distance strategy produces 1,362,561 significant results with long edge of 0.67 percentage points and short-side mean reversion of 1.29 percentage points. The Supertrend line, because it adapts to volatility through ATR, serves as an effective reference level for measuring overextension. This study completes six indicators and 99 million tests. The consistent finding across the series is that indicators contain the most useful information when they measure how far price has deviated from a dynamic reference, not when they generate directional signals. VWAP, Bollinger Bands, and now Supertrend all point to the same principle. The trend is your friend in the academic sense. Momentum is a documented and robust factor in financial markets. But the Supertrend indicator, as it is typically used on a single chart, captures that factor most effectively through distance measurement, not through the green and red arrows. The trend is the reference line. The opportunity is the deviation from it. And that distinction is the foundation for building strategies that the data actually supports. References Baltas, A.N. and Kosowski, R. (2013). Momentum strategies in futures markets and trend-following funds. European Financial Management, 19(3), pp. 1-44. Huang, D., Li, J., Wang, L. and Zhou, G. (2020). Time series momentum: Is it there? Journal of Financial Economics, 135(3), pp. 774-794. Jegadeesh, N. and Titman, S. (1993). Returns to buying winners and selling losers: Implications for stock market efficiency. Journal of Finance, 48(1), pp. 65-91. Moskowitz, T.J., Ooi, Y.H. and Pedersen, L.H. (2012). Time series momentum. Journal of Financial Economics, 104(2), pp. 228-250. Wilder, J.W. (1978). New Concepts in Technical Trading Systems. Trend Research, Greensboro, NC. Zakamulin, V. (2014). The real-life performance of market timing with moving average and time-series momentum rules. Journal of Asset Management, 15(4), pp. 261-278.