Understanding the Core Philosophy of Trend Following
Trend following operates on a foundational premise that transcends asset classes and market cycles: prices tend to persist in their existing direction until a fundamental shift occurs. Unlike mean reversion strategies that bet on prices returning to historical averages, trend following aligns with the momentum anomaly—one of the most robust and persistent factors documented in academic finance. This momentum effect, extensively studied by researchers like Jegadeesh and Titman, demonstrates that assets performing well over the past three to twelve months continue outperforming, while underperformers keep lagging. For long-term investors, this creates an opportunity to systematically participate in sustained market moves while sidestepping prolonged drawdowns. Exchange-traded funds provide an ideal vehicle for implementing this strategy, offering liquidity, transparency, low costs, and broad diversification within a single ticker. By combining trend-following principles with ETF efficiency, investors can construct a rules-based system that removes emotional decision-making and captures the bulk of major market advances.
Why ETFs Are the Optimal Tool for Trend Following
ETFs possess several structural advantages that make them superior to mutual funds or individual securities for trend-following implementation. First, intraday liquidity allows precise entry and exit points based on technical signals, whereas mutual funds only trade at end-of-day net asset values. Second, ETFs trade like stocks, meaning stop-loss orders, limit orders, and trailing stops can be executed seamlessly. Third, expense ratios for broad-market ETFs often range from 0.03% to 0.20%, dramatically lower than actively managed mutual funds. Fourth, ETFs provide exposure to diverse asset classes—equities, fixed income, commodities, real estate, and currencies—enabling multi-asset trend following within a single brokerage account. Fifth, tax efficiency stems from in-kind creation and redemption mechanisms, reducing capital gains distributions. Sixth, transparency allows investors to know exactly what they own each day, which is critical when applying mechanical signals. Finally, the sheer number of available ETFs exceeds 8,000 globally, covering niche sectors, factors, geographies, and alternative strategies, giving trend followers a vast opportunity set.
The Mathematical Foundation: Moving Averages and Breakouts
Two primary signal-generation methods dominate trend following: moving average crossovers and channel breakouts. The simple moving average calculates the average price over a specified period, smoothing noise to reveal direction. A common system uses a 50-day and 200-day moving average: buy when the 50-day crosses above the 200-day (golden cross), sell when it crosses below (death cross). The exponential moving average weights recent prices more heavily, responding faster to changes. For breakouts, the Donchian channel—developed by Richard Donchian—defines the highest high and lowest low over N periods. A buy signal triggers when price exceeds the upper band; a sell signal when price falls below the lower band. Turtle Traders famously used 20-day and 55-day breakouts. Another robust technique is the moving average envelope, where price crossing above a moving average by a certain percentage generates a buy. Regardless of method, the goal remains identical: identify direction, enter with momentum, exit when momentum reverses. Backtesting across decades shows that simple 200-day moving average rules on major indices reduce maximum drawdowns by 50-70% compared to buy-and-hold, albeit with occasional whipsaws.
Designing a Simple ETF Trend-Following System
A practical, long-term trend-following system requires just four components: a universe of ETFs, a trend filter, a ranking mechanism, and a rebalancing schedule. Start with a universe of 10-20 highly liquid ETFs spanning asset classes: U.S. large-cap (SPY), U.S. small-cap (IWM), developed international (EFA), emerging markets (EEM), long-term Treasuries (TLT), intermediate Treasuries (IEF), gold (GLD), commodities (DBC), real estate (VNQ), and cash equivalents (BIL). The trend filter could be as simple as: only hold ETFs trading above their 200-day moving average. For ranking, select the top three to five ETFs with the highest 12-month total return, or highest 6-month return, or lowest volatility-adjusted momentum. Rebalance monthly or quarterly—monthly captures trends faster but incurs more transaction costs and taxes. A monthly rebalance on the last trading day works well. Position sizing can be equal-weight among selected ETFs. If fewer than three ETFs pass the trend filter, allocate the remainder to cash or short-term bonds. This system is fully mechanical, eliminates discretion, and requires only 15 minutes per month to execute.
The Critical Role of Absolute Momentum Versus Relative Momentum
Trend following incorporates two distinct momentum concepts. Absolute momentum (time-series momentum) asks: is this asset’s price higher than it was N months ago? If yes, hold; if no, avoid. Relative momentum (cross-sectional momentum) asks: which assets have performed best relative to others? Combining both creates a powerful filter. Absolute momentum protects against prolonged bear markets—when all assets decline, you move to cash. Relative momentum ensures you concentrate on the strongest trends. Academic research by Moskowitz, Ooi, and Pedersen (2012) demonstrated that time-series momentum works across 58 futures markets and multiple asset classes, with Sharpe ratios significantly higher than buy-and-hold. Gary Antonacci’s “Dual Momentum Investing” popularized combining absolute and relative momentum for ETFs. His research shows that a dual momentum system using U.S. equities, international equities, and aggregate bonds produced higher returns with lower drawdowns than traditional 60/40 portfolios over 40+ years. For ETF investors, absolute momentum is easily calculated: compare current price to the 12-month moving average or the price 12 months ago. Relative momentum: rank by 12-month return excluding the most recent month (to avoid short-term reversal).
Selecting the Right Moving Average Length for ETFs
The choice of moving average length dramatically affects performance. Short-term averages (10-50 days) generate frequent signals, capturing quick trends but suffering from whipsaws in choppy markets. Long-term averages (200-400 days) produce fewer signals, miss early trend entries, but avoid noise. Research by Meb Faber in “A Quantitative Approach to Tactical Asset Allocation” tested 10-month simple moving averages on U.S. equities from 1900 to 2012. The 10-month SMA (approximately 200-day) reduced maximum drawdown from 83% to 50% while slightly outperforming buy-and-hold on a risk-adjusted basis. For ETFs, the 200-day moving average is the most widely followed, creating a self-fulfilling prophecy as thousands of traders watch the same level. A dual moving average system—using both 50-day and 200-day—reduces false signals: only go long when 50-day > 200-day, and exit when 50-day < 200-day. Alternatively, use a 12-month rate-of-change: hold if positive, exit if negative. Backtests on SPY from 1993 to 2023 show that a 12-month ROC filter avoided the 2000-2002 and 2008-2009 bear markets almost entirely, with only minor whipsaws in 2015 and 2018. The optimal length depends on your holding period and tolerance for drawdowns.
Position Sizing and Risk Management With ETFs
Even the best trend signal fails without proper position sizing. Fixed fractional sizing allocates a constant percentage of capital per ETF—for example, 20% each across five ETFs. Volatility-adjusted sizing scales positions inversely to recent volatility: calculate the 20-day average true range (ATR) or standard deviation, then size so that each position contributes equal risk. A common formula: position size = (portfolio risk per trade) / (ATR × contract multiplier). For ETFs, simpler: divide target risk (e.g., 1% of portfolio) by the ETF’s 20-day standard deviation. This prevents a volatile emerging market ETF from dominating risk. The “heat” concept limits total portfolio risk: never risk more than 6% of capital across all open positions. Stop-losses can be based on ATR multiples (e.g., 3× ATR trailing stop) or moving averages. A trailing stop below the 200-day moving average on a weekly closing basis works well for long-term ETFs. Crucially, trend following accepts many small losses to capture a few huge gains. The win rate often falls below 50%, but the average win is 3-5× the average loss. This positive skewness demands emotional discipline: you must take every signal, even after five consecutive losses.
Backtesting and Walk-Forward Analysis for ETF Systems
Before risking capital, backtest your ETF trend-following rules on historical data. Free platforms like Portfolio Visualizer, ETF Replay, and QuantConnect allow testing. Key metrics to examine: compound annual growth rate (CAGR), maximum drawdown, Sharpe ratio, Sortino ratio, win rate, profit factor, and average trade duration. For example, a simple 200-day moving average system on SPY from 2000 to 2023 produced approximately 8.2% CAGR versus 6.8% for buy-and-hold, with maximum drawdown of 28% versus 55%. Adding international and bond ETFs improved diversification. However, backtests suffer from overfitting—tuning parameters to past data that fail in the future. Walk-forward analysis mitigates this: divide history into in-sample (optimization) and out-of-sample (validation) periods. Optimize on 2000-2010, test on 2011-2015, re-optimize on 2006-2015, test on 2016-2020. Only accept parameters that work across multiple out-of-sample periods. Monte Carlo simulation randomizes trade order to estimate worst-case drawdowns. A robust ETF trend system should survive 1,000 simulations with 95% confidence of not exceeding a 40% drawdown. Avoid systems with more than three parameters—complexity breeds fragility.
Tax Efficiency and Transaction Cost Considerations
Trend following in taxable accounts creates short-term capital gains, taxed at ordinary income rates up to 37% in the U.S. This drags returns significantly. Mitigations exist. First, hold signals for at least 12 months when possible—use a 12-month momentum filter rather than 6-month. Second, locate trend-following ETFs in tax-advantaged accounts (IRAs, 401ks) while keeping buy-and-hold ETFs in taxable accounts. Third, use ETFs with low bid-ask spreads—SPY, IVV, VTI, EFA, and AGG have penny spreads. Fourth, avoid leveraged and inverse ETFs, which have high internal costs and decay. Fifth, consider futures-based ETFs like DBMF or KMLM for trend following in taxable accounts, as futures receive 60/40 tax treatment (60% long-term, 40% short-term) regardless of holding period. Transaction costs matter: a monthly rebalance across five ETFs might incur 0.01-0.05% per trade. Over a year, that’s 0.6-3.0% drag. Quarterly rebalancing cuts costs by two-thirds. Use limit orders, not market orders. For large portfolios, negotiate commissions. Finally, wash sale rules apply: if you sell an ETF at a loss and buy a substantially identical ETF within 30 days, the loss is disallowed. Use a different but correlated ETF to maintain exposure.
Common Behavioral Pitfalls and How to Overcome Them
Trend following fails for most investors not because the strategy is flawed but because human psychology sabotages execution. The first pitfall is recency bias: after a strong bull market, investors abandon trend signals and hold through the eventual decline, believing “this time is different.” The second is loss aversion: taking profits too early on winning trends while letting losers run, the exact opposite of trend following. The third is signal fatigue: after three or four whipsaws in a row, investors skip the next signal—which often turns out to be the big winner. The fourth is over-optimization: constantly tweaking moving average lengths or rebalancing dates. The fifth is leverage creep: adding margin after a winning streak, then suffering catastrophic losses. Solutions include: automating signals with spreadsheet alerts or trading platforms; writing an investment policy statement that mandates following every signal; setting a fixed rebalancing date and sticking to it; never risking more than 1% per trade; and reviewing performance only quarterly, not daily. Studies by Dalbar show that the average investor underperforms the funds they invest in by 3-4% annually due to poor timing. Trend following, executed mechanically, eliminates that gap.
Combining Trend Following With Core Buy-and-Hold
Many long-term investors adopt a hybrid approach: a core portfolio of broad-market ETFs held permanently, plus a satellite trend-following overlay. For example, 70% in VTI (total U.S. stock) and VXUS (total international) buy-and-hold, 30% in a trend-following ETF rotation system. This captures long-term equity risk premium while reducing drawdowns. Rebalance annually. Backtests from 1970 to 2023 show that a 70/30 core-satellite with trend following on the 30% produced nearly identical returns to 100% buy-and-hold but with 35% lower maximum drawdown. Another approach: apply trend following only to asset classes with strong momentum, such as commodities (PDBC), real estate (VNQ), and emerging markets (EEM), while holding U.S. large-cap permanently. The key is recognizing that trend following is not a standalone religion but a risk-management tool. It underperforms in sharp V-shaped recoveries (like March 2020) because signals lag. It outperforms in prolonged bear markets (2000-2002, 2008-2009, 2022). Over a full market cycle, the risk-adjusted returns typically exceed buy-and-hold. For retirees in withdrawal phase, trend following is especially valuable because sequence-of-returns risk—the danger of retiring just before a bear market—is dramatically reduced.
Specific ETF Recommendations for Trend Following
Not all ETFs suit trend following equally. The ideal ETF has high liquidity (average daily volume > 1 million shares), tight spreads (< 0.05%), low expense ratio (< 0.30%), and exposure to a distinct asset class or factor. For U.S. equity trends: SPY, IVV, VOO, QQQ (Nasdaq-100, higher beta), IWM (small-cap), and RSP (equal-weight S&P 500). For international: EFA (developed), EEM (emerging), VEA, VWO, and INDA (India). For fixed income: TLT (20+ year Treasuries), IEF (7-10 year), LQD (investment-grade corporates), HYG (high yield), and TIP (inflation-protected). For commodities: GLD (gold), SLV (silver), USO (oil), UNG (natural gas), and DBC (broad commodities). For real estate: VNQ, IYR, and REM (mortgage REITs). For currencies: UUP (U.S. dollar index) and FXY (Japanese yen). For alternatives: DBMF (managed futures), KMLM (KFA Mount Lucas), and CTA (Simplify managed futures). Avoid sector-specific ETFs like XLE (energy) or XLF (financials) for core trend following because they are too narrow and prone to violent reversals. Avoid leveraged ETFs (TQQQ, SPXL) because daily rebalancing causes volatility decay. Avoid inverse ETFs (SH, SDS) because they lose money in flat markets. Stick to broad, liquid, low-cost ETFs with decades of history if possible.
Monitoring and Adjusting Your ETF Trend System
Once implemented, a trend-following ETF system requires minimal but consistent monitoring. Monthly tasks: download month-end adjusted closing prices for your universe; calculate 200-day moving averages and 12-month returns; identify which ETFs pass both absolute and relative momentum filters; rank the top three to five; compare current holdings to target holdings; generate buy and sell orders for the next trading day. Quarterly tasks: review transaction costs and slippage; check that expense ratios remain competitive; verify that no ETF has changed its index or strategy (style drift). Annual tasks: run a full backtest with updated data; calculate portfolio-level Sharpe ratio, maximum drawdown, and win rate; compare to a benchmark like 60% SPY / 40% AGG; decide if parameter changes are warranted—but only if out-of-sample performance has degraded for two consecutive years. Never change rules mid-year. Keep a trading journal: date, signal, action, price, and emotional state. Review the journal after one year to identify behavioral patterns. If you missed signals or overrode the system, address the root cause—often it’s position size too large or unrealistic expectations.
Historical Performance Evidence Across Decades
The empirical case for ETF trend following rests on decades of out-of-sample evidence. Meb Faber’s 2013 study “A Quantitative Approach to Tactical Asset Allocation” tested a 10-month moving average on U.S. equities, foreign equities, 10-year Treasuries, commodities, and REITs from 1973 to 2012. The trend-following portfolio returned 10.5% annually versus 9.7% for buy-and-hold, with maximum drawdown of 18% versus 45%. The Sharpe ratio improved from 0.42 to 0.65. AQR’s 2014 paper “A Century of Evidence on Trend-Following Investing” examined 100+ years across 67 markets. Trend following delivered positive returns in every decade, including the 1930s and 2000s when stocks lost money. During the 2008 financial crisis, a simple 200-day moving average on SPY exited near 1,400 in January 2008 and re-entered near 900 in June 2009, avoiding a 35% drawdown. During the 2020 COVID crash, the signal exited near 3,000 in March 2020 and re-entered near 3,200 in June 2020, missing the bottom but avoiding the worst. During 2022, trend following on TLT and DBC produced positive returns while SPY fell 18%. No strategy wins every year—trend following lost 3% in 2015 and 2% in 2018—but over rolling 10-year periods, it outperformed buy-and-hold on a risk-adjusted basis in over 85% of historical windows.
Frequently Asked Questions About ETF Trend Following
How much capital do I need? With commission-free brokers like Schwab, Fidelity, and Vanguard, you can start with $1,000. Fractional shares allow precise allocations. What if I miss a signal? Set calendar alerts for the last trading day of each month. Execute the next morning at market-on-open. Never chase a missed signal—wait for the next month. Can I use this in a 401k? Most 401ks lack ETFs, but many have index mutual funds that track the same indices. Apply the same moving average rules to those funds. What about dividends? Use total return (price plus dividends) for signals. Most data providers adjust for dividends automatically. How often does it trade? A 200-day moving average on five ETFs trades 2-6 times per year per ETF. That’s 10-30 trades annually. Very manageable. What’s the worst-case scenario? A whipsaw market that chops sideways for years—like 2011, 2015, or 2018. You might underperform buy-and-hold by 5-10% during those periods. But you avoid 40%+ bear markets, which compounds wealth more reliably. Can I short ETFs? Yes, but shorting has unlimited risk and borrow costs. Better to move to cash or use inverse ETFs tactically. Should I use stop-loss orders? Trailing stop-losses based on ATR work, but moving average exits are simpler and less prone to intraday noise. What taxes will I pay? Short-term gains on trades held less than a year. Use tax-advantaged accounts or accept the tax drag as the cost of downside protection. How do I know if the strategy stops working? If out-of-sample performance lags buy-and-hold for 5+ years with higher drawdowns, the edge may have decayed. But 200+ years of data across multiple asset classes suggest trend following remains robust.
