What Is Trend Following?
Trend following is the oldest and most extensively documented systematic investment strategy in existence. The premise is simple: prices trend, trends persist, and systematic rules can capture those trends while avoiding the worst declines. It requires no forecasting, no fundamental analysis, no opinions about where markets should be. It asks only one question: is the price going up, or is it going down?
This simplicity is deceptive. Behind the straightforward concept lies a strategy that has been rigorously tested across more than a century of data, across dozens of markets, and in live trading by some of the most successful investment firms in history — including commodity trading advisors (CTAs) who manage hundreds of billions of dollars using trend-following principles.
The Core Concept
Trend following rests on a single empirical observation: asset prices do not move randomly. They exhibit persistent directional movements — trends — that last weeks, months, or years. These trends are caused by the gradual incorporation of information into prices, by the self-reinforcing dynamics of capital flows, and by the behavioral tendency of market participants to underreact to new information and then overreact once a trend is established.
A trend-following system does not predict where prices will go. It identifies where prices are going — right now — and positions accordingly. If the current trend is up, the system holds the asset. If the current trend is down, the system exits to a safe-haven position. The assumption is not that the trend will continue forever, but that it has a better-than-random chance of continuing in the near term — and that the payoff from riding trends is larger than the cost of the occasional false signal.
The Academic Evidence
The evidence for trend following is extraordinary in both depth and breadth.
Moskowitz, Ooi, and Pedersen (2012) published "Time Series Momentum" in the Journal of Financial Economics, documenting significant momentum effects in 58 liquid futures markets across equities, bonds, commodities, and currencies. They found that assets with positive returns over the past 12 months continued to outperform, and assets with negative returns continued to underperform. The effect was statistically significant, economically large, and present in every asset class studied.
Hurst, Ooi, and Pedersen (2017) extended this analysis to "A Century of Evidence on Trend-Following Investing," using data from 1880 to 2016. Their findings were striking: trend following has been profitable in every decade for over 130 years. It performed best during the most extreme market environments — the worst equity drawdowns, the largest bond selloffs, the most volatile commodity markets. When other strategies failed, trend following delivered its strongest returns.
This is not a data-mined artifact. The strategy's profitability across 130 years of data, across dozens of markets, and in live institutional trading makes it one of the most robust findings in all of quantitative finance.
How Moving Averages Detect Trends
The most common implementation of trend following uses moving averages — a smoothed representation of an asset's recent price history. As we detail in our comparison of SMA vs. EMA moving averages, the choice of moving average type affects signal timing and sensitivity, but the core logic is the same.
The Simple Moving Average (SMA) Signal
The classic trend-following rule: if the asset's price is above its N-month simple moving average, hold the asset. If below, move to a defensive position (typically Treasury bills).
The most studied lookback period is 10 months (approximately 200 trading days). Meb Faber's widely cited research, discussed in our overview of the Faber GTAA strategy, demonstrated that this single rule, applied across five asset classes, delivered equity-like returns with dramatically reduced drawdowns from 1973 to 2012.
Why 10 months? The signal captures intermediate-term trends that correspond to economic and business cycle dynamics. Shorter periods (3–5 months) are more responsive but generate more false signals. Longer periods (15–20 months) are smoother but slower to react to genuine trend changes. The 10-month SMA sits in a sweet spot that balances responsiveness against noise reduction.
The Exponential Moving Average (EMA) Signal
Exponential moving averages place more weight on recent prices, making them more responsive to trend changes. An EMA trend signal will react faster to a reversal than an SMA of the same lookback period. This faster reaction reduces the lag at trend turning points but increases the frequency of whipsaw signals during choppy markets.
Dual Moving Average Crossovers
Instead of comparing price to a single moving average, some systems use two moving averages of different lengths. When the shorter-period average crosses above the longer-period average, a buy signal is generated. When it crosses below, a sell signal triggers. This approach smooths out noise more effectively than the single-average method but adds additional lag.
Why Trends Persist
The persistence of trends is not accidental. Multiple reinforcing mechanisms create and sustain directional price movements.
Behavioral underreaction. When new information arrives — an earnings surprise, a policy shift, a geopolitical development — investors tend to underreact initially. They anchor to their prior estimates and adjust slowly. This creates a drift in the direction of the new information that unfolds over weeks and months. The drift is the trend.
Institutional capital flows. Large institutional investors move capital slowly. A pension fund that decides to increase its equity allocation does not execute in a day — it phases in purchases over weeks or months. This sustained buying pressure creates and extends trends. Similarly, forced liquidations during crises (margin calls, fund redemptions) create sustained selling pressure that extends downtrends.
Central bank policy transmission. Monetary policy operates with long and variable lags. The effects of interest rate changes take 12 to 18 months to fully transmit through the economy and financial markets. This slow transmission creates persistent trends in rate-sensitive assets — bonds, REITs, gold — that trend-following systems are well-positioned to capture.
Feedback loops. Rising prices attract attention, media coverage, and new investors. The additional capital inflows push prices higher, attracting more attention. This self-reinforcing cycle sustains trends beyond what fundamentals alone would justify — creating the "bubble" phase that is a regular feature of financial markets. Trend following captures the profitable phase of these cycles and exits when the trend reverses.
What Trend Following Cannot Do
Intellectual honesty requires acknowledging the strategy's limitations.
Flash crashes and overnight gaps. Trend following uses end-of-period prices. If a market gaps down 10% overnight due to a sudden event, the trend-following system will register the damage at the next signal evaluation, not prevent it. The strategy protects against sustained declines, not instantaneous shocks.
Whipsaw markets. When prices oscillate around the moving average without establishing a clear trend, the system generates repeated buy and sell signals that produce small losses. These whipsaw periods — typically during trendless, range-bound markets — are the primary cost of trend following. Historically, whipsaw costs average 1–3% per year during non-trending periods.
Exact timing. Trend following always enters late and exits late. It will never buy the exact bottom or sell the exact top. It sacrifices precision at turning points in exchange for capturing the body of the trend — which is where the bulk of the returns reside.
| Trend Following Strength | Trend Following Limitation |
|---|---|
| Avoids sustained bear markets | Cannot prevent flash crash losses |
| Captures majority of bull market gains | Enters after bottom, exits after top |
| Works across all asset classes | Suffers during choppy, trendless markets |
| No forecasting required | Requires discipline to follow signals |
| 130+ years of positive evidence | Periods of underperformance vs. buy-and-hold |
Trend Following Across Strategy Types
On PortfolioWiser, trend following is not a single strategy — it is a component that can be applied across the entire platform. As explained in our overview of tactical asset allocation, trend detection serves as either the primary signal or a secondary filter in the majority of tactical strategies.
As a primary signal: Pure trend-following strategies like the Faber GTAA use moving averages as the sole decision mechanism. Each asset is independently evaluated against its own trend. No cross-asset comparison is performed — the signal is purely absolute.
As a secondary filter (trend health): Many strategies use momentum to rank and select assets, then apply a trend-following filter to the selected assets. If the top-ranked asset is trending negatively (below its moving average), it is rejected in favor of a defensive asset. This layered approach combines the selection power of momentum with the protective power of trend following.
The Strategy Builder on PortfolioWiser exposes this distinction explicitly. You can select the trend detection method (SMA, EMA, or composite), the lookback period, and whether trend health operates as a pre-filter (applied before ranking) or a post-filter (applied after ranking). These configuration options let you test how different trend-following implementations affect both returns and drawdowns for any strategy universe.
Trend following is not glamorous. It does not make bold predictions. It does not catch tops and bottoms. It simply asks whether an asset is going up or down, and positions accordingly. This simplicity is its greatest strength — it is robust precisely because it does not depend on complex models, proprietary data, or accurate forecasts. It depends only on the single most reliable phenomenon in financial markets: trends persist.