Amber Quant Lab research
Overview / Strategies / Trend following
Baseline · family trend

Trend following

Baseline  A well-known reference strategy. Not a candidate: it sets the bar new ideas must clear.

Validation Sharpe0.46train 0.73
Validation CAGR4.3%train 6.4%
Max drawdown-15.0%train -10.7%
Correlation to 60/400.49validation
Deflated Sharpe76%discounted for 2 trials
PerformanceHow it works

Performance

Backtest after costs. Train is where the idea was developed; validation is where it was judged.

Growth of $1Trend following60/40SPY
trainvalidation$0.5$1$1.5$2.5$42008201020122014201620182020Trend followingTrend following $2.2460/4060/40 $3.49SPYSPY $4.51
Drawdown from peakTrend following60/40
-30%-20%-10%0%Trend following-15% (Mar 2019)60/402008201020122014201620182020
Calendar-year returnsTrend following60/40
-20%0%20%2007 Trend following: +15.3%2007 60/40: +7.5%’072008 Trend following: +2.7%2008 60/40: -17.9%’082009 Trend following: +1.4%2009 60/40: +12.8%’092010 Trend following: +9.3%2010 60/40: +13.5%’102011 Trend following: +13.5%2011 60/40: +7.7%’112012 Trend following: +9.1%2012 60/40: +11.2%’122013 Trend following: +2.4%2013 60/40: +15.6%’132014 Trend following: +4.2%2014 60/40: +11.8%’142015 Trend following: +0.2%2015 60/40: +1.7%’152016 Trend following: +0.5%2016 60/40: +7.7%’162017 Trend following: +15.5%2017 60/40: +13.7%’172018 Trend following: -11.9%2018 60/40: -2.0%’182019 Trend following: +13.4%2019 60/40: +22.0%’192020 Trend following: +6.1%2020 60/40: +16.4%’202021 Trend following: +4.5%2021 60/40: +15.0%’21
MetricTrain 2007–15Validation 2016–21
CAGR6.4%4.3%
Volatility8.0%7.8%
Sharpe0.730.46
Sortino0.970.56
Max drawdown-10.7%-15.0%
Calmar0.600.29
Worst month-4.0%-5.3%
Months up60%62%
Turnover / yr4.6×5.1×
Avg gross exposure1.28×1.36×
Correlation to SPY0.090.37
Correlation to 60/400.220.49
Beta to SPY0.030.16

Average holdings

LQDUS investment-grade corporates25.5%
TIPUS inflation-protected Treasuries23.4%
IEFUS Treasuries 7–10y21.9%
SPYUS large-cap equities13.0%
UUPUS dollar index (long USD)11.5%
TLTUS Treasuries 20y+9.3%
EFADeveloped ex-US equities8.4%
GLDGold7.0%
EEMEmerging-market equities5.9%
DBCBroad commodities5.3%

How it works

Time-series momentum across asset classes: own each asset when its own past return beats cash (short it when allowed and it doesn't), size each position by inverse volatility, and scale the whole book to a volatility target. Monthly.

Why it should work. Investors under-react to news and then herd, so price trends persist for months (Moskowitz, Ooi & Pedersen 2012). The core of managed-futures funds such as Man AHL, Winton and AQR's trend funds; historically strongest in prolonged selloffs (2008, 2022).