Adaptive Momentum

This strategy combines relative strength momentum (choosing the best performing asset within the peer group) and pairs it with traditional trend following which is used to identify overall market trend.

Whereas the original Core Momentum strategy is a more rigid “on-off” type system, Adaptive Momentum is more flexible by choosing between the best ETF within a group and utilizing defensive ETFs if the trend is positive but the offensive ETFs are underperforming.

We use leveraged ETFs where possible:

  • US Equities: 3x large cap tech (TQQQ) and 3x small cap (TNA)

  • Foreign Equities: 2x international (EFO) and 3x emerging markets (EDC)

  • Alternative Assets: commodities (DBC) and 3x US real estate (DRN)

  • US Bonds: 3x intermediate treasuries (TYD) and 3x long term treasuries (TMF)

  • Cash: short term treasuries (BIL)

The logic

First, rank each asset from best to worst using a multi-period momentum score

If the overall market trend is down, buy the best defensive asset (TYD or BIL)

If the overall market trend is up, buy the top 4 offensive assets (TQQQ, TNA, EFO, EDC, DBC, DRN, TMF, TYD)

If the overall market trend is up but one of the top 4 offensive assets has negative momentum, allocate that portion to cash instead.

Backtested Performance

Adaptive Momentum is more risky than the S&P 500. Typically more risk = more reward so it’s not really fair to compare against the S&P.

For an apples-to-apples comparison, we use a portfolio that’s roughly 50/50 UPRO (3x S&P 500) and cash rebalanced every 5% deviation.

Key points:

  • 2.5x higher annual growth rate

  • Half the drawdown of the benchmark portfolio and the same drawdown as the general market

  • 0.45 market correlation

  • 3x higher Sortino Ratio (risk-adjusted return)

Annual Returns

There are some losing years and periods of underperformance but most importantly there are no catastrophic down years like the benchmark portfolio experiences.

Drawdowns

Some thoughts:

  • Losing 40% in 2008 is uncomfortable but less than what the general market did (let alone the benchmark portfolio).

  • Other drawdowns (2011-2018) are pretty in line with the benchmark.

  • The March 2020 Covid drawdown was much better

  • Flat for most of 2022 but a little slow to pick up the rally from October

Rolling Returns

Five year rolling returns are consistently between 10 and 40%.

Either much higher than the benchmark during recessions (2008, 2022) or in line with it (2014-2020).

How it works

At the end of every month, the algorithm determines the number of positions and position size (typically between 1-4).

During bullish months, allocations are typically equally split between four risk assets. During bearish months, it concentrates more in bonds (TYD, TMF) and cash (BIL).

See example below (notice how it goes from mostly bearish in 2008 to mostly bullish in 2009):

Final thoughts

Like all investing strategies, there are pros and cons with Adaptive Momentum.

Pros:

  • Excellent risk-adjusted returns

  • Not very correlated with the market (0.45 vs 1.00)

  • Risky enough to where a small allocation contributes meaningfully to overall portfolio performance

Cons:

  • Goes through periods of relative underperformance while the market trends higher

  • Quite risky, better to use as a small part of portfolio

  • No guarantee it works in the future

  • Kind of a pain buying up to 4 ETFs - made much easier if using something like M1 Finance though

The core tenet of Modern Portfolio Theory is buying uncorrelated return streams and rebalancing occasionally. While we used to be 100% trend advocates, it’s psychologically difficult to stay committed to because of tracking error. “Perfect is the enemy of good enough”, etc.

No one really cares too much if their portfolio is up when the market is down, but if you’re down while the market’s up (especially for a long time!), that’s painful. You’ll likely have FOMO (fear of missing out) and jump ship at the worst time. Better to have a mix of good strategies (including buy and hold) and rebalance occasionally.

For more info, you can check out this PortfolioVisualizer report of Adaptive Momentum here.