The idea
Time series momentum asks one question per market, once a month: is the price higher than it was a year ago? If yes, own it. If no, stay out, or sell it short. It ignores every other piece of information.
It is the slowest strategy in this library and the one with the most research behind it. Funds that follow trends across dozens of futures markets run versions of exactly this rule.
Watch it play out
Step by step
The rules
Setup
- Once a month, compare each market's price with its price 12 months earlier.
Entry
- Hold (or buy) the markets whose 12-month return is positive.
Stop-loss
- No price stop: the monthly check is the exit.
Target
- Step aside, or go short, when the 12-month return turns negative.
Filters that help
- Spread it over many unrelated markets.
- Size each market so it contributes similar risk: more of a calm market, less of a wild one.
Why it can work
Markets absorb big changes slowly. When the economy, interest rates or supply shift, prices tend to move in the same direction for months as investors adjust, which leaves a trend that a simple 12-month rule can follow.
Its biggest benefit is in long bear markets: the rule steps aside after the first leg down and avoids much of the rest, at the cost of missing the start of the recovery.
Worked example
At the end of a month a stock index stands at 4,800, against 4,500 a year earlier: the 12-month return is positive, so you hold it for another month.
Some months later it stands at 4,100, against 4,400 a year before. The return has turned negative, so you sell and wait in cash until the comparison turns positive again.
When it works, when it fails
Works best
In long, persistent moves in either direction: bear markets that unfold over many months, commodity booms, rate cycles. It has tended to do well in the worst years for stocks.
Struggles
In sharp V-shaped turns: it sells after a fall that reverses quickly and buys back higher. In sideways years it switches in and out at small losses.
Common mistakes
- Checking every day instead of once a month, which turns a slow rule into a noisy one.
- Running it on a single market, where one whipsaw year can feel like proof it is broken.
- Abandoning it after missing a rebound, just before the next long trend.
Variations
- Dual momentum
- Hold the stronger of two markets, and only if its 12-month return beats cash.
- Moving average version
- Hold when price is above its 10-month (about 200-day) average: a very similar signal.
What the evidence says
Evidence: ResearchedPublished studies exist. Results depend on the details and on costs.
A 2012 study across 58 futures and forward markets found that the past 12 months' return predicted the next month's in every asset class, and that a diversified strategy built on it made money with little correlation to stocks.
A later study extended the test back to 1880 across 67 markets and found trend following positive in every decade. Returns have been lower in the 2010s than before, and results depend heavily on diversification and costs.
Questions
- Does time series momentum work on stock indices alone?
- The research covers stocks, bonds, currencies and commodities together. On a single stock index the rule has helped most in long bear markets and cost a little in choppy years, so its value there comes mainly from avoiding the deepest falls rather than from adding return.
- What is time series momentum?
- A rule that holds a market when its own return over the past year is positive and exits or shorts it when negative. It compares the market with its own past, unlike cross-sectional momentum, which ranks markets against each other.
- Why 12 months?
- The research found the strongest effect around a 12-month look-back. Many funds blend 1, 3 and 12 months to smooth the switches.
- Can I use this on a single stock or coin?
- The research covers diversified futures markets. On one market the idea still applies, but the results swing far more, and long losing stretches are likely.