TenachineAI-Powered BacktestingSign in
Guide · 5 min read

Dual momentum trades about once a year, and its median ticker still lost money

Dual momentum 12 minus 1 trades about once a year per ticker, keeps one of the smallest drawdowns and tightest tails on this site, and its win rate and risk to reward ratio both look healthy. The median ticker still finished the decade slightly underwater.

Two hundred and fourteen tickers completed a run of Tenachine's dual momentum 12 minus 1 study between January 2015 and December 2024, and between them they generated 2,125 trades, 9.9 per ticker across the whole decade. That is roughly one new position a year per stock, the lowest turnover of any strategy on this site with a full return distribution behind it. A strategy that trades this rarely should be easy to sum up in one number. It is not.

The average return across those 214 tickers is positive 0.41 percent. The median ticker, the one sitting exactly in the middle once every result is sorted, lost 0.05 percent. Both numbers are correct, and the gap between them is not the wide kind. Tenachine's guide on triple ma pullback covers a strategy where the mean and median land 1.18 percentage points apart on opposite sides of zero. Here they are 0.46 points apart, and the sign still flips. A strategy this quiet can tip either way on almost nothing.

Dual momentum 12 minus 1, 214 completed tickers, by return bucket
Dual momentum 12 minus 1, 214 completed tickers, by return bucketShare of 214 completed tickers by total return bucket: 0.0 percent lost more than 10 percent, 1.9 percent lost 5 to 10 percent, 32.2 percent lost 1 to 5 percent, 29.0 percent finished within 1 percent of flat, 29.9 percent gained 1 to 5 percent, 6.5 percent gained 5 to 10 percent, and 0.5 percent gained more than 10 percent.Lost more than 10%0.0%Lost 5% to 10%1.9%Lost 1% to 5%32.2%Flat, within 1%29.0%Gained 1% to 5%29.9%Gained 5% to 10%6.5%Gained more than 10%0.5%
Not one of the 214 tickers lost more than 10 percent, and only one gained more than 10 percent. That is a tighter cluster near zero than every strategy on this site except rsi mean reversion and nr7 volatility breakout, the two strategies ahead of dual momentum on both average drawdown and maximum single-ticker gain.

Sixty two of the 214 tickers, 29.0 percent, finished the decade within 1 percent of flat in either direction, the single largest bucket in the study. The median ticker's negative 0.05 percent return sits inside that bucket, a hair's width from the positive side of the same range. Move one ticker from the losing half of that bucket to the winning half and the median flips positive without the average moving by any amount a reader would notice. That is a different kind of fragility than a fat tail pulling a headline number away from a typical result. Here the typical result and the near typical results are so close together that the sign is close to a coin flip.

Two other numbers make this strategy an outlier in the other direction. Its average maximum drawdown, negative 2.78 percent, is the third smallest of the twenty two strategies published here, deeper only than rsi mean reversion's negative 2.68 percent and nr7 volatility breakout's negative 0.38 percent. Tenachine's guide on nr7's drawdown covers that smallest case, also a loss at the median. Dual momentum's largest single ticker gain, positive 11.53 percent, is the third smallest maximum of any strategy here for the same reason: a rule built to sit out most of the time and exit within a month of losing its edge does not give any one ticker the room to run the way a breakout strategy does.

The trade level numbers do not point to an obviously weak rule either. The win rate is 45.68 percent, higher than all but a handful of the other strategies on this site, and the aggregate risk to reward ratio is 1.14, trades that win pay back fourteen percent more than trades that lose cost. A win rate above 45 percent paired with a risk to reward ratio above 1.0 usually describes a strategy with a real edge on paper. Here it describes one that still lands on the wrong side of zero at the median, because the edge on each of those infrequent trades is thin enough that ten years of it barely moves the number either way.

The rule itself explains why trades are this rare. It scores each stock by its 12 month return minus its most recent 1 month return, the classic academic momentum measure built to ride a trend while skipping the short term reversal that often follows a big move. It only opens a new position once every 30 trading days, and only if that score is positive and above its own 21 day average. Once a month it checks again and exits if the score turns negative, falls below its own average, or the position has been held 250 trading days. Every stop sits 2 average true ranges below entry, sized so a stop out costs exactly 1 percent of account equity. Tenachine turns a written rule like that into the exact code that produced these 2,125 trades and the distribution above, which is also why the study can publish 9.9 trades a ticker across a full decade rather than one illustrative example.

A low turnover, small drawdown, small tail strategy is not automatically safe at the level that matters to a single reader holding one ticker, the median outcome. This is not a signal to trade on its own, and ten quiet years are not a forecast of the next ten. Check where the median sits relative to zero before treating a positive average as proof that a typical position made money, especially when the distribution is compressed this tightly around it.