Plain Momentum (Rate of Change): The Boring Indicator That Keeps Ranking Well

Why You've Probably Dismissed ROC

Rate of Change is about as simple as a technical indicator gets: it measures how much price has moved over a fixed lookback period, expressed as a percentage. No adaptive smoothing, no proprietary weighting, no black box. Because it looks too elementary, most traders reach instead for something that sounds more sophisticated—multi-layered oscillators, smart money concepts, confluence systems. That instinct turns out to be expensive.

IndicatorEdge ran 660,005 out-of-sample backtests across 903 assets and 382 indicators, covering 1-Hour, 4-Hour, Daily, and Weekly timeframes. ROC was part of that test set, and it kept showing up in the results in a way that many of the fancier alternatives did not.

What the Data Actually Shows

Among all the indicator-asset combinations tested, only 26% beat buy-and-hold after realistic transaction costs. That is the baseline you are competing against. ROC (30) ranked in the top performers for the Index asset class—placing second in that category behind DeMarker. For a plain percentage-change calculation with a single parameter, that result is harder to dismiss than most traders expect.

Across the full 903-asset universe, 63% of assets had at least one indicator that beat buy-and-hold, with a median best Sharpe of 0.62. ROC belongs to the minority of indicators that actually show up in that winning column rather than generating impressive-looking backtests that collapse under cost pressure.

The Complexity Trap: High Win Rate, Low Real Edge

The data exposes a pattern worth understanding before you reach for anything more elaborate than ROC. Several heavily-marketed indicators carry win rates above 70%—Murrey Math Lines at 74.3%, Holy Grail Confluence at 73.3%, Ultimate Oscillator at 72.7%—yet each of them beat buy-and-hold on fewer than 11% of assets tested. A high win rate does not mean an indicator actually makes money after costs; it often just means entries cluster near price reversions that would have recovered anyway.

Smart money concept indicators showed the same problem: not a single SMC-based system in the test set consistently beat buy-and-hold across the asset universe. Meanwhile, a 30-period rate-of-change calculation—no secrets, no subscription required—placed in the Index category top five. Simplicity does not guarantee an edge, but complexity clearly does not guarantee one either.

Where ROC Fits and Where It Doesn't

ROC's Index-class performance makes intuitive sense. Broad indices tend to exhibit sustained momentum phases driven by macro flows rather than microstructure noise. A 30-period lookback is long enough to filter short-term chop while remaining sensitive enough to catch meaningful trend acceleration. That environment suits a plain momentum measure well.

The picture is different elsewhere. In the Forex category, Fisher Transform dominated with 17 asset wins. In Crypto, MA Envelope led with 5. In Commodities, Keltner Mean-Reversion topped the table. No single indicator wins everywhere, and ROC is no exception. The honest takeaway: ROC earns serious consideration for index-oriented strategies; for other asset classes, check the asset-specific rankings rather than applying it by default.

These results are hypothetical backtests, not a promise of future returns and not financial advice. Backtests reflect past price behaviour under the tested conditions; real trading involves execution slippage, position sizing, and market regimes that differ from history. Use this data as a filter for what is worth investigating, not as a guarantee of what will work.

How to Actually Use This

Start with the question the backtest answers: does ROC (30) appear in the top results for the specific asset you trade? If you are working with an equity index, the data suggests it is worth serious testing. If you are in Forex or Crypto, the asset-level rankings will point you toward what actually led for that class.

Keep the parameter simple. The 30-period setting is what appeared in the results. Optimising aggressively over a long lookback until a setting looks good in-sample is precisely the kind of process that produces the 74% win rates you saw in the trap data above—wins that do not survive out-of-sample. One lookback, one signal rule, tested on data the model has never seen—that is the discipline the data rewards.