Ten momentum oscillators, each measuring some form of "how stretched is current price from its recent average." They differ in how they normalize the reading, how much they smooth, and which extra signals they produce.
The big idea: oscillators work brilliantly in range-bound markets and disastrously in trends. Every oscillator strategy needs a regime filter before it's tradable.
At a glance
| Indicator | Scale | Default period(s) | Lines | Best for |
|---|---|---|---|---|
| RSI | 0–100 | 14 | 1 | Default oscillator — universally useful |
| Stochastic | 0–100 | (14, 3) | 2 | Range-bound crossovers, divergences |
| Williams %R | −100–0 | 14 | 1 | Inverted Stochastic — same math, cleaner read |
| CCI | Unbounded | 20 | 1 | Statistical extremes, cyclical instruments |
| PPO | Unbounded % | (12, 26, 9) | 3 | MACD with cross-instrument comparability |
| Ultimate Oscillator | 0–100 | (7, 14, 28) | 1 | Multi-period confluence, fewer false signals |
| StochRSI | 0–100 | (14, 14, 3, 3) | 2 | Earlier signals than RSI, more noise |
| TSI | −100–+100 | (25, 13, 7) | 3 | Smoother, lagged momentum — high signal quality |
| CMO | −100–+100 | 14 | 1 | Pure unsmoothed momentum |
| Fisher Transform | ±3.8 (saturating) | 10 | 2 | Sharp reversal peaks, statistical normalization |
How to choose
You need a general-purpose momentum oscillator. Start with RSI. It's the default for a reason — well-understood thresholds (70/30), balanced smoothing, single-line simplicity. Switch only when you have a specific reason.
You want crossover signals. Reach for Stochastic or StochRSI. Both expose %K/%D lines built for the standard crossover pattern. StochRSI fires more often; Stochastic fires more reliably.
You want fewer false signals at the cost of more lag. TSI or Ultimate Oscillator. Both filter aggressively — when they signal, the signal means more.
You want comparable signals across multiple instruments. PPO. Its percentage normalization makes readings on different price-scale instruments directly comparable — useful for portfolio-level analysis and cross-instrument screens.
You want sharp visual reversal peaks. Fisher Transform. The Gaussian transformation creates spikes at turning points rather than gradual curves. Reversal moments are unmistakable when they occur.
You want a binary momentum-direction flag. CMO's Direction output (1 or −1) gives this directly. The cleanest "is momentum positive or negative?" flag in the library.
You don't know which to pick. RSI. Every other oscillator solves a specific problem RSI has, but RSI's defaults work for ~70% of strategies and any other oscillator can be substituted later without restructuring the rest of the system.
The shared failure mode
Every oscillator on this list will lose you money if you trade its extremes during a trending market. RSI > 70 in a strong uptrend doesn't mean "sell" — it means "the trend is working." The same applies to Stochastic > 80, CCI > 100, Williams %R > −20, and so on.
Three ways to handle this:
- Add a trend filter. EMA(50), SMA(200), HMA(21) — any of these as a regime gate. Only allow countertrend signals when the trend filter agrees.
- Use a regime indicator. KAMA's Efficiency Ratio is the cleanest — only trade oscillator extremes when ER < 0.3 (range regime).
- Use the oscillator differently in trends. In trends, treat 70+ readings as confirmation (buy strength) rather than reversal warnings (sell extremes). Wire via Conditional Flow to switch behavior by regime.
Every oscillator page in this section covers these patterns under "Three setups" and "Advanced patterns in AlgoLift."
Common patterns
The two-line crossover. Stochastic, StochRSI, PPO, TSI, and Fisher Transform all expose a signal/main line pair. The crossover pattern (line crosses signal in an extreme zone, gated by a trend filter) is the most common momentum strategy structure in the entire reference. Build the wiring once, swap oscillators to test variants.
The divergence pattern. Price makes a new high but the oscillator doesn't follow. Implemented with BarsAgo Modifier to compare current values against historical values N bars ago. Works on every oscillator on this list, but most robust on RSI, TSI, and Ultimate Oscillator.
The midline filter. RSI's 50-line, CMO's zero-line, TSI's zero-line, PPO's zero-line. Using the midline cross as a "what direction is momentum biased?" flag for unrelated strategies. Simple, no tuning required, works across most market conditions.
The dual-oscillator confluence. Requiring two oscillators with different smoothing structures to agree before entering a trade. Common combinations: RSI + Stochastic, RSI + Ultimate Oscillator, CMO + Fisher Transform. The confluence dramatically reduces false signals at the cost of fewer total entries.
Oscillators don't predict reversals — they measure how stretched the current price is from its recent average. Whether stretched-equals-reversal depends entirely on regime. Pair every oscillator strategy with a trend or regime filter, and you'll avoid the most common failure mode in retail systematic trading.