OptiNod Academy

JMA — Low-Lag Adaptive Average

The smoothest trend line: powerful while a trend is underway, but less reliable at turning points.

JMA is a moving average that reacts quickly while keeping the curve smooth. During an active trend, it absorbs minor pullbacks well and shows the trend direction cleanly.

At turning points, however, that smoothness becomes a weakness. The price structure may already have broken down while the JMA line is still slow to turn. Use JMA only as a guide while a trend is in progress, and confirm trend reversals separately with faster signals such as price structure, RSI, or MACD.

The right way to use it starts with separating its role. When the trend is still intact, treat pullbacks into the JMA as entry candidates. When the JMA starts to flatten, reduce new entries. If you wait until the JMA turns, you will usually recognize the reversal too late.

It is the most reliable guide while a trend is underway

When a trend is clearly in progress, JMA follows price with very little noise. Even when pullbacks appear inside the trend, the JMA barely wavers, so its slope holds steady. That makes it one of the cleanest tools for reading the current direction and strength of an active trend.

This is where it works well as a pullback-entry guide. If you buy when price pulls back to the JMA, the line’s smooth behavior helps avoid getting stopped out unless the trend has truly ended. Minor pullbacks are absorbed by the JMA, with stops placed 0.5 to 1 ATR below the line.

> After SOL breaks above $200 on the 4-hour chart in November 2024 and trends around $230-$250,

> JMA(14, 0, 2) maintains an upward slope,

> price pulls back into the JMA, then a candle closes back above it.

> Enter long at that candle’s close, with the stop 0.5 ATR below the JMA line.

> If the JMA slope clearly flattens or price closes below the JMA, treat the setup as invalid.

The key condition is that the trend must still be underway. JMA is strong at absorbing pullbacks inside a trend, but once the trend itself weakens, the same smoothness becomes a liability.

The Paradox of Smoothness — Turning Points Are Hard to See

When you make a moving average smoother, it absorbs major noise and major real changes in the same way. The same mechanism that filters false pullbacks also works during a real trend ending. As a result, even after the trend ends and price starts moving the other way, the JMA slope changes slowly.

After SPY made a high near $565 in July 2024, the daily JMA(14) did not clearly turn negative until price had already fallen to $535. Around the same period, EMA(14) flattened about a week earlier, and the RSI close below 50 came even before that. JMA’s smoothness cleanly absorbed two small July pullbacks without false signals, but it absorbed the real trend reversal the same way, delaying recognition.

That gap is JMA’s structural limitation. The tool that feels fastest while a trend is underway can become the slowest tool at the turning point. The same indicator behaves in opposite ways depending on where it is used.

That is why roles must be separated. Use price-structure breaks, MACD histogram divergence, or RSI closes below 50 to identify turning points. Use JMA only as a guide for following the trend after the turning point has passed. If you wait for the JMA to turn, every decision comes late.

Phase — A Setting That Can Go Negative

JMA’s second parameter, phase, is a phase-adjustment value ranging from -100 to +100. At 0, lag and overshoot are balanced. Positive values reduce lag but increase overshoot. Negative values do the opposite. The ability to set this value below zero is a major difference from EMA and SMA.

EMA lag cannot be adjusted directly. EMA(7) has less lag than EMA(21), but within EMA itself, the only way to reduce lag is to shorten the period, and shortening the period brings in more noise. JMA, by contrast, lets you adjust lag directly through phase. With the same length of 14, setting phase=+50 reduces lag while preserving smoothness.

The problem is that reducing lag through phase does not make JMA detect turning points earlier. Phase pulls the curve closer to price while a trend is in progress; it does not meaningfully bring forward the moment when the curve turns. Even with phase=+90, JMA will still usually recognize a trend ending later than EMA.

Phase is best used to fine-tune pullback entries during an active trend. As the JMA tracks closer to price, it creates more entry locations on smaller pullbacks. Raising phase to catch turning points is the wrong use of the parameter.

Length, Phase, and Power — The Overfitting Risk of Three Parameters

JMA has three adjustable parameters: length, phase, and power. EMA has one. That gives JMA three times as much room for fine-tuning to the same dataset, and it also clearly increases the risk of overfitting. A combination such as (14, +30, 2) may look strong in a backtest and then fail on the next dataset.

A practical rule is simple: if you optimize all three at once, the result is hard to trust.

• length: Choose within the 7-21 range based on the asset’s volatility cycle.
• power: Keep it fixed at 2, because its effect on smoothness is smaller than phase.
• phase: Fine-tune only within the -50 to +50 range.

By asset type, phase can be set like this. For assets with long, smooth trends such as SPY and QQQ, values near phase=0 usually work well. For assets such as BTC and ETH, where fast trends and deep pullbacks often mix, phase=+30 to +50 tends to fit pullback areas more often. For altcoins that are often range-bound, keep phase at 0 or negative to reduce false signals. Pushing it positive adds more false signals.

Reproducibility Problems From a Proprietary Algorithm

The original JMA algorithm is commercial and proprietary. Jurik Research is the only source for an official license, while TradingView community versions, MetaTrader ports, and open-source recreations are all reverse-engineered estimates.

You need to know two things. First, even with the same data and the same (length, phase, power), the line can differ slightly across platforms. A setup that worked well on TradingView may not line up at the same point in MetaTrader. Second, backtest results cannot be reproduced as directly as with EMA or SMA. Because the formula is estimated and implemented differently across platforms, the values vary slightly.

Before building a setup, always plot your platform’s JMA against EMA and DEMA on the same data at least once. Visually check how smooth the line actually is. If you copy someone else’s JMA setup exactly, you may not get the same signals.

JMA Is Not a Universal Tool

There is no statistical proof that JMA is always better than EMA. It can produce better results in smooth-trending conditions, such as strong trend phases in assets like SPY. But in fast-turning conditions, such as sharp altcoin moves or immediately after news events, it clearly performs worse than EMA.

JMA has a specific place where it works well. Use it only as a guide for following an active trend in low-volatility, smooth-trending conditions. Outside that role, it can produce worse results than simply using EMA.

The Three Most Common Ways JMA Is Misused

• Simple crossover entries: A JMA(7) crossing JMA(21) usually appears after the trend has already progressed. It has fewer whipsaws than an EMA crossover, but the entry is still late. Keep crossovers only for confirming trend direction, and leave entries to price-structure breaks.
• Trying to catch turning points with JMA: At turning points, price structure, MACD, and the RSI 50 line react much faster than JMA. Waiting for JMA to turn delays recognition of the trend ending.
• Raising phase too aggressively: When phase is pushed above +70, lag decreases but overshoot increases with it. If an asset often stops you out on pullback entries, lower the phase.