Hull moving average: less lag, more overshoot

[Fairman]

Hull moving average: less lag, more overshoot

The Hull moving average was designed to reduce lag. It uses weighted moving averages in a particular combination to track price more closely than a simple or exponential average of the same length. On a chart it looks smooth and turns quickly.

That speed comes at a cost. Because of how it's calculated, the Hull average can overshoot price at turning points. After a sharp move, it can keep going briefly beyond where price actually is before turning back. That makes it look very responsive on historical charts, but in real time, a Hull turn can be followed quickly by a turn back.

It's often used with the colour changing when the slope flips. That's easy to read but produces a lot of colour flips in choppy markets. On M5 in a range, the Hull line can switch direction several times an hour.

Where it seems most useful is as a trend direction tool on a higher timeframe, where its speed helps catch changes early and noise is lower. On very low timeframes, the overshoot and frequent flips make it harder to use.

If you're comparing it with an EMA, try the same length on both and watch them over a few sessions. The Hull will turn earlier. The question is whether those earlier turns are right more often than they're wrong.

[Fairman]

A useful test is counting slope flips per session for the Hull and an EMA of similar responsiveness. The Hull of a given length is roughly as responsive as a much shorter EMA, so comparing same-length values isn't quite fair. Matching them by how quickly they react to a move and then counting false turns in a range gives a fairer picture. In my experience watching both, the Hull tends to be cleaner in trends and noisier in ranges, which is what you'd expect from its design. Matching by responsiveness rather than by length makes the comparison fair, and it usually changes which one looks better.

[Fairman]

Repainting comes up with Hull indicators sometimes, but the standard calculation doesn't repaint. It only uses past data. Some versions on TradingView or MT4 use extra smoothing or look-ahead tricks that do repaint, though, so it's good to check the code or testing in replay. If the line on a closed candle changes when new candles form, that version isn't suitable for anything beyond a visual guide. The basic formula is simple enough that you can compare against a known clean version. Replay mode on TradingView is the quickest way to check a version you've downloaded.

[FXS Prop Desk]

The overshoot is the trade-off. Hull reacts fast, but in a sharp move it can overshoot price and then reverse, giving a false signal. On M1, that happens often during news or the London open.

Using a longer period helps, or using the Hull only as a direction filter rather than for crossovers. Some traders also combine it with a slower moving average, taking trades only when both point in the same direction.

[Shadow Trader]

A scalper watching a calendar for the payrolls number refreshes the page and waits. The number appears. By then, the market has already moved.

Calendar sites collect data from official sources, sometimes by hand and sometimes by script. Even the fast ones lag the official release by a second or two. Free services can lag by much more.

That delay is a trap for anyone trying to use the number as a trigger. The price reacts to machine-readable feeds that deliver the figures in milliseconds. A human reading a web page is competing with a very different class of participant.

Calendars are better used for scheduling and preparation. Knowing when an event is due, what the forecast is and how significant the market considers it provides context. The number itself arrives through the price.

Some traders pay for dedicated data feeds or squawk services to cut the delay. For a retail scalper, the benefit is questionable. Even a fast feed can't beat algorithms, and the price already carries the information.

If trading around releases, reading the price reaction is faster and more reliable than reading the headline.

Has a calendar delay ever caused a bad entry, and how big was the lag?

[Fairman]

Combining the Hull with a slower average can help with the overshoot. For example, only taking Hull signals in the direction of a 50 period EMA reduces the number of false turns, since many of the Hull's quick flips happen against the slower trend. You lose some responsiveness, but the signals that remain tend to be better. The Hull then acts as a timing tool within a direction set by something steadier, which plays to its strength without relying on it alone. I'd test the combination on a few months of data before relying on it.

[Fairman]

What period do people use for the Hull on M5? I've seen everything from 9 to 55. Shorter periods make it very responsive but noisy, and longer ones bring it closer to a regular moving average in behaviour, just with a bit less lag. I'd test a few periods on the pair and timeframe you trade, looking at how often it flips and how many of those flips lead to a move of a decent size. The right period depends a lot on the volatility of the pair and how long you hold trades.