What Is the Hull Moving Average?
The Hull moving average is a weighted moving average built from three others: twice the weighted average of half the length, minus the weighted average of the full length, smoothed again over the square root of the length. It turns sooner than a simple moving average of the same length.
Every moving average is late. This one is an attempt to be less late, and the interesting part is what it gives up in exchange.
How it works
Three weighted averages, in one formula:
HMA(n) = WMA( 2 × WMA(n/2) − WMA(n), √n )
A weighted average leans on recent bars. Over 3 values, the newest gets weight 3, then 2, then 1 — so the newest bar counts three times what the oldest does.
The subtraction is the trick. 2 × WMA(8) − WMA(16) takes the fast average and pushes it further
in the direction it is already moving, by the amount the slow one is behind. Then the outer WMA
smooths the result.
What it buys
Measured on this chart: 3 bars against 10.
Both turned after the high, because both are averages of the past. But seven bars is a substantial difference in when you would have acted, and it is the entire case for using this.
It also sits closer to price throughout — a mean distance of 0.51 against 0.84 for a simple average of the same length.
One thing it does not do, despite the formula: overshoot. The subtraction looks like it should throw the line past price at turns, and measured on this data it sits outside a bar’s own high-low range 19 times against the simple average’s 24 — less often, not more.
What it costs
In a range: 7 changes of direction against the simple average’s 3.
That is not a bug, it is the same property. A line that reacts sooner to a real turn reacts sooner to a fake one, and there is no setting that separates the two — because at the moment it happens, they are the same data.
So the honest summary is one sentence: it is better in a trend and worse in a trading range, and you have to know which one you are in from market structure rather than from the line.
The settings
One number: the length. The trade is the ordinary one — shorter turns sooner and is wrong more often.
What is different is where you start. A Hull at 16 behaves roughly like a simple average at something considerably longer, so if you are switching, the equivalent Hull length is shorter than the one you were using. Matching the number rather than the behaviour is how people end up with a line that flips constantly and conclude the indicator is broken.
What most people are actually running
Worth separating, because the popular version is not the formula above.
The script most traders load is “Hull Suite”, not a plain Hull moving average. It wraps the same idea in a colour-changing band and offers three variants: HMA, EHMA and THMA.
EHMA swaps the weighted averages for exponential ones — smoother, and it turns slightly later. THMA uses a three-part construction that is faster still and correspondingly twitchier.
The colour is the part that changes behaviour, and not in the arithmetic. A line that turns green and red converts a continuous reading into a signal you feel obliged to act on, and that is a presentation decision rather than a measurement.
If you are comparing yours to someone else’s, check which variant they are on. Three different lines are sold under one name, and they turn at different bars.
A worked example
Establish the condition first. Trending, so this line is worth having on the chart.
The line is rising and price is above it. That is the whole positive reading, and it is a moving average reading — nothing about the Hull changes what the line means, only how quickly it says it.
Pullbacks that hold above the line leave the read intact. This is where the closer tracking matters: at 0.51 from the close rather than 0.84, a pullback that reaches the line is a smaller pullback.
The line turning down is the signal to be out — not to be short. The same distinction as Supertrend.
And it will be 3 bars late. Plan for that rather than being surprised by it.
The original data
Across our study of 24,971 trading videos, 98 cover the Hull moving average. The median one gets 2,848 views, 92% never pass 50,000, and the median length is 7.5 minutes.
That 92% is among the highest saturations measured here, level with position trading, above Bollinger Bands at 85% and behind only Chaikin Money Flow at 96% — on a small field of 98 videos.
The corpus carries description text for 58 of those 98, and across those 58, five mention invalidation, failure, or what a bad read looks like — about one in twelve, the second-highest rate here after average true range (ATR).
When it fails
Sideways it is worse than what it replaced
Covered above with numbers: 7 direction changes against 3. If you are using it in a range you have chosen the wrong tool for the condition, and the responsiveness you paid for is working against you.
It is still an average of the past
Three bars is better than ten and it is not zero. No arrangement of past closes can turn before the thing it is averaging has turned, and any tool claiming otherwise is either repainting or lying.
You matched the length instead of the behaviour
Covered above. A Hull 50 is not a simple 50 — it is much faster, and swapping one for the other by number gives you an indicator behaving nothing like the one you meant to replace.
You looked once the trend had run
Every turn it caught early is obvious in retrospect and every false turn looks obvious too — afterwards. At the pause above, the line is still rising and there is nothing on the chart that separates the two cases.
Related
Moving average is what this is a variation of, and the comparison is the only way to see what has actually changed.
Trading range is the condition where its advantage becomes its problem.
And Supertrend is the other way to get a fast trend read, built from volatility rather than from averaging.
This is the one I actually have on my charts, and the reason is narrow: I want a line that tells me a trend has turned without waiting half a month to say so. What I had to learn is that the same responsiveness means it flips around in a quiet market, so it is a trend tool and I do not look at it when nothing is happening. If someone is using it to generate entries in a range they will lose money slowly and blame the indicator.
— Michael Whitman, from this video
This page is educational, not financial advice. Test every idea on your own charts before risking money.