Home / Wiki / H / Hull Moving Average (HMA)

Hull Moving Average (HMA)

Trading and SpeculationUpdated 16 Nov 2026

The Hull Moving Average (HMA) is a trend-following indicator used in technical analysis. It combines several weighted moving averages in a way that removes most of the delay found in an ordinary moving average while keeping the line smooth. It was devised by the Australian trader and author Alan Hull and published in 2005.

Every moving average faces a trade-off. A long average gives a smooth line but turns well after price has turned; a short one reacts quickly but is jagged. The simple moving average lags most, and the exponential moving average somewhat less. Hull's method attempts to get both properties at once, and it belongs to a group of low-lag averages that includes the double exponential moving average and the Arnaud Legoux moving average.

01

Formula

The indicator is built from the weighted moving average, written here as WMA, in which the most recent price receives the largest weight. For a chosen period nn the calculation is:

HMAn=WMAn(2×WMAn/2−WMAn)\text{HMA}_n = \text{WMA}_{\sqrt{n}}\left(2 \times \text{WMA}_{n/2} - \text{WMA}_{n}\right)
  • WMAn\text{WMA}_{n} is the weighted moving average of price over the full period.
  • WMAn/2\text{WMA}_{n/2} is the weighted moving average over half the period, which follows price more closely.
  • 2×WMAn/2−WMAn2 \times \text{WMA}_{n/2} - \text{WMA}_{n} is a raw series that takes the faster average and pushes it further in the direction of its lead over the slower one, cancelling most of the lag.
  • WMAn\text{WMA}_{\sqrt{n}} is a final weighted average of that raw series over a period equal to the square root of nn, which smooths out the noise the second step introduces.

Half-periods and square roots are rounded to whole numbers. With the common setting of 16, the three averages have lengths of 16, 8 and 4.

02

Worked example

Take a period of 4, so the half period is 2 and the square root is 2, and five closing prices rising steadily: 10, 11, 12, 13 and 14. A weighted average of four values uses weights 1, 2, 3 and 4, which sum to 10; a weighted average of two values uses weights 1 and 2, which sum to 3.

At the fourth price, the four-period average is (10+22+36+52)÷10=12.00(10 + 22 + 36 + 52) \div 10 = 12.00 and the two-period average is (12+26)÷3=12.67(12 + 26) \div 3 = 12.67. The raw value is 2×12.67−12.00=13.332 \times 12.67 - 12.00 = 13.33. At the fifth price, the four-period average is (11+24+39+56)÷10=13.00(11 + 24 + 39 + 56) \div 10 = 13.00, the two-period average is (13+28)÷3=13.67(13 + 28) \div 3 = 13.67 and the raw value is 2×13.67−13.00=14.332 \times 13.67 - 13.00 = 14.33.

The final step is a two-period weighted average of the two raw values: (13.33+2×14.33)÷3=14.00(13.33 + 2 \times 14.33) \div 3 = 14.00. The Hull Moving Average equals the latest price of 14. Over the same four prices the weighted average stands at 13.00 and the simple average at 12.50, both trailing the market. On a steady trend the indicator has no lag at all, which is the property it was designed for.

03

How traders use it

The line is read mainly for its direction. A rising line indicates an uptrend and a falling line a downtrend, and many charting packages colour it accordingly. A change in slope is treated as a possible entry or exit point. Because the line turns quickly, it is used on short time frames and by traders who want to react early to a reversal.

Longer settings are used as a filter to define the prevailing trend, with trades taken only in that direction. The indicator can also replace conventional averages in a moving average crossover system, with a short and a long version plotted together. Hull himself suggested judging the turn of the line, not the crossing of price through it, because the line hugs price so closely that crossings are frequent and of little meaning.

04

Limitations

  • Overshoot. The step that removes lag does so by extrapolating. When price reverses sharply, the line can run past the actual price before turning, showing values that were never traded.
  • False signals in ranges. In a sideways market the slope flips repeatedly, and each flip may be read as a signal. Like other tools for trend following, the indicator performs poorly without a trend.
  • Sensitivity to the period. Results depend heavily on the chosen length, and a setting tuned to past data through backtesting may not hold up afterwards.
  • Less use as a level. Slower averages are often watched as a support level or resistance level because many traders follow the same line. The Hull version stays too close to price to serve that purpose well.

Reduced lag does not amount to prediction. The indicator is a transformation of past prices and carries no information beyond them, and there is no reliable evidence that it produces better trading results than simpler averages once costs are counted. Adaptive alternatives such as Kaufman's adaptive moving average approach the same problem differently, slowing down in noisy markets instead of extrapolating.

Last updated 16 Nov 2026