The Hull Moving Average: Less Lag, Bought With Overshoot

The Hull Moving Average is smoother and faster than an ordinary moving average at the same period, which sounds like it should be impossible. It is not impossible; it is a trade, and understanding what is being traded away is the difference between using it well and being caught by it.

Hull Moving Average at a glance

What it is and why it works

Every moving average has the same problem: it is an average of past prices, so it necessarily trails behind the current price. Make the period shorter and it trails less but becomes noisy. Make it longer and it smooths well but trails badly. Alan Hull’s idea was to attack that trade-off directly rather than accept it.

The formula has three steps. First, calculate a weighted moving average over your chosen period: say 16 bars. Second, calculate a weighted moving average over half that period, 8 bars, and double it. Third, subtract the first from the second and take a weighted moving average of the result over the square root of the period, which for 16 is 4.

The middle step is where the lag goes. Twice the fast average minus the slow average is a linear extrapolation. If the fast average is above the slow one, the market has been accelerating, and the formula projects that acceleration forward. It is not seeing the future; it is assuming the recent rate of change continues. The final short smoothing then cleans up the noise that extrapolation introduces.

Which tells you exactly where it fails. Extrapolation is correct while the trend continues and wrong at the moment it turns, so the Hull MA overshoots at reversals. It will run past the turning point, then whip back sharply. The smoothness of the line makes this feel like a decisive signal rather than what it is; a projection being corrected.

On redundancy: this is still a moving average. It occupies the same slot on your chart as an EMA, a DEMA or a TEMA, and those last two use a very similar extrapolation trick. Running an HMA alongside an EMA and treating their agreement as confirmation is running two smoothings of the same closes. Choose one, and choose it because you understand what its particular lag-versus-overshoot balance costs you.

How to trade it, step by step

1. Install a Hull MA, because MT5 does not have one. Source a custom indicator from the MQL5 community or code it. Before relying on it, verify the implementation uses linear weighted averages at all three stages and the square-root final period, substituting EMAs is common and produces a visibly different line under the same name.
2. Pick a period, and prefer a perfect square. The final smoothing uses the square root of the period, so 9, 16, 25, 36 and 49 avoid rounding decisions that different implementations handle differently. If two Hull MAs disagree on the same chart with the same period, this is usually why.
3. Use the slope, not the crossover, as the primary read. The Hull MA’s advantage is that it turns early and cleanly. A rising Hull is an uptrend state, a falling Hull is a downtrend state. Waiting for price to cross it discards the speed advantage you installed it for and gives you a laggier signal than the line itself provides.
4. Colour it by direction if your implementation allows. Most community versions colour the line green when it is rising and red when it is falling. This is not decoration; it makes the one useful reading immediate, and it stops you interpreting a flat, ambiguous line as a trend.
5. Add a slower conventional average as a state filter. A 50 or 100 period EMA on the same chart tells you which direction the Hull’s turns are worth acting on. Hull turns against a clearly established longer trend are frequently the overshoot correcting rather than a genuine reversal.
6. Do not take the first Hull turn at an extreme. Because it extrapolates, the line rolls over at the top of a move and then often resumes. Requiring a closed bar beyond the turn, or a break of a structural level from market structure, filters out a large share of the overshoot whips at the cost of a little speed.
7. Stop hunting for a period that removes the problem. Shorter periods turn sooner and overshoot more; longer periods overshoot less and lag more. The trade-off is structural. Pick a period that matches your holding time and spend your effort on the state filter instead.
8. Check it against a plain EMA once, honestly. Put a 16 Hull and a 16 EMA on the same chart and scroll through a few hundred bars. You will see the Hull turn earlier in trends and whip harder in ranges. That comparison tells you more about whether it suits your market than any amount of reading.

Size every one of those entries with the position size calculator and check the trade is worth taking with the risk/reward calculator before you commit.

The conditions it needs

Sustained, directional trends

This is where the extrapolation assumption is correct. In a market that keeps going, projecting the recent rate of change forward is a good approximation, so the Hull MA tracks price closely with much less delay than a conventional average and gives back far less at the end of the move.

As a trailing reference in a trend-following approach

Its combination of smoothness and responsiveness suits trailing better than most averages. A conventional EMA either lags so far that it gives back a large share of the move or is so short it whips out on every pullback. The Hull sits between those, which is a genuinely useful place to be.

Higher timeframes on trending instruments

The overshoot problem is proportionally less costly when each bar represents a meaningful stretch of trading and the trends last for many bars. On a daily chart of a trending instrument the Hull is close to its best; on M1 in a range it is close to its worst.

As a replacement for a shorter EMA rather than an addition

If you have been running a fast EMA and finding it noisy, the Hull does the same job with far fewer false turns, because the final smoothing stage removes much of the noise that extrapolation introduces. Swap one for the other. Running both gives you two views of the same closes.

When it fails

• It overshoots at reversals, by construction. The lag reduction comes from projecting the recent rate of change forward, and at a turn that projection is wrong. The line runs past the turning point and then snaps back. Because the line is smooth, this looks decisive rather than mistaken, which is what makes it costly.
• In a range it whips continuously. A market oscillating around a mean produces a Hull MA that turns every few bars, each turn looking clean and each one wrong. The smoothness that makes it attractive in a trend makes it actively misleading in sideways conditions.
• Its smoothness manufactures unearned confidence. A jagged line makes you hesitate; a clean one does not. The Hull MA does not know more than an EMA does, it is the same closing prices with a different weighting, but it presents its output with a certainty that is not backed by additional information.
• The current bar’s value moves, and people mistake that for repainting. The Hull MA does not repaint: once a bar closes, its Hull value is fixed forever. But the value on the forming bar changes with every tick, and because the extrapolation amplifies recent movement, it changes more visibly than an EMA would. That is normal live behaviour, not revision of history.
• Implementations disagree. With no MT5 standard, community versions vary in how they round the square-root period and whether they use weighted or exponential averaging internally. Two indicators both labelled HMA(20) can plot noticeably differently, which makes shared settings advice unreliable.
• It is not independent evidence from another moving average. Hull, EMA, DEMA and TEMA are all smoothings of the same closing prices, and DEMA and TEMA use a closely related extrapolation. Stacking them and waiting for agreement produces a feeling of confluence with no additional information behind it.

Markets it is most informative on

• NAS100 (Nasdaq): Strong, persistent intraday trends are exactly where the extrapolation assumption holds.
• GBP/JPY: Long directional runs where reduced lag saves a meaningful part of the move.
• BTC/USD (Bitcoin): Extended trends punctuated by sharp turns: the Hull’s best and worst conditions in one instrument.
• Gold (XAU/USD): Trends hard enough to reward a fast average, provided a state filter keeps you out of its range phases.

For different levels of experience

If you are brand new

A moving average is just the average price over the last so many bars, drawn as a line. Because it is an average of the past, it always trails a bit behind where price actually is. That delay is the reason moving-average signals so often arrive after the move.

The Hull Moving Average tries to fix that. It looks at how fast the average has been changing and projects that speed forward a little. The result is a line that keeps up with price much better than a normal average.

The catch is worth knowing before you use it. Projecting forward works while the market keeps going and fails at exactly the moment it turns, so the Hull runs past the top or bottom and then snaps back. It looks very confident while it does this. Use the line’s colour or direction to tell you what state the market is in, but do not take a trade just because it changed direction, especially if the wider market has been going sideways.

If your results are inconsistent

Two adjustments make a large difference here. The first is to stop using price crossing the Hull as your signal. You installed a faster average, then added the lag back by waiting for a crossover. The Hull’s slope changes before price crosses it, so read the slope and you keep the advantage you paid for.

The second is to put a filter above it. The Hull’s failure mode is not random; it is specific and predictable. It overshoots at turns and whips in ranges. A slower average, or a read of higher-timeframe structure, tells you whether the current Hull turn is happening inside a trend or inside a chop, and those two situations need opposite responses.

Also settle the repaint question for yourself, because it comes up constantly in forums. The Hull does not repaint. A closed bar’s Hull value never changes. What people are seeing is the forming bar’s value moving on every tick, which happens with every indicator, and moves more here because the extrapolation amplifies recent change. Check it yourself: note the value on a closed bar and come back an hour later.

If you are experienced

The construction is a second-order finite-difference extrapolator wrapped in a low-pass filter. The 2×WMA(n/2) − WMA(n) term is a linear forecast one half-window ahead, and the √n final WMA is the noise suppression that makes the forecast presentable. Framed that way, the phase-lead is real and quantifiable, and so is the cost: any extrapolating filter has gain above unity in part of its passband, which is precisely the overshoot, and it amplifies high-frequency content that the final smoothing only partially removes.

Compared with DEMA and TEMA, which apply the same lag-cancellation idea with exponential kernels, the Hull’s weighted kernel gives it a more compact impulse response and therefore less ringing, at the cost of a harder cutoff and more sensitivity to a single outlier bar entering or leaving the window. Which you prefer depends on whether your instrument’s pathology is persistent noise or occasional bad prints.

For systematic use, two points matter. First, the overshoot is a deterministic function of the recent second difference, so it is estimable rather than mysterious; you can compute the expected overshoot and require price to exceed it before accepting a turn, which is a better filter than lengthening the period. Second, the √n rounding is a genuine reproducibility hazard: implementations differ, and a strategy fitted on one HMA build will not reproduce on another. Pin the implementation, or write your own, before any parameter work.

Risk management for this strategy

The Hull MA’s specific risk is that it makes marginal setups look clean. A smooth, decisively turning line encourages larger positions and faster entries than a jagged one would, and that psychological effect is real and worth naming, because the underlying information is no better than an EMA’s.

Practically, the overshoot dictates stop placement. Entering on a Hull turn and placing the stop just beyond the line puts the order exactly where the correction to the overshoot will travel. Stops belong beyond the structural extreme, the swing high or low that would genuinely invalidate the idea, which is usually further away. Size from that distance using the position size calculator, and accept the smaller position rather than moving the stop closer.

Finally, treat consecutive Hull turns as a warning rather than as more signals. When the line has changed direction three times in a short stretch, the market is ranging and the indicator’s assumption has failed. The correct response is to stop taking its signals, not to take the next one with more conviction because the previous ones lost.

Where Market Structure Pro fits

The Hull Moving Average is excellent in a trend and actively harmful in a range, and it gives you no help at all in telling which one you are in. That single gap accounts for nearly every loss traders take with it.

Market Structure Pro exists to fill that gap. Its dedicated ranging and chop filter has one job: to say NO TRADE when the market is oscillating rather than trending. Applied to a Hull-based approach, that is the difference between taking the turns that continue and taking the turns that whip. The verdict arrives as TRADE, TRANSITION or NO TRADE with a confidence percentage, an A/B/C grade and a plain-English explanation of what is driving it, rather than as another line whose slope you have to interpret.

The TRANSITION state deserves a specific mention here, because it maps directly onto the Hull’s weakness. The moments where the extrapolation is most likely to be wrong are precisely the moments where a market is changing character: leaving a trend, entering a range, or the reverse. Having that state named rather than inferred from a smooth line is worth more than any refinement of the moving average itself. And because MSP locks its state on the closed bar and does not repaint, you can go back afterwards and see exactly which state each of your Hull turns occurred in, which is how you find out whether the filter is working.

Frequently asked questions

What is the Hull Moving Average?

It is a moving average designed by Alan Hull to reduce lag without becoming noisy. It calculates a weighted average over a chosen period and another over half that period, doubles the fast one and subtracts the slow one to extrapolate the recent rate of change forward, then smooths the result over the square root of the period.

Why is the Hull Moving Average faster than an EMA?

Because it does not simply average the past, it projects forward. The step that doubles the half-period average and subtracts the full-period average is a linear extrapolation of how fast the average has been changing. That projection is what removes the lag, and it is also why the line overshoots when a trend turns.

Does the Hull Moving Average repaint?

No. Once a bar has closed, its Hull value is fixed and will never change. What traders often mistake for repainting is the value on the current forming bar moving as new ticks arrive, which happens with every indicator and is more visible here because the extrapolation amplifies recent price movement.

Is the Hull Moving Average available in MetaTrader 5?

No, it does not ship with MT5. MetaTrader provides Simple, Exponential, Smoothed and Linear Weighted averages along with AMA, DEMA, TEMA, FrAMA and VIDYA, but no Hull. You need a custom indicator from the MQL5 community or your own implementation.

What is the best Hull Moving Average period?

There is no universal answer, but periods that are perfect squares such as 9, 16, 25 or 49 avoid rounding differences in the square-root smoothing step, which is a common source of disagreement between implementations. Beyond that, shorter periods turn sooner and overshoot more, longer ones the reverse.

What is the main weakness of the Hull Moving Average?

Overshoot at reversals. Because it works by projecting the recent rate of change forward, it is correct while a trend continues and wrong at the moment it ends, so the line runs past the turning point before snapping back. In ranging markets this happens repeatedly and every false turn looks convincing.

Should I use the Hull MA crossover or its direction?

Its direction. The slope of the Hull line changes before price crosses it, so waiting for a crossover reintroduces the lag the indicator was designed to remove. Most implementations colour the line by direction for exactly this reason, and reading the colour change is the intended use.

Is the Hull Moving Average better than the EMA?

It is faster at the same period and gives back less at the end of a trend, which is a genuine advantage in trending markets. It is also more prone to false turns in ranges. Neither is better in general, and because both are smoothings of the same closing prices, running them together adds no independent information.

Related reading

• Moving Averages: The family this belongs to, and the lag problem it was built to attack.
• Supertrend: Another trend-state tool that turns cleanly, and whips in the same conditions.
• Trends vs Ranges: The single distinction that decides whether the Hull helps or hurts.
• Breakout Trading: Where a fast-turning average adds most, and where its overshoot costs most.