ACADEMY / Indicators and signals

WaveTrend indicator: settings, cross signals and what a test shows

The WaveTrend indicator is a smoothed oscillator that shows how far price has stretched from its recent average, drawn as two lines. Its crosses are worth trading only as a written rule tested with costs: the plain cross fires too often on short bars, while a zone-filtered cross or longer bars leave room for the round trip.

By Soren Huang · Published 2026-11-06 · updated 2026-11-06 · Editorial policy

Contents

How does the WaveTrend indicator work?

The WaveTrend indicator is a smoothed oscillator that shows how far price has stretched from its own recent average, scaled by the usual size of that stretch, and drawn as two lines. Its crosses are useful only as a written rule tested with costs: the plain cross fires too often on short bars, while a zone-filtered cross or longer bars leave room for the round trip.

An oscillator is an indicator that swings above and below a centre line instead of following price upward. The WaveTrend indicator page gives the full definition; this article spends its words on settings, signals and testing. Think of a thermometer that does not report degrees but reports how unusual today is compared with the last two weeks. A reading of +60 means price sits far above its average by the standard of recent swings; a reading of −60 means far below. The main line is called WT1. The second line, WT2, is a short average of WT1. When WT1 crosses WT2, the stretch has started to shrink or grow. Where WaveTrend sits among the wider family of technical analysis tools and indicators you can test is covered in that guide.

WaveTrend shows how stretched price is from its own average. The two lines, the zones at ±53 and ±60, and the crosses inside those zones are the parts a test can use.

How is WaveTrend calculated, step by step?

WaveTrend is built in four ordered steps, and the second line is only a short average of the first, which is why the two lines cross so often. Each step uses the previous one.

Average the typical price. The typical price of a bar is the average of its high, low and close. An exponential average of it over the channel length (10 bars by default) is the centre that price is measured against.
Measure the typical distance from that average. Take the distance between each bar's typical price and the centre, ignore the sign, and average those distances over the same channel length. This is how large a normal swing has been lately.
Scale today's distance by the typical distance. This turns a distance in points into a number that means the same thing on a quiet week and a wild week:

scaled distance = (typical price − centre) ÷ (0.015 × typical distance)

Here typical price is today's bar, centre is the average from step 1, typical distance is the average from step 2, and 0.015 is a fixed constant that only sets the size of the numbers so that the usual swing lands near the ±50 to ±60 area.

Smooth twice. An exponential average of the scaled distance over the average length (21 bars by default) is WT1. A simple average of WT1 over 4 bars is WT2.

The four ordered steps that turn price into the two WaveTrend lines. Only the last step produces the lines a chart shows.

What are the default WaveTrend settings, and should you change them?

The widely used defaults are a channel length of 10, an average length of 21 and a signal smoothing of 4, and whether to change them belongs in a test, not in a chart you look at by eye. A chart shows you the crosses that worked and hides the ones that did not. Lag is the number of bars between a turn in price and the turn in an indicator built from it. Every averaging step adds lag, so the two length settings trade earliness against noise.

Setting	What it controls	Shorter value	Longer value
Channel length (default 10)	How many bars define the centre and the typical distance	Centre follows price closely; more crosses, earlier, more noise	Centre moves slowly; fewer crosses, later, smoother
Average length (default 21)	How much WT1 is smoothed	WT1 turns sooner and bends more often	WT1 turns later and bends less often
Signal smoothing (default 4)	How far WT2 lags WT1	Crosses fire sooner and more often	Crosses fire later and less often
Zone levels (default ±53 and ±60)	Where overbought and oversold begin	More bars count as extreme	Fewer bars count as extreme

A workable way to decide is to test a range rather than a single value: channel 6 to 20, average 13 to 40, smoothing 3 to 5. If results are good only at one exact pair of lengths and poor at the neighbours, the pair fitted noise. If a whole block of neighbouring settings behaves alike, the rule has something real to stand on.

Shorter lengths cross earlier and more often; longer lengths cross later and less often. The same 60 bars give three different trade counts.

Which WaveTrend signals can you test: overbought, oversold or the cross?

There are three testable rules from the same two lines, and the zone-filtered cross trades far less often than the plain cross, which changes the cost picture completely. The overbought zone is the area above +53 (more strongly above +60) and the oversold zone is the area below −53 (more strongly below −60). A zone alone is not an order; a line can stay in a zone for many bars during a strong trend. The cross is what gives a bar to act on.

Rule	Buy when	Sell or exit when	How often it fires
A. Any cross	WT1 crosses above WT2 anywhere	WT1 crosses below WT2 anywhere	Most often
B. Zone-filtered cross	WT1 crosses above WT2 while both are below −53	WT1 crosses below WT2 while both are above +53	Least often
C. Zero-line cross	WT1 crosses above 0	WT1 crosses below 0	In between

Three rules from the same two lines fire at very different rates. The trade count decides how much of the move each rule must give to costs.

Rule B needs an exit of its own, because the opposite zone cross may not come for weeks. Common choices are an exit when WT1 crosses the zero line, a time exit after a fixed number of bars, or a fixed stop. Rule A and rule C are always in the market, long or short, unless you add a flat condition. Trade count matters because every trade pays commission and slippage, and the rule that fires most often has the smallest move per trade to pay them from.

Hypothetical example

Suppose rule A fires 60 times a year on 15-minute bars and captures $25 of gross move per trade, while rule B fires 12 times a year and captures $90 per trade. With a round-trip cost of $30, rule A nets −$5 per trade, or −$300 a year. Rule B nets $60 per trade, or $720 a year, from one fifth of the trades. The gross backtest ranks A above B; the net backtest reverses the order.

How does WaveTrend differ from RSI and CCI?

WaveTrend is a smoothed cousin of CCI with a signal line added, while RSI measures the balance of up and down closes, and the practical difference between all three is lag and trade count. The CCI indicator divides the distance from a simple average of typical price by the average distance, using the same 0.015 constant, with no smoothing afterwards. WaveTrend replaces the simple averages with exponential ones and then smooths the result over 21 bars, so it is slower and far less jagged. The RSI indicator works from a different question: over the last 14 bars, how much of the movement was up closes versus down closes? It is bounded between 0 and 100, like the stochastic oscillator and its %K and %D lines, while CCI and WaveTrend have no hard limit.

	WaveTrend	RSI	CCI
What it measures	Scaled distance from an exponential average	Balance of up and down closes	Scaled distance from a simple average
Scale	Open; zones at ±53 and ±60	0 to 100; levels at 70 and 30	Open; levels at +100 and −100
Smoothing and lag	Two averaging steps after scaling; most lag	One smoothing step; medium lag	None after scaling; least lag
Built-in signal line	Yes, WT2	No	No
Trade count from a cross rule	Lowest	Medium	Highest

What each oscillator measures, how it is bounded and how it lags. The smoother line crosses less often and later.

The two-line structure makes WaveTrend closer in use to the MACD indicator and its signal-line cross than to RSI: in both, the signal is a smoothed line crossing its own short average. The fair comparison between any of these is not a chart impression. It is each indicator's cross rule, with the same costs, on the same bars, over the same period. Park and Irwin (2007) reviewed decades of such tests and found that results depend heavily on costs and on the period, which is exactly why the comparison has to be run rather than guessed.

What timeframe and market suit the WaveTrend oscillator?

WaveTrend produces usable swings on bars long enough that the average trade covers several times the round-trip cost, so daily and multi-hour bars are the first place to test it and very short intraday bars are the last. The indicator has two averaging steps after scaling. On 1-minute bars that lag is a few minutes; on daily bars it is several days. The lag is the same number of bars either way, but the move captured per trade grows with the bar size while the commission and slippage per trade do not.

Markets with wide, clean swings relative to their tick cost suit a stretch-and-return indicator best: liquid index futures such as ES, NQ and TAIEX futures, gold, and the major currency futures. Thin contracts add slippage that eats the small edge a cross rule has. If you want to test it intraday anyway, the trade-count and cost reasoning in what testing shows about the best indicator for intraday trading applies without change: start at 30- or 60-minute bars, not 1-minute bars, and compare the average trade after costs at each bar size before trusting any of them.

What does a backtest of a two-line cross rule on NQ show?

A stand-in two-line cross rule on NQ daily bars produced a positive but uneven result after costs, with a drawdown that is a large part of the net profit and long flat stretches between trades. AlphaPrime (Valynox) ran the search on NQ daily bars from 2019-01-02 to 2025-06-30, long and short, one contract, with $20 commission and $10 slippage charged on every trade. The rule tested was a MACD cross used as its only entry and exit, with no stop, no target and no filter. It stands in for the WaveTrend main-line cross because both rules are a smoothed line crossing its own short average; the run did not test WaveTrend itself.

Source: AlphaPrime product engine (CPU), run 2026-10-31. NQ daily bars, 2019-01-02 to 2025-06-30, long and short, 1 contract, $20 commission and $10 slippage per trade.

Number	Result
Net profit (closed trades)	$96,490
Max drawdown	$41,205
Trades	146
Win rate	56.16%
Profit factor	1.373
Average trade	$660.89
Time in market	17.7%

Look at the slope changes along the equity line, the depth of the largest drawdown against the climb before it, and whether the out-of-sample stretch keeps the shape of the earlier part. NQ · daily · 2019-01-02 to 2025-06-30 · CPU engine backtest

This is the one candidate the search kept out of 1,024 it tried, chosen for its return against its drawdown. Picking the best of many after the fact flatters the kept one: Bailey, Borwein, López de Prado and Zhu (2014) show that the best backtest in a large search carries a selection bonus that new data does not repeat, and Sullivan, Timmermann and White (1999) found the same effect across thousands of simple technical rules. The numbers above describe this one run on this one history. They do not say what the rule or the indicator earns, and they are not something a reader can expect.

What the run does show is the shape a two-line cross rule produces on daily bars. There were 146 trades across six and a half years, a swing pace, not a day-trading pace. Time in market of 17.7% means the rule was flat most of the time, so a large part of the index's movement was simply not traded. The average trade of $660.89 is many times the $30 of commission and slippage per trade, which is the margin a daily-bar rule has and a 5-minute rule does not. The profit factor of 1.373 means the winners paid for the losers with a little over a third left over, and the max drawdown of $41,205 is what an account would have had to sit through to see the rest.

How do you set up your own WaveTrend test?

Write down the settings range, the entry rule, the exit rule, the costs and the pass threshold before the first run, then test the three signal types side by side on data you have not tuned on. The order matters because anything decided after seeing a result becomes part of the tuning.

Pick the bar size and market. Start with daily or 60-minute bars on a liquid index future, for the cost reasons above.
Write the settings range. Channel 6 to 20, average 13 to 40, smoothing 3 to 5. Test the range, not one pair.
Choose one signal type and one exit. Rule A, B or C from the table above, with its exit written out. Act on the open of the next bar after the cross, because the current bar's WaveTrend value moves until that bar closes.
Enter commission and slippage per trade. Use your broker's commission and at least one tick of slippage per side; rerun with double the slippage to see whether the result depends on cheap fills.
Write the pass threshold, then run. A minimum number of trades and a minimum average trade after costs, written before the first run. Keep the most recent block of data out of every run for one final test.

The order of decisions before the first run. Going back from the result to the settings range turns the test into tuning.

Steps 2 and 3 produce a large grid: three settings ranges, three signal types and several exits multiply into thousands of variants, and AlphaPrime's strategy generation takes exactly these inputs, the indicator, the parameter ranges, the entry and exit conditions and the search constraints, and turns the grid into candidates that are each evaluated with the same costs. Whether you run the grid by hand or with a generator, the pass threshold from step 5 is what keeps the best-looking cell from being mistaken for a tested rule.

A good first run to do today: load daily bars of one index future, keep the last twelve months out, and test rule B with a zero-line exit over the settings range above. Count the trades, read the average trade after costs, and compare it with rule A on the same bars before you look at a single equity curve.

Frequently asked questions

Does the WaveTrend indicator repaint?

On closed bars it does not. WaveTrend is built from averages of the typical price, and once a bar has closed its high, low and close do not change, so neither do the WT1 and WT2 values of that bar. On the bar that is still forming, the value moves with every new price until the close, so a cross you see during the bar can disappear by the close. A backtest and a live rule should both act on the next bar after a cross on a closed bar.

Is WaveTrend better than RSI?

Neither is better in general; they measure different things and trade at different rates. WaveTrend measures scaled distance from an exponential average and smooths it twice, so it crosses later and less often. RSI measures the balance of up and down closes over 14 bars and is bounded between 0 and 100. The only fair answer for your market is each indicator's cross rule tested with the same costs on the same bars.

What timeframe does WaveTrend work on?

Start on daily or multi-hour bars. The indicator has two averaging steps after scaling, so its lag is several bars at any bar size, and on very short bars the move captured between crosses is often smaller than commission and slippage. On longer bars the move per trade grows while the cost per trade stays the same. Test 60-minute bars before 5-minute bars, and compare the average trade after costs at each size.

What settings should I use for WaveTrend?

The common defaults are channel length 10, average length 21, signal smoothing 4, and zone levels at ±53 and ±60. Rather than searching for one perfect pair, test a range such as channel 6 to 20 and average 13 to 40. If a block of neighbouring settings behaves alike, the rule has something real behind it; if only one exact pair works, it fitted noise in that history.

References

Bailey, D. H., Borwein, J., López de Prado, M., & Zhu, Q. J. (2014). Pseudo-Mathematics and Financial Charlatanism: The Effects of Backtest Overfitting on Out-of-Sample Performance. Notices of the American Mathematical Society, 61(5), 458–471.
Sullivan, R., Timmermann, A., & White, H. (1999). Data-Snooping, Technical Trading Rule Performance, and the Bootstrap. The Journal of Finance, 54(5), 1647–1691.
Park, C.-H., & Irwin, S. H. (2007). What Do We Know About the Profitability of Technical Analysis? Journal of Economic Surveys, 21(4), 786–826.