UniversalMA | Trend Testing | Comparison with MA

The Moving Average technical indicator is able to predict future trends. We proved this in one of our previous articles.

But the classic MA is a very outdated indicator. Will modern adaptive smoothing algorithms be able to better identify the trend? Or is it still too early to scrap the good old MA and nothing will be better?

In this article, we will try to understand these issues.:

Let's create an expert using the UniversalMA indicator.
Let's calculate the “RBTI” rating for the UniversalMA indicator.
Let's compare the ratings of the indicators and analyze in detail their strengths and weaknesses.

Content

What is UniversalMA?

UniversalMA trend indicator of technical analysis

From the point of view of calculations, UniversalMA differs from the usual moving average by additional averaging methods and an expanded list of data on which this averaging is carried out.

We will not go deep into issues that do not directly affect our trading profits. Therefore, we will briefly consider the main averaging methods added to UniversalMA.:

Adaptive Smoothing JMA (JJMA) is a method that automatically adjusts smoothing parameters depending on market volatility.
JRSX Ultralinear Smoothing (JurX) is an advanced smoothing method that provides more accurate trend tracking while maintaining smoothness of the line.
Parabolic approximation (ParMA) is a method of smoothing data using a parabolic curve, which makes it possible to better track the acceleration of price movement.
The Tilson Algorithm (T3) is a method based on a combination of several exponential moving averages with different periods to improve the filtering of market noise.
Variable Index Dynamic Average (VIDYA) is a method that uses the Average true range (ATR) to adapt to market volatility, making it particularly effective in a volatile market.
Adaptive Moving Average (AMA) is a method that automatically adjusts the averaging period depending on market volatility, providing an optimal balance between sensitivity and smoothing.
FATL (Fast Adaptive Trend Line) is a fast adaptive trend line. Each bar has its own unique weight assigned.

Configurable parameters of UniversalMA

UniversalMA example of how the Step parameter works

The main parameters are inherited from the Moving Average indicator: The period, the Method, the type of price to build, and the Offset.

We also have new options in UniversalMA.:

The phase parameter, which varies from -100 to +100, affects the quality of the transition process. It should be noted that this parameter does not work with all types of indicator smoothing. But this is not important for us, since the tester's capabilities in MetaTrader 5 are enough to try all the options.

Step – The minimum step of changing the value of the indicator line. It leaves the indicator value the same in case of insufficient movement. Visually, pronounced steps form on the indicator line. It is indicated in the paragraphs of the current symbol.

Let's calculate the RBTI rating for the UniversalMA indicator

The expert is ready for the tests. It, as well as a set of settings, can be downloaded in the Downloads section on this page. When we tested the classic Moving Average, we found more profitable trend interpretation methods. It would be fair and correct to add the application of these methods to these tests too.

The number of parameters that need to be selected has become quite large. Therefore, we will have to resort to genetic optimization. Fortunately, the strategy tester built into MetaTrader 5 has such capabilities. It makes sense to repeat the optimization 3 times, this will be enough for the algorithm to determine the best parameters. Gold (XAUUSD), timeframe H1 and 13 years of real history - let's go.

UniversalMA the result of working as a trend indicator

Learn more about the calculation method RBTI

Let's select the parameters on the site for optimization

The results obtained are worse than those of a regular moving average. How can this be, because inside UniversalMA there are the same settings as in a regular MA?!

It's all about genetic optimization. Despite the enormous power of this tool, a complete search of the parameters finds the absolutely best parameters, and the genetic algorithm is average.

However, this does not affect the rating calculation in any way. On the contrary, it gives indicators with fewer settings an advantage. This is well deserved, since the fewer the optimization parameters, the less likely it is to retrain.

When we tested the same indicator using the RBTI v1.0 method, we got a phenomenal result on the AMA averaging. So, he's still there, but we just didn't find him this time. And even if they had, its extremity would have been smoothed out 3 times. This is one of the reasons why we introduced RBTI v2.0.

The parameters that showed the best balance and IMP_opti_maxBalance = 3.59

Period = 111
Mode = T3
Price = TrendFollow_2
Phase = 80
StepMA = 10000
Shift = 5
OpenStrategyBuy = color change
OpenStrategySell = color change

The parameters that showed the best ISP and IMP_opti_maxISP = 3.59

Period = 111
Mode = T3
Price = TrendFollow_2
Phase = 80
StepMA = 10000
Shift = 5
OpenStrategyBuy = color change
OpenStrategySell = color change

Parameters that showed the best RF and IMP_opti_maxRF = 3.56

Period = 80
Mode = JurX
Price = Open
Phase = 85
StepMA = 5000
Shift = 10
OpenStrategyBuy = ghost cross ma
OpenStrategySell = color change

The forward test showed a drain on the flute and an increase in the trend area. It is quite normal for a classic trend indicator.

Average_IMP_opti = 3.58

IMP_all_maxBalance = 3.48

IMP_all_maxISP = 3.48

IMP_all_maxRF = 1.43

PC = 78%

RBTI = 2,8

Conclusions

1) In the competition to determine the trend, the Moving Average indicator lost to its descendant UniversalMA. But to be honest, he had no chance of winning, at best a draw. Since all the functions from the classic Moving Average are fully present in UniversalMA.

2) Genetic optimization is not able to find the absolutely best parameters. When possible, you should strive (it makes sense to increase the iteration step) to use a full iteration of the parameters.

3) The best averaging method for trend identification tasks is T3

4) The Step parameter proved to be in demand and fully justified its presence in the indicator.

5) The additional trend interpretations we suggested in the article about the classic version of the Moving Average turned out to be the best here.

6) The UniversalMA indicator has been assigned an RBTI rating of 2.8. The result will be recorded in the leaderboard.

Battle for Trend or Moving Average vs. UniversalMA