#037: HOW DO AUTO-ADAPTIVE INDICATORS WORK?
One of the first challenges that every trader has to face in the beginning is setting up the period of indicators. Almost every indicator is calculated from the last several bars in the chart and the indicator period defines how many bars should be used for the calculation. For every period value, the indicator looks different. What is the right approach? And can auto-adaptive indicators help us?
First, let me tell you something about the period setup of indicators.
Basically, there isn’t any general, universal, recommended value (neither it is the value 14 that is used as a default value for many indicators). The optimal period for each parameter depends on many factors, like timeframe, expected length of trades (scalping, short-term, medium-term, long-term trades, ...), or even the computer optimization of the period. Generally speaking, we can say that for short-term trades the optimal period is approximately 2-20, the medium-term 21-50 and for long-term trades it is at about 51-200. But it really depends on the specific situation, indicator, system and the timeframe. Often it can be beneficial to combine different periods in one system - for example when one indicator is used with one lower and one higher period - to get the short-term, medium-term and long-term view on the market. In general, you need to remember that the lower is the period, the more market noise you will get - you can filter this out by looking at the higher period (or timeframe) to get a more complex view on the market situation (e.g. power and direction of a trend).
There is quite an interesting theory about the best period published by Perry Kaufman in 1995, who observed following:
When the market trends, it is mostly strong and clean move (I disagree with him on this point as it also depends on the timeframe and other circumstances), which doesn’t contain too much noise. In that case, we can work with lower periods of indicators.
When the market doesn’t trend (it is choppy), charts contain a lot of noise and it is much better to use a higher period of indicators.
Perry Kaufman also advanced from theory into practice (as one of few) and created an indicator (which I consider to be one of the first, or maybe even the very first auto-adaptive indicator), called Adaptive Moving Average (abbreviated to AMA or also KAMA), which solves the issue of the optimal period in a new, original, way - it dynamically changes the period and adapts to the situation in the market - depending on if the market is trending or not. Creating such indicator isn’t complicated and AMA (or also KAMA) is a standard part of many trading platforms.
Auto-adaptive indicator construction
When constructing auto-adaptive indicator, you need to add to the “standard” indicator one additional component - the part that will tell you if the markets are in trending or non-trending phase. There are many indicators that can provide this information, but Perry Kaufman decided to use another from his own indicators, the one that he calls Efficiency Ratio (ER). This indicator fluctuates between 0 and 1. The closer it is to number 1, the more the market trends, the closer it is to number 0, the less the market trends.The second step is quite simple - we use any of the moving averages (Kaufman uses modified EMA) and choose the range of the values that should be used for the period - let’s say from 2 to 50. When connected with ER indicator, the auto-adaptive version of the moving average uses higher values of the pre-defined range (in our case values close to 50), whenever ER indicator gets closer to 0 (when it reaches 0, the EMA period will be 50). This is because there is too much market noise and low period values are not suitable. On the other side, the lower periods will be automatically used for EMA every time ER gets closer to value 1 (when it reaches value 1, the EMA period will be 2).
As you can see from the example above, the EMA values are not fixed, but they dynamically change in the pre-defined range (in our case 2-50), according to the behaviour of the market.
In practice, the setup of auto-adaptive indicator looks quite simple. For example, AMA has 3 parameters:
The first parameter is the period that should be used for ER indicator calculation. The second and the third are the minimum and maximum value of the EMA period that will automatically adapt to the current market situation (based on ER indicator).
How does it look on the chart? Let’s have a look:
On the 1-minute chart of e-mini Russell 2000, we can see 3 moving averages. The blue EMA is using period 2, red EMA using period 50 and yellow AMA with setting 2-50. You can clearly see on the picture when in the choppy, non-trending situations, AMA is closer to EMA 50, and when the market trends, it is closer to EMA 2. Everything works really well and the indicator is really auto-adaptive - EMA adapts to the current market situation without any problems.
Other auto-adaptive indicators
We can build almost any auto-adaptive indicator on this logic. Unfortunately, such indicators are rarely standard part of the trading software, and you need other auto-adaptive indicators search on the internet, or program them in given software - luckily in these days this is not an issue and you can create your own indicators on most of the trading platforms. Basically, the sky is the limit - this article is about the auto-adaptive version of EMA (called AMA or also KAMA), but you can use the same principle basically for any other indicator, oscillator, etc.
It is a universal approach that can be used almost on any other indicator.
I can confirm that I really have good experience with AMA indicator and I consider it to be one of the best moving averages. This idea that you can dynamically (and also automatically) change the period of any indicator based on the market behaviour, is really good. And I plan to make more research on other versions of auto-adaptive indicators (that I have already prepared). But at the moment I have more important trading related stuff to complete, so it can take some time before I get to it. Still, I can really recommend the auto-adaptive indicators as a valid approach how to improve your trading.