Salve a tutti,
with this post I’d like to explain the different trend scoring strategies that I developed and implemented in Crescendo and Crescendo Ultimate.
This is the first of a set of posts that are made to explain in details the new strategies so that you’ll be able to use Crescendo not only with the default settings (that will be those “good for everyone”) but also as a tool for automatic trading.
Crescendo, like no other commercial EA out there, is not a “simple EA”, but also a real tool for automatic trader. Some of you, also thanks to Crescendo, are starting to discover a new way of trading and becoming more and more what’s called a “quantitative trader”.
First, why trend scoring? Giving a “score” to the trend helps mainly in filtering entries. You can decide to enter trades only when the trend is very strong, for example. The more you filter them, the less trades you have but the more “surefire”. On Crescendo Ultimate scoring is also used to exit trades via a trailing strategy that I’ll explain in details in the next posts.
Actually I developed 6 different strategies for trend scoring but many many more can be developed with the same concept.
Below there’s a picture showing a comparison of the 6 algorithm I actually developed and tested:
The trend score goes from -5 (strong short) to +5 (strong long).
To do so I calculate the value of an “indicator” 5 times (different time frames and/or different periods) and based on a set of rules we give a +1 if the indicator gives a bullish trend, and -1 if it gives a bearish trend. Summing up the 5 values we have a score that can go from -5 to +5.
1) Hull Moving Average
We evaluate if 5 Hull Moving Averages are going up or down. There are a couple of approaches to that.
The first one is use the same period (like 9 or 10 periods) to different timeframes (like 1H, 4H, 1D, 1W and 1MN).
The second one is to use the same timeframe (like 1H) but with different periods (like 9, 25, 36, 169 and 441).
To evaluate if the HullMA is bullish or bearish we compare the value of the HullMA of the present bar (or the last closed one) and the bar before it. If the value did increase then the HMA is bullish, if it’s decreasing the it’s bearish. Very simple
I won’t go more in details why I did choose the HullMA but in simple words it is because is a fast moving average and very smooth. So, fast (not much lagging) and thanks to smoothness it doesn’t give many false signals.
2) Average Hull Moving Averages
We create a new moving average coming from the average of the value of the 5 HullMAs. So we sum the value of the 5 HullMA and divide by 5. We do that for the current bar (or the last closed one) and for the previous one.
This time we won’t have a score coming from the value of 5 HullMAs but from only one calculated on the average of the 5 HullMA. If the new composite HullMA is pointing up we give a +5 score, if it’s poiting down then we give a -5 score.
So, with that second algo we’ll have only two scores: -5 and +5.
It’s like the 1st algo (Hull Moving Average) but we use the “speed” of the price. In reality we use a smoothed version of the speed of price (smoothed using the Hull algorithm).
The score is then calculated giving a +1 for each speed if it is increasing and -1 if it’s decreasing.
Speed is the perfect momentum indicator and works quite well as a trend indicator.
Same as for Speed but using acceleration instead of speed.
Acceleration is the speed of speed and so it’s a very good momentum indicator with some “forecasting” ability. Sometimes even too much
5) Average of Speed and Acceleration
In this case we calculate 5 speeds and 5 accelerations. We sum them (so we can have scores from -10 to +10) and we divide the value by 2. That way we’ll have scores from -5 to +5 but unlike with algo 1-3-4 we also have odd values: -5, -4, -3, -2, -1, 0, +1, +2, +3, +4 and +5.
Why use both speed and acceleration? I decided to try with that to smooth the “forecasting” coming from acceleration. Also the fact that gives all the values from -5 to +5 (0 included) gives a better way to read the trend.
6) Parabolic SARs
Like for algo 1-3-4, but using the Parabolic SAR indicator.
This is the last indicator that I’m using for trend scoring and it is giving very good results.
Difficult to say which one is best. Based on my experience it depends much on the the way Crescendo is set (fixed target profit or trailing, currency pair used, etc).
I’m thinking about taking out the trend scoring algorithms from Crescendo, and let it rely on an external indicator (like the one shown in the above picture) and provide you with the code of it. That way if you want to add your own trend scoring algo or modify the existing ones you’re free to do that. And also I can provide you with new algorithms simply by sending you an updated version of the “trend indicator”.