AI ProSoft has an ongoing focus at generating as stable a return as possible. Generating a 20% return is only relevant if it can be achieved as a result of many trades and if the relative as well as the absolute drawdown has been low. In the video link above, I elaborate over the past 4 weeks with the improved AI Robot that has the explicit goal of generating stable return.
Results of 4 weeks
It is now precisely 4 weeks since the thorough update of the old ASA_ML algorithm to its new version. The changes are that substantial that comparisons with earlier MT reports do not make sense.
The last 4 weeks can be summed up by:
- 313 trades
- 54% succes rate
- Absolute and elative drawdown of 2.54%
- Return of 4.4%
If the months to follow yield the same results it certainly starts to look good as the relation between the generated result and the risk (drawdown) is ‘healthy’. And this is only the start. As already mentioned, a new CorrelML algorithm is on its way and is to be activated within the coming weeks.
The new CorrelML algorithm so far has generated 19 winners out of 20 trades, and this algorithm can strongly assist in providing the wanted stability.
MT4 rapport for improved AI Robot
Use the following link to access trades from the previous 4 weeks. MT4 rapport 14 juli 2018 (37 downloads)
The road towards a stable return
The video contains a more detailed walk-through of the new CorrelML algorithm that is to be activated in the coming weeks.
The CorrelML algorithm is not the only algorithm at the drawing board. I have also researched new algorithms to analyze the opening of markets as well as the receiving of fundamental news – and in both cases, there seem to be interesting patterns in the market.
This is the case for sentimental (0-15 minutes before arrival of news) and it is the case as well up until 4 hours after the news. This knowledge can be used to optimize the existing algorithms, and it can be used to specify new algorithms of their own. Prototypes of such algorithms are tested at a demo account during this week with a success rate between 72% and 90% during last week.
A geopolitical ML Algorithm
I have also wished to develop an algorithm that is founded in assessments from top traders and experts. This approach has been delayed for now, as those so-called experts did very poorly in their guessing/forecasting during the last 4 weeks on Forex Forecast.