Zamolxis report 2016

11 months passed since Zamolxis was launched. Before talking about its performances or what’s left to be done to make it even more profitable, I will say once more that forex should be treated as a business instead of a getting rich quick scheme.

For example, suppose you buy something cheaper, from China and resell it on ebay at a higher price.  Your net profit is 30%, I think it’s a decent amount. But there are some risks and expenses involved: website and hosting expenses, SEO expenses, rent expenses, employees plus the risk of not selling all your stocks.

If you trade forex you don’t have all that expenses but there are a few risks: drawdown, slippage, spread, VPS crashes, broker dependency. In every business, with only a few exceptions, risk is directly proportional to the amount of money you want to earn.

So far Zamolxis produced 70% per year. Let’s say it keeps producing the same amount of money for the next 9 years. Our starting balance is $10,000. Every year, the lot size increases in such a way to earn 70% per year out of your total balance. The results:


Impressive, right? Never underestimate the power of compounding. And, with a little patience (time goes anyway) this target is very achievable.

So far Zamolxis went almost as expected: 5% per month, 51% trades won, 49% trades lost, the percent of losing trades is 2% lower then the percent of winning trades, and, what’s important, the profit (TP) for every trade is greater that the loss (SP). All these makes Zamolxis very safe and profitable.

There is only one weakness: the lot size is increased after 4 consecutive losses and if the expected number of consecutive losses is reached then the strategy is invalidated and the robot should be stopped. It didn’t happen so far (although on some brokers the number of consecutive losses was higher than other brokers, but that’s about the degree of broker dependency) but there is no guarantee this won’t happen in the future. In order to prevent that, there is a built in function which allows the trader to set their targets. However, I consider this as a weakness (every strategy has its weaknesses) and I’m planning to eliminate it without altering the profitability. I’m working hard to eliminate this without altering the profitability or the faster recovery and I think I know how to achieve this, please read the TODO list.

Zamolxis is one the first public AI (Artificial Intelligence) forex robot. Unlike other AI based methods, mine works great, as everyone can see. Bayesian filters properly built can predict the market with good accuracy, this was the first phase of the experiment. The second one consists in adding more neural network and bayesian filters and slowly remove the recovery function because the main goal is to keep the profitability at a constant level without prolonging the recovery time.


There is one downside of using AI, the optimization is very time consuming and can’t be done in MT4 optimizer. Here is how its done: we first need a large set of winning trades (winning trades only) and for every trade we need to know the initial entry and exit conditions (all timeframe volatility, various indicator status, previous open and close and so on). Then, all these data should be tested against future unseen data and validated only if the percent of winning trades, losing trades, consecutive losses and drawdown remains the same. This involves a huge work, but I’m willing to take it to the next level.

In the following months I’m going to work on releasing Zamolxis 2.0 for my customers. I will rent an amazon cloud and start computing until I come up with a truly forex brain composed of bayesian filters and neural networks used for classifications. The final goal is to eliminate the mentioned weakness thus making it almost risk free without lowering its profitability rate of 70% per year.

If you are interested in my work and wanna be part of it and support it, please subscribe to my newsletter and buy the current Zamolxis version.