Table of Contents
How to analyse a robot test?
Once we have analysed certain criteria presented on this page, we are led to take an interest in a more concrete analysis. To do an analysis, simply download the demo of the robot in question and run the test for this robot in the Metatrader 5 software.
To start the analysis, it is good to put yourself in the conditions close to reality, for example the leverage and the initial deposit that you will use in order to determine if your deposit will be sufficient. To avoid slip-sensitive bots I set a random delay. Do not forget to enter the currency of the trading account, because if a conversion must be made it is better to simulate it.
Once these parameters are ready, I personally set the modelling to 1 minute OHLC, this greatly accelerates the speed of the test, but you will be less precise.
I like to analyse over the largest period possible so I select the entire history to date. I assume that an effective trading robot must be protected against all the economic crises that have occurred in recent years. If the robot fails to overcome a passing economic upheaval, it will not handle the next upheaval that will hit the stock market.
Before starting the test, I find out about the robot and these parameters, each robot has different parameters and different operations. It is important to respect the author’s recommendations for a test in the best conditions, I am not currently trying to improve anything.
Run the test
We can run the test.
There are 2 possibilities, the test stops before the end, because the trading
account has lost all the money or the test ends at the end. If the account
exploded I think it’s because the robot falls within the criteria that I
explained in this tutorial: Analysis criterion and therefore they are not robots that have
protections, it is dangerous to use them.
In this tutorial we are interested in the robot that arrived at the end and
which presents a test report that will help you understand and analyse, because
I have already noted that the authors want to highlight some of these figures
without let them themselves know that it is not good that this figure is
Everything you can find on the official page of Metatrader 5: Testing Report
The texts that will be in italics will be a copy/paste of the page presented
above, but I want to avoid going back and forth between these two pages if you
don’t have 2 screens.
I will take as an example the image of the test present on their page. I will go over the points which are the most important.
It is important to consider the balance and funds drawdown.
Ratio of the gross profit to the gross loss. A value of one means that these parameters are equal.
In our example, the profit factor is 1.12 To calculate this profit factor, divide the Gross Profit by the Gross Loss.
In our example:
Gross Profit: 1,310.51
Gross Loss: 1,168.49
1,310.51 / 1,168.49 = 1.12
The profit factor or profit factor in French is a performance index of a trading strategy which is calculated by dividing the gross profit by the gross loss (commissions included) for the whole of a given period. This performance measure relates to the amount of profit per unit of risk with higher values indicating a profitable system. A ratio between 1.25 and 1.75 represents a mark of minimum security, but will not be able to support the various expenses over the years and transactions (brokers’ commission, platform fees, taxes, bank commission, data expenses Steps…). A profit factor of 2 is the minimum to have if you want to survive in the financial markets. On the other hand, if the ratio is greater than 4, this is too ambitious in real life. In addition, it will be necessary to privilege a profit factor to the real capital gain. There are still some disadvantages in the use of the profit factor: it does not show whether the strategy is profitable in relation to the total capital nor the maximum risk incurred nor the size of each gain or loss. It is for this reason that it is recommended to use other measures in combination with the profit factor to have a more coherent overall picture.
The value reflects the riskiness of the strategy, i.e. the amount of money risked by the Expert Advisor to make the profit it obtained. It is calculated as the ratio of gained profit to the maximum drawdown.
Recovery Factor (RF) is a statistical indicator, which is calculated through dividing the aggregate profit by the maximum drawdown. This indicator is important in assessing the trading system stability.
In fact, RF shows how fast the trading system may exit from the maximum drawdown.
- If RF is more than 2, the system has a certain profit margin and you might consider this system for trading.
- If RF is less than 2, it is better to update it, since its stability is questionable.
- If the system’s RF is more than 10, you can consider this system stable and use it with a bigger capital.
In our example the Recovery Factor is equal to 1.47, to calculate it you must take the total net profit and divide it by the Equity Drawdown Maximal.
In our example:
total net profit: 142.02
Equity Drawdown Maximal: 96.58
142.02 / 96.58 = 1.47
Arithmetic mean of a trade (change in percents). Arithmetic mean of equity changes per trade. The arithmetic mean usually overestimates the profitability of a trading system as compared to the geometric mean. If the geometric mean implies the multiplication of results of each trade, the arithmetic mean just sums them. The value in percents is given in brackets. It is positive if the trading system is profitable. The negative value means that the system is losing.
Geometric mean of a trade (change in percents). Geometric mean shows by how many times the capital changed after each trade in average. The relative equity change is often a more objective estimation than the expected payoff. Capital change in percents is given in brackets. A negative number in brackets means that on the average the capital is reduced on each trade.
A classic measure which is commonly used to evaluate the performance of a portfolio manager, fund results or a trading system. The ratio is calculated as (Return – Risk-Free Rate)/Standard Deviation of Return. In the strategy tester, the Risk-Free Rate is assumed to be zero. The ratio values are usually interpreted as follows:
- Sharpe Ratio < 0 — the strategy is unprofitable. Bad.
- 0 < Sharpe Ratio < 1.0 — the risk does not pay off. Such strategies can be considered when there are no alternatives. Indefinite.
- Sharpe Ratio ≥ 1.0 — this can mean that the risk pays off and that the portfolio/strategy can show results. Good.
- Sharpe Ratio ≥ 3.0 — a high value indicates that the probability of obtaining a loss in each particular deal is very low. Very good.
Series testing (the probability of correlation between trades). The series testing allows to estimate the degree of correlation between trades and evaluate whether the trade history includes more/less periods of consecutive profits/losses than normal distribution implies. The detected correlation allows to apply the methods of money management and/or change the trading system algorithm to maximize profit and/or to remove the dependence. Both non-finding the real correlation and finding a nonexistent correlation between trades are dangerous. The Z score indicates deviation from normal distribution in the sigma. A value above 3 indicates that a win will be followed by a loss with the probability of 3 sigma (99.67%). A value below -3 indicates that a win will be followed by a win with the probability of 3 sigma (99.67%).