Analysis of the Fringe AI.
Table of Contents
Introduction.
For this analysis I will take the 2 tutorials, I made for this purpose. There will also be additional criteria that I did not put on the tutorials, because they are not important.
Analysis Criterion and Strategy test analysis.
Fringe AI is designed by an artificial intelligence. In general, artificial intelligences use 80% of the data to train and 20% to be tested. In general, EA developed by an intelligence have a concern that it is over-optimized in backtesting, but does not work over time.
I would like to emphasise that it is your responsibility to purchase the product, you will have 7 days to withdraw and be reimbursed. Gains and losses are not the author’s responsibility, all the criteria to avoid a bad decision are present, it’s up to you to be careful.
As explained in my tutorial, I will start by preparing my backtest on the Metatrader 5 software by downloading the demo.
To do my tests, I will use these parameters and I will do 3 tests over periods that seem to come back often. The trading systems will be tested over the entire market period, from 2010 to today and from 2015 to today. These time units come back regularly and I think it is good to compare certain criteria over several periods to avoid that the trading system is optimised for one period.
Test results
A colour code for the following will make it possible to differentiate these 3 periods.
You can find the 3 graphs and their 3 backtest reports below.
Opening the image in a new tab will allow you to zoom in on the image and see all the analyses that will follow.
Entire history
Colour code : #000000
2010.01.01 to 2023.01.09
Colour code : #0514FB
2015.01.01 to 2023.01.09
Colour code : #8000F9
Analysis criterion
During the test the curves do not present a martingale or grid type strategy.
We note that each open position has an SL and a TP systematically and that the ratio between the SL and TP is greater than 1, which means that each strategy aims to earn more than what is risky.
We can see in the inputs parameters allowing managing the size of the positions to risk a percentage of the capital and no more.
Strategy test analysis
I’m going to use the strategy analysis tutorial to analyse where the strategy fits in the outcome criteria.
Average Win / Loss
You just need to have a ratio between the average win divide the average losers.
This reflects the risk of each position, in general one wishes to gain more than one has risk.
1.80 = 36 846$ / 20 404$
1.68 = 37 643$ / 22 284$
1.53 = 32 677$ / 21 245$
Conclusion Average
For all the periods, the ratio remained above 1 and the difference between the periods shows stability in the ratio.
Profit Factor
Profit Factor = 5.33
This is too ambitious in real life.
Profit Factor = 5.35
This is too ambitious in real life.
Profit Factor = 5.45
This is too ambitious in real life.
Conclusion Profit Factor
Le profit factor montre que c’est trop ambitieux.
Recovery Factor
Recovery factor = 205.52 is more than 10, you can consider this system stable and use it with a bigger capital.
Recovery Factor = 203.93 is more than 10, you can consider this system stable and use it with a bigger capital.
Recovery Factor = 176.85 is more than 10, you can consider this system stable and use it with a bigger capital.
Conclusion Recovery Factor
The Recovery Factor is simply huge!
Sharpe Ratio
Sharpe Ratio = 6.30 a high value indicates that the probability of obtaining a loss in each particular deal is very low. Very good.
Sharpe Ratio = 6.91 a high value indicates that the probability of obtaining a loss in each particular deal is very low. Very good.
Sharpe Ratio = 8.65 a high value indicates that the probability of obtaining a loss in each particular deal is very low. Very good.
Conclusion Sharpe Ratio
For all periods the sharpe ratio shows very good risk management.
Conclusion
All the numbers are huge, even the profit factor which is way too big. It looks like over optimization. If not, the numbers are incredible.
The best thing to do is to wait and see how the live signal develops over time. We have to see if in July 2023 the signal is still good.
Come and discuss the strategy on the community discord.