About a month ago, I’ve started to run a few Forex EAs I own in the hope of finding the right combination. I put them on Cent accounts rather than Demo accounts since the latter is immensely inaccurate and cannot be taken as a real measure for reference. How many EA’s are winners in demo but total losers live? I’ve seen far too many schemes like that, try avoiding demo testing, not worth it and you also end up losing months of tracking/monitoring. I was very lucky to have a friend of mine give me a spare PC which I converted as my own dedicated trading machine. No screens, no keyboards, no mouse, just the power and ethernet cable which feeds the PC.
Here is an overview of my 4 systems currently into testing:
S9 is a bit of a rocky road at first sight but does look promising. The main factor which I care in choosing a system is its drawdown level, growth rate and # of trades. By filtering the data, you actually end up with this:
Isn’t this way more appealing?! Out of 20 trades, 17 were winners(85%), we managed to gain 4% while risking 0.32% in total, that’s a +10:1 Reward/Risk ratio. Sure, my JPY pairs are increasing my exposure for unnecessary reasons but if you look closely at the graph, none of them move(open/close) positions at the same time which means that from the data, when one pair is on ”standby” another JPY pair is putting a position. Since not all 4 pairs are ”activated” at the same time, usually 2 or less, this means I have continuous trading setups. I am not sure how I will manage this later, either take the best pair and increase its lots or keep minimal lots across 4 correlated pairs.
I wanted to push my research a little bit more and thought it would be a good thing to compared backtesting data vs live data. Since I already have almost 2 months of live data I can just compare with the actual data from Myfxbook for the same period of time, December 10th – January 26th.
From a very quick analysis, we can see that if we choose the same pairs as the one selected on my testing account on myfxbook, we end up with +44$ rather than the reported +20$. Can we say performance from backtesting and live testing is as much as 2x off? Maybe, but duly noted. Now if we also compare the drawdown, on live account we have 0.32% while backtesting gives us 0.18% that’s again almost 2x off. Again, duly noted. I think what will become an interesting case of study is if I continue doing this cross reference analysis on a weekly basis. To me, the results tells me that performance in backtesting will be at best, twice as much than in real life and the drawdown value will be twice as less than the real expected live value. I will keep you guys updated next week and see how performance correlates.
S10 is my next flagship EA. Yes, it uses martingale but there is a multitude of entry/exit strategies used which so far has prevented it from crashing where other live accounts have suffered these past 2 weeks.
You have to take into account this was put on a 500$ and generated +250$(+50% gain) in so-called Cent profits in a bit lest than 2 months. The max drawdown level was at -34%. At first I didn’t believe it because I thought the stats were wrong or MyFxbook had a bug because I regularly check the ”health” of my machines, ensuring everything works correctly and I have never seen it go more than 10-15% drawdown so 34% was shady. However, this week, GBPAUD made a big bullish move and my bot was on the wrong side all the way at the bottom.
Here was the play:
This essentially means that if I had put 2000$ rather than 500$ like this current testing account, my max drawdown would’ve been 8.5% rather than 34% and my gains 12.5% instead of 50%, within 2 months. Sounds more reasonable no? I very much think so. If I’m putting my own 2000$ and because it is not a big amount, I am willing to risk/have a drawdown ~20%. With 2000$ in play, based on a 300 pips bad move, this means I can handle 2 pairs gone wrong to hover at around -20%. Now, how often does it occur when there are 2 pairs doing +300 pips move on the same week? Most likely more often than you think so. We always have to think in the worst possible case. If you go to any school of engineering, they will teach you how to use processes/methods to ensure there is a minimum standard of quality being done which in this case is the worst possible case, basic math, rough approximation. Here we apply the same, the column “Drawdown % difference” is the difference between the previous data from drawdown. D3=C3-C2 and so on. From the table shown, I believe 2000$ or 3000$ is the sweet spot as they both share the same ratio. 2000$ would be more considered as aggressive while 3000$ would be more safe, the numbers don’t lie here. At 2000$, we are looking at a very possible occurrence of -20% drawdown. at 3000$ we are looking at ~-15% drawdown, obviously the gains are inversely proportional. By choosing to start with 500$, it allows us to test its behavior with minimal amount of cash, rather than focusing on how much we can make.
Now to all the guys who knows someone who knows someone about how great of a Forex trader they are. Here’s some food for thought:
If the professionally audited funds are underperforming, how many do you really know who is legit? Not many. Don’t trust all the guys and their phone screenshots or IG screenshots. You are better off learning how to trade yourself and acquire a lifelong skill.
Global markets soared in the first half of 2017, largely due to the momentum gained from the November 2016 U.S. election results. The U.S. equity market posted strong gains across all cap ranges, with the S&P SmallCap 600® posting 22.47% over the 12-month period as of June 30, 2017. The S&P MidCap 400® and S&P 500® followed, reporting gains of 18.57% and 17.90%, respectively.
During the one-year period, the percentage of managers outperforming their respective benchmarks noticeably increased, compared to results from six months prior. Over the one-year period, 56.56% of large-cap managers, 60.69% of mid-cap managers, and 59.55% of small-cap managers underperformed the S&P 500, the S&P MidCap 400, and the S&P SmallCap 600, respectively.
Another quite useful document where I found how much money professional funds are willing to risk. Turns out its around ~15%, not too far off my own 20%. The Barclays Currency Traders Index represents 58(as of 2017) programs/funds trading solely currencies and being audited by Barclay’s. Of course there are more firms doing audits but this is good enough information. You will also notice they have about 0 correlation between S&P 500, US Bonds or World Bonds performances, it’s a different ball game. Another piece of information which they make available is the historical data regarding the number of funds registered annually.
From 2000 to 2008, the number of funds kept increasing and since the crash of 2008, it has surprisingly been decreasing every single year for the past 10 years.
This should already tell us that if you don’t have proper money management and trade setups structured you have basically 0 chance. You must be prepared before putting money in the market.
My goal is to put 2000$ by April 1st and let it grow completely automated. By combining both systems S9 and S10, I have greater control of my expectations based on live forward testing and backtesting using 99.90%(which doesn’t mean anything btw). Backtesting is to validate a proof of concept. If it doesn’t pass backtesting it most likely won’t pass demo and even less forward testing. If it passes backtesting and demo, it doesn’t mean live testing is a go.
Having a Cent account is the only close thing to reliable measure against putting down your hard earned money as experimental.
Don’t forget to sign up using the form below!