Currently, there are now 7 EA worth considering, the least pf is 1.54 and the rest very good pf values. 

Two EA with infinity Pf because haven't lost a single trade yet.   All these Ea trade at 0.1 lot size.

It is worth keeping an eye on these EA, since it already traded for 2mths.

I'll keep this testing going for the 3rd mth and then update you guys here.

After the 3rd mth, I'll create some "variants" EA with different SL or TP setting and I probably be using these EA and install them across different symbols.  For eg 93060 (GBPUSD) be applied in the other remaining 5 symbols etc.

However I will stop this project if I don't see anyone joining me to contribute cos I don't want to be working hard while the rest of FSB community sit back and take without giving back to the community at large, I don't reward such behaviours and thus I'll stop if I think this project is not encouraging more participants.

I'll then return back to my own EA development and mind of own business, nothing personal.  I already have many successful portfolio, so I don't really need anyone's EA (that's why I don't mind being the 1st person to give out the EA).

I'm not asking those of you who manually work your EA to share cos I won't do that myself, so neither will I expect anyone to reveal their trade secrets here.  However, if you have been using EA Studio to generate those EA, honestly, if 10 people were to work together, with the same effort they can achieve 10 times more efficiency.  There is power in synergy.

I'm here to create an environment whereby FSB users can join forces to help each other to save time and resources.  However, if most of you prefer to work on your own, that's completely fine with me (no offence at all to me, I respect your preference).

If you have any suggestions do let me know.  If you are interested and wish to protect your work by only sharing with those who contribute instead of openly sharing your EA like I did. Then we can create a Ryver team whereby only those who want to contribute has access to members uploaded strategies.  I'm open to any types of collaboration as long as members can start come together to speed up each other's success.

Regards
Hannah

First of all...Sorry for my bad German. Your English is fine...great, in fact...much better than my German! In fact, I don't know any German, so I'll just type in English. Never a need to apologize.

Using EA Studio is going to be a hit-or-miss process. There seem to be two theories as to the best way to proceed.

The first is to generate over short, most recent time periods and only In Sample. Set several generators into motion for a certain length of time and keep only the top few results of each instance. Place these onto a test server and see how they do. The idea is that a portion these strategies will be profitable over the short-term, until the market corrects the underlying bias that these relatively simple EAs exploit.

I cannot state one way or another whether or not that may yield any significant results. (I have chosen to start with the Second Method, below) It seems that we can search so many strategies so quickly and come up with such a mass of possible strategies that some of those strategies MUST be functional. Or perhaps we are searching so many strategies that we find mostly heavily curve-fitted strategies that fail with deployment.

Determining how to weed out these strategies and when to place them into a live account will be extremely difficult.

At any rate, if we expect these EAs to fail after a certain (relatively short) period of time, we don't have much time for live testing. We need to determine which EAs are profitable in as short of a time span as possible. I believe there will be a problem with this approach in that many strategies will appear profitable in the short term, just by dumb luck. If we have only a small sample of trades, nearly any strategy could be profitable. So for every 100 EAs we are testing, we may get more false positives than actual profitable EAs within a reasonable testing period.

You will inevitably get a bunch of EAs at the start of the test that are in profit, and nearly all will inevitably turn to a loss. Some may even "perform well" for a longer time; don't be too quick to call those "winners". Small sample size leads to many erroneous conclusions!

The combination of Portfolio Expert and MT4Tracker (https://forexsb.com/forum/topic/6728/mt … portfolio/) (thank you sleytus!) is an absolute Godsend for this method, and the second, which is (currently) my preferred method...

Second Method: Run over a large amount of data and include an OOS period. Visually inspect the OOS period to ensure it is "acceptable". Hopefully, this leads to more enduring EAs. Optimize over a portion of the most recent OOS period. I will need to experiment to determine what the optimal period is for optimization. It will actually vary with each EA, but there is not way with EA Studio to effectively determine what period that would be.

For my method, I've decided that it worthless (for me) to generate over less than M30 (my preferred TF). I set 14 instances to generating overnight and they examined a total of 1.7 EAs; after deleting a few for bad OOS, I was left with...four EAs. Not worth the time (at least not at this point!).

I'm generating on M30 over 178,398 bars. From From 2003-05-05 03:00, to 2017-08-11 23:30
50% OOS

My Acceptance Criteria are currently as follows:
Max. Amb. Bars: 10
Min. Net Profit: 10
Min Trades: 600 (a little less than one trade per week minimum)
Min. SQN: 1.6
Min. W/L: 0.30 (never gets this low anyway!)
Max Balance Dev: 15 (too high; will modify. This is a VERY interesting criteria.)
Max Stag%: 15
Max DD%: 10

I'm going to compare this with generation from 2014 to the present (less OOS) and see what happens.

Keep talking sleytus, I'm edging closer to your way of thinking!!

What Time Frames are you generating over and what Acceptance Criteria are you using?

I like your "100 trades" goal. How much data are you finding that it takes to reach that goal? And how quickly do strategies generate that you are comfortable using?

I still cannot see a way to weed out the majority of strategies that could perform poorly in the immediate future, leaving mostly ones that have good profit potential. This is ultimately the most important aspect of the generation, so how do we start to approach this?

WARNING:

I've recommended the use of Fx Synergy for it's Scaling In features.

I've use the trial to test out Fx Synergy (to ensure I can create my expected workflow in Fx Synergy).

I wrote to the customer support yesterday and was informed that Scaling In function can only be applied to MANUAL TRADING.

In another words, you need to click on a specific open position and apply Scaling In for that particular position. 

I can't be automatically applied to positions opened by your EA.  The upgraded version will able to apply Scaling In automated trades as claimed by the customer relation officer.

However, I may still go ahead to buy it before the upgrade version is made available to the public (before the end of 2017) because the price will be doubled by then.

For those who are interested to get Fx Synergy, kindly PM me cos I'm trying to get a group discount.