Appreciation and proposal(s)
Results 1 to 8 of 8
 1 Attachment(s)    

Thread: Appreciation and proposal(s)

  1. #1
    Junior Member
    Join Date
    Feb 2009

    Thumbs up Appreciation and proposal(s)

    Hallo people,

    this is my first post here, although I am a quite addicted reader of this forum, and I cannot wait longer for expressing my appreciation and many thanks to Funyoo to have founded it. I feel quite comfortable here in comparison to other cyberplaces, because I find a quite good match, as a convinced “systematic” trader, with the philosophy.

    I want to make some proposal and ask some questions.

    If I have not counted wrongly (a little index number, maybe?) in the “Top Free Expert Advisors” section of the site, just looking at the systems with a ratio > 1, there is a list with about 200 experts.

    To be clear: again congratulations and thanks for all this work, shared with the public! It is really rare finding a so well referenced and ranked list, and this forum, much better than others, contains usually very nice descriptions of the mechanisms behind the various scripts, besides the code. And a lot of testing. But I have some improvement to propose.

    First I am a bit concerned that the “ratio” used for ranking these systems ((Total net profit/Balance)/RDD) is not the best measure. Especially because it doesn’t really account for the difference between profitable systems and “suckers”. In addition, a simple average of such ratio over different testings, each one with its own set of parameters, currency, even TF, length and period of time tested… doesn’t make much sense, to myself.

    One observation (and see below for how I would select a system, now) is that probably a unique variable for ranking EA’s is not sufficient-meaningful. If this must include a mix of profitability and risk (and this can be a first approximation of a “good” ranking parameter), then we can discuss how to produce the better metric.

    And the second observation is that I think it should be better, for such a comparison, adopting very standardized testing procedures, such as a portfolio of some defined “model instruments/pairs”, a basket of defined time periods (depending on the “speed” of the system, but anyway consistent for similar systems), and a defined set of parameters. This can well be the best found after different approaches of optimization, but must be clearly defined, and, instead of making averages of different sets of parameters, it is probably better to keep just the best one.

    What I think it could be nice to do, as a collaborative work, is to help Funyoo in collecting and ordering data is some meaningful form. I started yesterday, just for fun, making an excel file. I attach it, but, please, it’s really just a draft and very very far from being complete. For each script I would include, besides the “metric” that can be chosen as the more representative for a ranking, also some other essential parameter useful for ranking and choosing a script, such as for instance profit factor, frequency of trading, maximal drawdown, and the like, but also kind of money management, kind of indicator/pattern/strategy used, risk factors (e.g. the absence of a stop loss), the "category" of the system (e.g.: trend-following, countertrend, channelling-ranging, breakout, adaptive...).

    This as far as the information is amenable to standardization across different systems. This list could then be linked also to single “EA reports” where more details can be included, always in standardized form. Good candidates are currency tested (each with their results), more analytical quantitative parameters (max profit/max loss…), some more information about mechanics, a description of options/parameters, some notes about the code (availability of source, easy/difficult to understand/modify) and everything that can be useful.

    My experience of yesterday is that, there is so much information about some of these systems, that compiling such a report alone would prove quite impossible. Together, with all the “experts of the experts” (LOL) contributing a bit, could become quite easy.

    Good. I go on in separate messages with the other topics.

    Hope this is useful for discussion. Hope not to hurt anybody and if I am to invasive I will go back to my reading chair!
    Attached Files Attached Files
    Last edited by fsprea; 08-27-2009 at 13:06. Reason: Added a category

  2. #2
    Junior Member
    Join Date
    Feb 2009

    Default Second topic: Optimization

    I am very glad to join a community where people like testing and optimizing scripts. Somebody is sometimes a bit pessimistic about these efforts, like Metatrader7. But you too, Metatrader, are convinced that the lack of success depends, besides of the "hard environment" (full of competition) where this business takes place, to inconsistencies and faults made by people that does not have the patience or the method to study things deeply enough.

    By common knowledge (most financial institutions trade automatically, at least for a protion of their portfolio) and navigation on several forums (this one comprised: well done Jezzer!) that profiting of the markets with algorithms is indeed possible. Would you, Metatrader7 (and anybody else), not agree with me that the main reason of failures is a not-methodical-enough, or not-"scientific"-enough, or simply not-patient-enough DEVELOPMENT of systems? Development is what starts when invention has taken place. Tweaking, testing, improving, optimizing, measuring and reproducing.

    Funyoo and anybody else here have a lot of merits, because just by sharing inventions (EA's & systems) they allow people (Like me! ) refining their knowledge of the market by giving examples of how a trading system can be structured and how different indicators/systems/methods allow following market dynamics. But, as several people here posted, being then profitable depends on several further factors, like making your own system (automatic or manual), apply it carefully, managing capital, or... further developing some of the potential gems that are available here (and elsewhere). So: I strongly believe in this development path.

    Recently I had the fortune of reading R. Pardo's "The evaluation and optimization of Trading Strategies" (2008). Do anybody else know this book and want to comment about it and its contents? I found it extremely interesting although not exaustive.

    My proposal would be to set up maybe a sub-forum (or a super-thread??? LOL) where optimization can be discussed in all the facets that it is composed of. Not just data integrity... Metatrader is not immediately the best tool for such a task, but, here as elsewhere, with all the brilliant coders we have, we can maybe also try improving it.

    Just as an example of the many topics that can be discussed (each one is huge!):
    • The design and optimization process
    • Tools for optimization
    • Stucture of a trading system: DirectionalDevice, TradeManagementDevice, RiskManagementDevice
    • Historical simulation: data and algorithms
    • Preliminary testing (profitability): the "backtesting" as meant ususally
    • Objective functions
    • Grid optimization or more advanced methods (e.g. genetic algorithms)
    • Optimization: range and steps of parameters e.g. The quest for robustness
    • Walk-forward analysis
    • Metrics of system performance
    • Overfitting: symptoms and cure
    • From backtesting to live: measuring performance, keeping it under control and decide when it is time to stop or re-tweak a system

    Would anybody be interested in discussing and further learning these topics? Is it just my flawed brain?

    A WARNING about myself because I must be realistic and humble. I am just a "part-timer": have a "regular job" and also a family. I hope to become a full timer as soon as possible, but this is not the case, yet. My ideas, plans and dreams badly overcome my time and energies. So: it is possible that, after "launching" some ideas, I will not so fast in developing them. But I feel that, even only this "launching" can be useful for me as well as for others, and that there are people around, much better than I, that can develop these ideas a lot.

    To the next (last) topic

  3. #3
    Junior Member
    Join Date
    Feb 2009

    Default Third: where to start from

    finally, as I would also make some honest money while I develop ideas (and my trading system), I would like to apply some of the methods presented above with a nice little script.

    Now: I know that Funyoo does not want to reccomend anything here. And I can very well understand why. "Which is the best EA" is a meaningless question. Because besides distinguishing between badly sucking and probably profitably systems, it is not possible to answer with a bit intelligent detail. Every one has different features that must adapt to your own trading style, psychology, capital availability, time availability and so on.

    I want therefore to ask something much more specific, also to give an example of the features that a ranking of EA's, according to my first post, should list, if possible.

    I would like that somebody, kindly, indicates me an EA:
    • That is available here (or elsewhere, but strongly prefer here) for free
    • That is present as a source code. I will never ever trade any "black box" around. I want to perfectly understand, at last, how my system works. I want to know how I am employing my money on the market.
    • Therefore, if possible, a quite straightforward and well explained one
    • Thus, again if possible, one that is not based on "too exotic" indicators/patterns/methods. I'm in this business since more than two years, now, and read quite a lot. So: any "classical" indicator is OK, comprising something a bit advanced, like Ichimoku, Chaos and the like. If it is something more sophisticated, please I would like to have some reference to the source of relevant information. Basically, all what is included in Kaufman's handbooks can be good.
    • I want low risk stuff, so: no Martingale, no stop-loss-less things. I do not want the risk of being pulled out of the market due to a single big shot, even if the probability that it happens is 1 in 10 years (you know: for the Murphy Law this will be for sure your second week of trading LOL )
    • I am perfectly aware that in finance low risk = low profit. That is like the gravity law of physics. My balance would be: well, let's try to beat a hedge fund...
    • Possibly I would like a moderate pace system: so, ideally one or two trades per day, regularly, every day. Not much more, not much less.
    • Ideally I would like trading the H1 timeframe. On the forex, due to the 24/7 trading hours, longer bars are different in any broker according to the time zone they have their server located in. This makes me a bit unsure about a method where my indicator has a value if I apply it on a chart of data calculated in London, and another value on data calculated in Tokio... But, on the other hand, I am definitely resistent in trading bars shorter than 15M
    • Ideally (unless it is a ranging-market suited system) I would not like a "scalper", i.e. very few pips/trade on average. Till now I was mostly a trade follower. I would not dislike other examples, like channels or similar. Adaptive systems, i.e. systems that change "system" when they recognize trending/ranging conditions, are actually what I'm aiming at. But I know that the road thereto can be VERY long and difficult. Anyway: an average of let's say, 5 pips/trade is really far from my wishes. You risk to give back in spreads all what you earn, and to be continuously in and out. "Big men" say that usually a lower frequency is better.
    • It doesn't matter that the system has produced humungous results in all backtestings so far. I will optimize it, and likely with at least some of the procedures indicated in message two. I am convinced that such process can increase system profitability of 10-100 folds.

    So: you see. Asking for "the best" EA can be a quite elaborate process, and I am sure the criteria of other people would be much different from mine. Therefore I think that a good ranking or listing of EA's should contain more values than not simply an "average ratio".

    From where I started to look at (and of course with 200 EA's out there the job is quite complicate!), I was now leaning towards the very simple three moving averages system. I considered the Taichi Demarker (just because it is the first of the list), but I discarded it for its "clustered" behavior and for an apparent lack in robustness. I considered the RSI Power Maximum, especially because I know pretty well all the indicators, but then immediately discarded when I saw it had no stop loss. I wanted to give a look to the ten wawes stochastic, but couldn't follow all the discussion about the different versions (if they are really different trading systems maybe it is better "separating" them in the evaluation list).
    The three MA's is really easy to understand, although I would prefer something else because I have already quite an experience in coding and testing trend-following systems based, among others, also on MA's. I was not completely satisfied with my results (that now, in retrospect, do not look so bad, actually), and so I would like something better promising, if anybody can give me a suggestion.

    My goal is starting making some money (not too much, probably) while I go on with my research. Will be glad and thankful if somebody helps!


    Last edited by fsprea; 08-27-2009 at 13:10. Reason: Added one criterion

  4. #4
    Administrator funyoo's Avatar
    Join Date
    Sep 2008


    Hi fsprea,

    Impressive first posts and interesting suggestions!

    I agree with you about the top, the ratio is not the best solution.

    However I had to choose for a formula that could be used easily and quickly.

    Your proposal for a better ranking would be indeed the ideal but as long as we can not automate the processus, the time required to achieve it would be monstrous.

    Note that MT5 will propose some new features to judge a system, if there is some potential or if it only sucks.

  5. #5


    People ask all the time which EA is the best $$? This forum doesn't answer that.
    This is #1 criteria to implement, then everything else will follow.
    Make a voting poll from the user end perspective.
    Talk about coder perspective later, now everything is upside down in this forum.
    Otherwise it can become an analisys paralisys and it is already.
    Google works, because they understand practical relevance of information. Same thing applies here.
    Establish prime purpose, prime objective first, then link everything else to that

  6. #6
    Junior Member
    Join Date
    Feb 2009


    Hi Funyoo and Metatrader,
    sorry for the delay in replying, but I was around in other forums in the meanwhile.
    Yes: making all manually is really a mess, and I am glad to know that probably MT5 will solve some of our problems.

    I still did not decide, yet, which system to start from, also because I am looking at some other interesting ideas. It appears, anyway, that fulfilling the "criteria" I exposed above is quite cumbersome, if nobody is giving a hint, because in this case I must make the "handwork" of looking at the different one by one, and it takes a bit of time...



  7. #7


    It's really very nice post.It's very impressive and interesting post.I really like this.thanks for sharing with us.keep up posting.

    Managed Services

  8. #8


    >>what's the best EA

    Every strategy has its trading session in the sun, and also its downfall. This is only more the case with automated trading, as programs don't adapt as fast as people do (well, some people). With this in mind, rather that trying to find the "best EA", it is more beneficial to identify which strategies work in what markets. Let the market show you the way, as Bill Williams might say.

    On this site, there is an decent trend-trading EA called vForce Like EA. I have just uploaded a refactoring of another poster's refactoring of Funyoo's original code. Since so many hands have been in it, I decided to rename it FunYooSalad v1.0:

    vForce Like EA

    It buys, it sells, it does pretty good when the market is trending.



Similar Threads

  1. Scalping Ergodic Ea proposal
    By herbstability in forum Ideas for expert advisors
    Replies: 0
    Last Post: 03-28-2013, 16:15

Tags for this Thread

100, automated trading, average, backtesting, breakout, broker, change, coding, color, comparison, design, expert, expert advisors, experts, forex, free, funyoo, grid, hedge, historical, how to, ichimoku, index, indicator, indicators, information, live, low, main, managed, manual, martingale, metatrader, money, mt5, pairs, pattern, profit, profit factor, profitable, research, rsi, scalper, scripts, sells, server, simple, source, stochastic, stop, strategies, strategy, study, system, systematic, systems, time, timer, tool, trader, trading, trading system, trend

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts

Appreciation and proposal(s)