How to keep in faith with a statistically profitable system?
Page 1 of 502 12 LastLast
Results 1 to 10 of 20

Thread: How to keep in faith with a statistically profitable system?

  1. #1
    Dear fellow traders, I have a query stuck and still can't figure it out by myself, maybe I did not learn enough.
    This query is made for any quants, please help me with this

    Here we go:
    Let's say we've test our platform for after, say it 100 transactions, and we have 90 winning transactions and 10 losing transactions (so we've simply 90% winning and 10% shedding in percentage terms)
    Then, we run the system at the account, and suddenlywe have like 30 consecutive losses. In accordance with our system it was 10 percent of failures, therefore we must theoretically stick with our platform and expecting that we shall at least have other 270 winners (when we're planning to make a 300 commerce in a year), am I correct? Please fix if anything wrong

    Nah, the inquiry is, how can we know the next trades will be 270 consecutive winning trades? It may be 100 consecutive losing trades then 900 winning trades in a row, and it'll still be 10 and a 90% winning.
    If so, how can we simply apply what we have analyzed systems, even it's a 99% winning platform, it also could be a 10 consecutive losing trades and 990 winning transactions, how can we maintain in religion with our platform which generates large numbers of losing trades in a row while it was statistically a profitable systems?

    Please help me with This Issue

  2. #2
    Quote Originally Posted by ;
    We don't. After dropping 30 trades in a row, we are still only 90 percent likely to win our trade. The numbers won't ever balance (i.e. experience 30 losses, but don't anticipate 270 winners to compensate for it), but the bigger our sample develops, the more likely it will begin to reflect our edge. Over time, those 30 losses will get a memory.

    That said, losing 30 in a row is beyond horribly unfortunate, and it is much easier for a 50% method to do it than the usual 90% method.
    Oh I don't understand, if it dropped 30 days in a row I wouldn't wager on the 31st one being a winner!

    There are just two possibilities -

    1. It really has a 90% chance of success and the rarest occasion ever has happened - i.e. something with a 10% chance of occurring (in this instance a loss) has just happened 30 times in a row.


    2. It does not really have a 90% likelihood of a trade that is successful and you believe it will due to curve matching, not analyzing it the market whatever or changing.

    I would say option 2 is millions of times more likely to be the true instance

    If a roulette wheel came up with the number 27, twenty five times in a row exactly what could you believe? The wheel or what a coincidence is rigged or broken in some manner? If a coin came up heads 200 days in a row could you believe it was a fair coin?

  3. #3
    Quote Originally Posted by ;
    You can replie much of live's plogy in demo, with practice. Invoke taxes? There's a world of room in paper, simulator and demo before studying with pennies, to hone the technical advantage of one.
    I do not mean plogy. And no thats not the exact same.

    I mean no matter what a broker tells you that the feeds from demo to reside don't wind up with the very same results.

    The charts may seem the same in retrospect however, the results will wind up different. I have tested demo against reside trades side by side, with results that were different.

    The demo trade magically works out and just a stop being hardly missed and the transaction functioning, although the live account always suffers more from a badly placed entry or simply does not have the exact same chance.

    Demo is a game without a traders or big boys playing and moving price. Nobody is stop hunting demo accounts.

    I'm throughly convinced that all brokers that offer free demonstration's rig them to create live trading appear easier than it is to sucker in as many new tiny accounts as possible.

    Look over any FX trading forum and you will discover skads of new traders that do amazing in demo and post something to the effect of, is it really this easy.

    Forward and backward testing at Demo to establish a W/L ratio is a COMPLETE waste of time. At best its good to work out the nuts and bolts of could be.

  4. #4
    Interesing thread, comparing coin moves with no memory to trading is not entirly fair, possibly in a 100% mechanical system but lose 30 transactions in a row and pyschology wise your screwed and can expect to make more and more losses or reduce your profits so brief they might aswell be declines because there tiny compared to a losing trade.

    A platform which might use 90% accuracy in some market conditions can quickly reverse to 90% losers in other market conditions, so as others have stated take 2-3 declines in a row then lower your size or back to demonstration and reevaluate your system to be certain it's still performing at the current problems.

    The likelihood on 30losers at a row in a 90% win rate can also be billions to 1 likely 10^30 = 1E 30

    Even at 50/50 to 1 the odds on 30 are 1,073,741,824

    Do not get me wrong I have had 30losers in a row, so it's cause I SUCKED and my system had a 99% failure rate, the poster using 34 losers in a row that the terms didn't package his system for a while, but he kept going until it returned to normal I suppose.

  5. #5
    Quote Originally Posted by ;
    Thanks for the table, what was the formula you used to calculate it?
    Formula is currently in attached XLS.

  6. #6
    Quote Originally Posted by ;
    What you may control occasionally depends partially on the results

    You can control your position dimensions, but the next few bars (The result ) can lead you to modify your position size so.

    So in a feeling, you have to care about both.


    yes, my place dimensions may change based on what happens next. But that has no bearing how I come back to make a desicion.

  7. #7
    Grkfx strikes the nail on the head.

    You need to get an edge that's somehow grounded in either a comprehension of the hows and whys that induce price movement, and/or the price patterns that are created as an outcome; or some other method based on strong abstracts, e.g. trend, S/R, momentum, volatility, significance, economic trends and motorists.

    Anything else, more than a big enough sample size, is just a workout hit or miss, and the transaction costs will eventually defeat you. Therefore I agree with. I haven't been able to invent a profitable method from (only ) data mining/optimization; I'm not saying that it can not be done, but I feel it's not a trivial task. So methods are based on this type of strategy, and the vast majority of these, given time, neglect. It's so simple to compose an EA that enters based on a combo of indys, and scales out with X pip TPs and/or Y pip trailing stops, not one of which bears any relation to how the heavyweight players are driving price.

    Many other EAs employ MM methods that do simply re-balance yield and risk, and/or redistribute wins and losses.

    And it's all too easy to be fooled by randomness: fluctuations can be such that a platform that returned an 80% win rate over a 500 trade evaluation, can easily yield a 40% win rate over the next 500 trades.

    Marv recently started an interesting thread about how randomly generated charts apparently exhibit exactly the exact same kind of trends, channels, double tops, pin bars, the respecting of trendlines and fibo amounts etc, and so on, just as real price charts do. Given that a generated chart can't be defeated by ANY system, this demones that to be profitable, a system must exploit inefficiences on price charts that are in some way qualitatively different from similar patterns on generated charts.

    Here's a horse racing analogy. The savvy punter knows that studying factors like the track state, weight carried, obstruction draw, and jockey's expertise can give him a real edge. That equates to what grkfx calls skill. The uninformed punter may (for example) run a historical evaluation on saddle color, and during a large data sample detect that the horse that wears a yellow saddle blanket wins more frequently on Thursdays, while one sporting a green saddle blanket wins more frequently on Tuesdays. The outcomes of the evaluation, whatever the error, are insignificant, simply because the underlying assumption has no real bearing on the outcome.

    Much the Exact Same applies to FX. Either the factors that are being tested reflect the market drivers, or they do not. When they don't, then - no matter how big or representative the sample size is - the trader is effectively creating random entries and exits, and the spread will eventually defeat him, just like the zeros on the roulette wheel ensure that the house always defeats the punter if the latter plays for long enough.

    And on top of all this, the trader still has to receive his plogy right.

  8. #8
    Quote Originally Posted by ;
    Formula is in attached XLS.
    Thanks very much

  9. #9
    Quote Originally Posted by ;
    Interesing thread, comparing coin moves with no memory to trading isn't entirly honest, maybe in a 100% mechanical system but shed 30 transactions in a row and pyschology wise your screwed and can expect to make an increasing number of losses or reduce your profits so short they may aswell be losses as there tiny in comparison to a losing trade.

    A system which may work with 90% accuracy in some market conditions can quickly flip to 90% losers in additional market conditions, so as others have stated take 2-3 losses in a row subsequently reduce your size or back to demonstration and reevaluate...
    You managed to find a system with a 99% failure rate? Awesome, please share it. Be great to take the other side of those transactions

  10. #10
    Quote Originally Posted by ;
    Be good to take the other side of these transactions
    Perhaps; not (see myth #1 ).

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
This website uses cookies
We use cookies to store session information to facilitate remembering your login information, to allow you to save website preferences, to personalise content and ads, to provide social media features and to analyse our traffic. We also share information about your use of our site with our social media, advertising and analytics partners more information