Bad beginings

Forums › ProRealTime English forum › ProOrder support › Bad beginings

Viewing 15 posts - 16 through 30 (of 33 total)
1 2 3
  • 03/22/2017 at 9:26 PM #29514
    TempusFugit

    Nicolas, thanks, very interesting sharings. But your answers bring up to me more and more questions, for instance:

    1. So which are the trendys schools in trading?
    2. Can money management by itself be a wining strategy?
    3. I understand that as less correlated strategies you use the better. But the more strategies you use is also bigger the risk of any/some of them are a bad apple. I mean, if you have confidence in a strategy, using it in different markets and different timeframes maybe do the trick. For instance 1 minute and 1 day TF should be using already quite different entrys/exits even in the same market/strategy, isn´t it?
    03/23/2017 at 8:36 AM #29545
    Nicolas
    1. All strategies that believe in trend following, whatever the way how do you define and trade it. You’ll find many materials on internet about it.
    2. Yes, coin flip can win with a decent money management strategy 🙂 (joke)
    3. Yes they can perform differently, but the correlation between them will be given by their results/equity curves/behaviour over time (same period).

    These are only mathematical and statistical theories, but can definitely help a trader to find the right balance to be more confident in his portfolio (less risks and probable same behaviour).

    1 user thanked author for this post.
    avatar TempusFugit
    03/23/2017 at 10:14 PM #29709
    AutoStrategist

    But what I want to to talk about here is that I noticed in backtesting is that this system and others, even with a long wining system, is quite common  to start the firsts months/weeks/days loosing money. Does it happen to you also? Why you think is that?

    Hi TF, I think there is some universal market law at work.  Every time I see that one or more of my strategies has had a good winning run and I decide to increase the position size and boom it starts losing.  It seems I need to do the opposite, increase the position size when a strategy has been in drawdown for a while but that feels like a horse race punter saying, hey that horse lost it’s last five races it must be due a win, I don’t think I could do it.  The opposite situation also occurred recently where a trend following strategy that had performed well for a few months went into drawdown so I thought ah market conditions have changed time to switch it off,  only to see it go on a winning run and recoup all the drawdown before I finally decided to switch it back on 🙁

    03/23/2017 at 11:11 PM #29712
    GraHal

    Sod’s law, but we all know it’s mean reversion? 🙂

    We lose out bottle cos we’ve bet too high a stake and so we close the deal … as you say AutoS, only to see it turn round in our favour had we left it running.

    I’m trying to discipline myself to small lots that I don’t bottle out on even if it goes the wrong way for a few days (after I miss a good exit due to distractions).

    Demo accounts have got a lot to answer for with unlimited virtual money … we get used to excitement of ‘big lot trading’ and it’s hard to come down! But that’s why brokers do it so we go Live and lose!

    On Demo manual at £100 per point, 5 point gain in a few minutes over breakfast left me feeling great, now Live at £2 – £6 per point I have to work a lot for not a lot! 🙂

    GraHal

     

    03/24/2017 at 10:53 AM #29745
    TempusFugit

    I see your point AutoS. Theoretically the answer is clear… constancy and be willing to endure drawdowns, but I almost have not experience with autosystems in real  yet so I won´t presume to know it right. What worries me at this point is… you can trust for sure the backtest, you can trust for sure a few live results (positive or negative)so… what do you trust? Seriously I am asking guys, what things do you trust in using a system in live? Because right now it seems to me like a jump on faith.

    Another thing that come across my mind. I am preparing my own autosystem, nothing to show off as I am learning here, but let´s fantasize that I get a marginal wining strategy. Then… It´s a good idea to share it here or wherever? I like the idea of sharing these things but can share it make the strategy lower its chances to win?

    03/24/2017 at 12:13 PM #29751
    GraHal

    TF sharing on here will increase the chances of success if other members point out possible problems (or even errors) with your strategy. Although often we are too ‘gentlemanly’ to do that lest we offend? But it can be done in a kind, constructive and helpful way.

    The only System I trust is one I have coded myself or by another if I fully / 100% understand how the code is working.

    Trust means how it meet Buy / Sell conditions … not make profit continously.

    If you Backtest & Walk Forward & Monte Carlo & Demo Forward Test for months (depending on Timeframe) and all results are good (good needs to be defined) then you may stand a chance of making money??

    I don’t think I’ve got enough time left in my life 🙂 for above on lots of prospective Systems … but that is the raison d’etre of this site (big thanks to Nicolas) … we all help each other develop each other’s talents towards winning Systems (I know you know this, but the words are flowing!)

    You may like to watch the Thread below developing … seems the right way to go to me, but requires dedication and skills in several areas. I hope Maz does not get disillusioned?

    https://www.prorealcode.com/topic/trading-system-boilerplate-collaboration/

    GraHal

     

    1 user thanked author for this post.
    avatar TempusFugit
    03/24/2017 at 8:12 PM #29782
    GraHal

    Here’s an idea we could try … if we can code it … should be straightforward, even I might be able to do it! 😉

    I can live with Drawdown as long as it’s not a ‘bad beginning’ so how about having a snippet in a Bot which exits a Trade if equity goes negative < – £50 or – £100 and then does not trade again until the average number of bars per Trade has elapsed for that particular Bot?

    Logic is that after average bars, conditions may be more favourable so the Bot makes a gain / good beginning and any future Drawdown will be out of Profit for that Bot?  If next trade makes equity negative < – £50 then repeat.

    I have used If equity negative < – £100 then Quit, but it required me to manually start the Bot again and I didn’t analyse the results or compare with the same Bot without ‘negative equity < – £100 then Quit’.

    With above at least we’ll know there’s a limit on ‘starting equity loss’ … bad beginnings may not be quite so bad? 🙂

    GraHal

    03/24/2017 at 8:35 PM #29783
    TempusFugit

    GraHal, maybe I am too square head but I don´t see the point in that. For me it would be the same that saying… “ok, as there is too much bad beginnings let start always later that we were going to”. No matter when you start, it´s a matter that there is always a begining, or a breakeven point. Sorry if I am to harsh, I just want to make my point clear. Is different if you think that in your strategy the loser trades use to come in groups, one after the other, in that case it can be a good idea. Am I missing something on it?

    By the way, I saw in your Pathfinder file that you use it in different markets than DAX. With the same paramerts or optimized for each one?

    03/24/2017 at 9:31 PM #29785
    GraHal

    No problem, it’s good to discuss …

    If a Bot is launched and takes a trade at, for example … a daily top, a 61.8% Fib retracement or the end of Elliot Wave 3 … then a reversal would be expected.

    If the Bot entry conditions were for a trade going the wrong way (from reversal above) then a loss could build over many bars and so we have a bad beginning which freaks me out cos I am thinking when is this going to turn?

    If I had several Bots running I could end up ‘account wiped out’ and margin calls etc. Drawdown from ‘Bot Profit’ would not result in wipe out etc.

    In my opinion, many Bots have lots of clever excellent logic / strategy, but do not take any account of prevailing market structure at trade entry.

    After elapse of average bars for that Bot then prevailing market structure may be more favourable to a trade entry.

    Re … same parameters or optimized for each one … I almost always optimise for each market unless I can’t understand the code at all! Even then I usually try to optimise a few – what look like variables – out of curiosity :).

    Cheers
    GraHal

    03/24/2017 at 10:28 PM #29786
    Derek

    From my experience the “flat” beginning can have several reasons:

    1) a system is only really profitable every other period from the start. For example every third year is a flat year.

    2) the initial observations the system has been build around occur more frequently in recent markets. Differently put: Why look at data from 2001 to design a system for 2017?

    3) during the design process we focus not on every single point of the equity line but on the most interesting periods with the highest gains and largest drawdowns. By the time we are finished and finally take a step back the very beginning enters our view “for the first time”

    4) lastly, and maybe most important for many strategies, auto-optimization of parameter values will force the equity curve around the time period where profits are biggest. Many strstegies in the library have only one or two good periods and are mostly flat.

    Personally I conclude for myself that there must not be a law that systems start of on the wrong foot. It’s merely a matter of pereception and the system development process.

    1 user thanked author for this post.
    avatar GraHal
    03/24/2017 at 10:38 PM #29787
    GraHal

    Derek, I agree with all you say above.

    My proposal (re Quit < – £50 above) is just a simple idea to limit reduction in starting equity. Your point 1. above would seem to support the rationale.

    I guess you’ve seen the Thread below … sounds like you could contribute to the collaboration.

    https://www.prorealcode.com/topic/trading-system-boilerplate-collaboration/

    Thank You
    GraHal

    03/25/2017 at 11:51 AM #29802
    GraHal

    Just wanted to explain further, when I said above
    If a Bot is launched and takes a trade at, for example … a daily top, a 61.8% Fib retracement or the end of Elliot Wave 3 … then a reversal would be expected
    You can see that price will have been heading in the same direction (e.g UP) for some time (longtime if on 4H or Daily TF) and a Bot based on ‘price derived indicators’ will likely meet conditions for a Trade in the same direction – UP.

    Bots I see do not take account of Elliott Waves, Fib Retrace, existing support & resistance levels which can set a price going in the reverse direction for seemingly no reason at all … apart from a self-fulfilling prophesy because the herd believes that is what will happen?

    Support and resistance levels are a more logical reverse point, but if a Bot does not take account then they are not there / invisible as far as the Bot strategy / intelligence is concerned.

    If we set a Bot Live and it encounters the scenario above on the first Trade then we have a bad beginning (due to trade up, but market reversal down) and so to limit starting equity loss we need a stop loss (e.g. Quit < – £50). Normal Drawdown from Bot profit would not result in equity below zero and so Stop Loss Quit < – £50 would not be applied except to bad beginnings.

    GraHal

     

    03/25/2017 at 1:16 PM #29810
    TempusFugit

    Grahal, you make really interesting arguments and made me think a lot, but sorry, still don´t buy it. My two cents are with Derek, is a matter of perception or system development or whatever, anyway can happen now or 100 hundred bars later if you postpone it. The boot can start loosing money with or without Elliot waves, Fib retraces… and those things can happen also 100 bars later. But hey, I hope I am wrong, why don´t you try to prove it in a system?

    03/25/2017 at 1:50 PM #29814
    AutoStrategist

    Some feedback from my experience although I make no claims on being an  expert, in fact most days  reveal to me how little I really understand about how it all works.

    TF, be very careful when you start out.  You are most likely to be starting with a single strategy so if that goes into drawdown you have no others to mitigate it.  I first went live with a system that in backtesting over the previous year had made £75K in total and lost £55K and all I focussed on was the difference between the 2 numbers and nothing else.  It only stayed live for about 4 days and it’s amazing I didn’t go broke.  That strategy by the way now at version 10 is one of my best performing systems although it doesn’t run very often due to all the filters and conditions I built into it to make it more effective.  As Nicholas said, ultimately diversification ( ‘the only free lunch on Wall Street’)  is the way to go and something I am actively looking for:  multi market, market neutral (long and short) plus multi-strategy; momentum, trend following, mean reversion etc.

    Regarding initial drawdown, yes I think we all need to tough it out sometimes and let the systems run through good and bad.  I don’t recommend you do this but I always put my new systems live based on backtesting (always using the maximum available data) without running them first  in demo, certainly focuses the mind on either getting them right, quickly correcting any issues or in some cases retiring them altogether (there have been a few). Also I always look at a strategy after it has run especially if it loses, it is incredible what you miss or didn’t notice and the market is very efficient at pointing these things out to you!  Watching strategies as they run in real time can also give some great insights for improvements.

    Hope this is helpful.

     

    1 user thanked author for this post.
    avatar manel
    03/25/2017 at 2:16 PM #29815
    Eric

    If running TS on demo and then choose the best performer and put it on live

    then you probably have a high risk for drawdown in the beginning live

     

  • Author
    Posts
Viewing 15 posts - 16 through 30 (of 33 total)
1 2 3

Find exclusive trading pro-tools on ProRealCode Marketplace

Similar topics:
how do i resolve bad performance at supertrend back test?

Create your free account now and post your request to benefit from the help of the community
Register or Login