Filters – What works for you?

Forums › ProRealTime English forum › General trading discussions › Filters – What works for you?

Viewing 15 posts - 1 through 15 (of 21 total)
1 2
  • 01/18/2018 at 8:43 PM #59835
    jebus89

    Hey guys, so im not sure if im in some sort of phase of learning, but my next big problem in creating strategies, is filtering out the bad parts of the market, for my system!

    So i’ve got a couple trade following strategies, and their main problem, and im guessing all of your trend following systems have the same problem, is that when markets go into sideways choppy movement the systems get fake signals and that triggers bad trades.

     

    What kind of market filters do you guys find useful? If you dont want to share filter-codes thats fine, but some insight is much appreciated.

     

    When i say filter i mean moving average 200 > moving average 500.. or ADX > 30 or whatever.. Anything inside your code that makes your system take less fake signals!

    01/18/2018 at 10:28 PM #59843
    Vonasi

    The problem with averages is the lag. Too fast an average and the lag is less but volatility of the average becomes too great to be useful. Too slow an average and you could be in a sideways market for a very long time before the average catches up and tells you that you are in a sideways market.

    Try moving away from the standard averages and use some of the alternatives.

    https://www.prorealcode.com/documentation/average/

    I’ve quite often found better performance with Time Series averages or if using two averages to confirm a trend then two different types of average sometimes works better. Create a pair of variables and use them in the AVERAGE instruction and let the optimization tool let you know what works best. For example:

    Every market and every time frame will be different and this way you are ‘tuning’ your averages to match them. Yes it is curve fitting but sometimes you just need an average or two to add as a filter against too many bad trades in a sideways market. You will never eliminate them completely with a lagging indicator unless you have a crystal ball and can tell the future.

    01/19/2018 at 1:41 AM #59851
    Nobody

    Ive found the best signals come in rising volatility at at least above a set point measurement of volatility   . Low volatility signals are more likely to be whipped . The best way to find out is measure ( quantify ) volatility vs signal efficacy  . I have lifted suboptimal Winrates below 50% to above 65% using these concepts  .  The basic premise is chop is noise and therefore not significant and more likely to be random , filtering that is a huge plus  … Sorry if i have offended anyone once again  😉

     

    Edit and as usual  i agree with vonasi ..   in the main averages are pretty average filters but also not totally useless . distance from average ( mean ) is a pretty handy metric imo

    01/19/2018 at 9:30 AM #59858
    jebus89

    Thank you both for ur contributions.


    @Brisvegas     Would you care to elaborate some more?  I agree, filtering out the choppy noice would increase my winrates significantly. (duh)

    I also agree that volatility would be a great way to filter out a lot of bullshit. When you say diff from avg/mean, do you mean stuff like bollinger%  indicator?

     

    Any tips are good tips 🙂

     


    @Vonasi    : The moving averages are indeed laggy, but unless youre trying to get super-crispy trades, i dont really mind a tiny bit of lag. Obviously the less is better, so ill look into using multiple different moving averages together. Im a big fan of wilder averages atm, the smoothing helps alot i feel. The problem with moving averages as filters are that (at least for me) your remove alot of the good trades as well as many bad ones.

    01/19/2018 at 9:36 AM #59859
    verdi55

    TRIX can be a good filter, sometimes better than common moving averages, because it is triply smoothed and does not change direction so frequently. Trade only when it has the right colour.

    RSI often works well : don’t buy when it is too high or v/v. It is important to optimize the period length, too.

    ADX in my systems usually gives insular results : small islands of improved results within a large parameter sea that does not work well. So I don’t use it.

    01/19/2018 at 9:56 AM #59866
    jebus89

    @Verdi55 , this is great, thanks. I havnt looked into TRIX as much, will definitly check it out 🙂

    Ive played with RSI but i think that the RSI is a better tool to use whent rading inside the choppy, rather then identifiying that the market is choppy. Maybe im just using it wrong.

     

    ADX is a filter i see many swear to, but i dont feel like i have gotten it right.. When im trying to use it as a filter i seem to be getting superlate into my trades and it seems more profitable to not use it and take some shitty failed trades, rather then using it and missing out big chunks of the trade. Again, maybe im just not using it right 🙂

    Ofc i optimize values 🙂

    Thanks for the tips guys, keep em coming! 🙂

    01/19/2018 at 10:19 AM #59871
    BC

    https://www.prorealcode.com/prorealtime-indicators/andrew-abraham-trend-trader/

    Take a look on this one.

    01/19/2018 at 10:37 AM #59874
    jebus89

    Hi Bin!

    Thanks for your reply and thanks for link to an interesting indicator! Unfortunatly im not sure if that would serve as a good filter, but rather maybe a good entry condition or maybe even a good exit ! 🙂

    But as a filter i think it seems to be swithcing over/under price rapidly when in chippy markets. Making it as good/bad as, say a supertrend..

    maybe im reading it wrong? please share the “how to” if youre using this as a filter today!

    01/19/2018 at 11:38 AM #59877
    BC

    Hi Jebus

    It work quite good for my algo, of course it is a must to optmize the main parameter to pitch in different kind of market and time frame (I prefer 1 hr or up). Example as below:-

    UpTrend = Close > Ret
    DownTrend = Close < Ret Then work together with your favourite momnetum or stochastic indicator to trigger buy or sell signal. Enjoy!

    01/19/2018 at 12:55 PM #59882
    jebus89

    Again, thanks for the quick reply Bin, and thanks for sharing. I’ll definitly give it a check with some of my existing algos and see what the results say! Thanks

    01/19/2018 at 6:14 PM #59913
    Nobody
    jebus89 wrote:

    Thank you both for ur contributions. @brisvegas Would you care to elaborate some more? I agree, filtering out the choppy noice would increase my winrates significantly. (duh) I also agree that volatility would be a great way to filter out a lot of bullshit. When you say diff from avg/mean, do you mean stuff like bollinger% indicator? Any tips are good tips 🙂 @vonasi: The moving averages are indeed laggy, but unless youre trying to get super-crispy trades, i dont really mind a tiny bit of lag. Obviously the less is better, so ill look into using multiple different moving averages together. Im a big fan of wilder averages atm, the smoothing helps alot i feel. The problem with moving averages as filters are that (at least for me) your remove alot of the good trades as well as many bad ones.

    Easiest way to measure distance from mean for me is to subtract a moving average ( mean ) from price  … IE   ….   Distance = CLOSE – MovingAverage (period)  . You can use more sophisticated mean distance measurement techniques to get some momentum / trend filter characteristics as well  .   As long as you dont dilute the original objective of a filter its handy attribute to make filters dual purpose for algo strategies . Measuring 2 or more key metrics  .

    I do use ATR for my chop filter   . I apply a moving average to ATR and either rising ATR or ATR above the moving average are the filters to turn on/off  algo signals  , I am sure there are other ways to utilize these concepts  , be aware i use this filter on intraday strategies on instruments that trade 24 hours , not a concept that will work on all timeframes   or instruments

     

    01/19/2018 at 9:18 PM #59926
    jebus89

    Awsome thats some good stuff. Ive been playing around with moving averages on different indicators such as the atr, havnt found a good way to work it just yet tho. And yes i agree that its very different depending on the timeframe and the time in trade. Im also working mostly on intraday systems! Thanks for the tip.

    01/19/2018 at 9:31 PM #59927
    verdi55

    I use ATR or historic volatility not as filters to open a trade, but as a decision when not to close a trade at breakeven.

    When volatility is high, I will not close an open position at breakeven, although other criteria may be met, but I wait until volatility decreases again. The reason is that volatility is often high around the opening of a position, but decreases later, after the position has run into positive terrain. This way, premature closing of a position can sometimes be avoided.

    01/19/2018 at 9:36 PM #59928
    jebus89

    oh that is very interesting verdi55, i havnt thought about that. Ive been thinking alot lately about my stop loss. I feel like its not dynamic enough at all, its basicly super simple.

    01/20/2018 at 12:47 PM #59948
    Yannick

    Hello

    If we go back to basics, there a 6 markets types (according to the author Sweeney or Van Tharp, i don’t remember from where I copied this picture).

    For trending market, you might have to consider moving average and momentum.

    For volatile market, you have to consider volatility, true range..

     

    Unfortunately, the author didn’t described what filter works the best in which market type .

    Regards

    1 user thanked author for this post.
    avatar Nicolas
  • Author
    Posts
Viewing 15 posts - 1 through 15 (of 21 total)
1 2

Find exclusive trading pro-tools on ProRealCode Marketplace

Similar topics:
you don’t know what you don’t know!

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