Mean reversion – How to…?

Forums › ProRealTime English forum › General trading discussions › Mean reversion – How to…?

Viewing 11 posts - 1 through 11 (of 11 total)
  • 05/04/2018 at 8:23 AM #69766
    jebus89

    Hi all, im trying to make some mean reversion strategies, but im not doing great hehe. Having a hard time, and was wondering if anyone was running mean reversion strategies live here?

    Mind coming up with a few tips and tricks? Not asking for a profitable code, just what to look for and how to look for it!

     

    Any good books/blogs to read? Podcast to listen to? Any GREAT indicators i should look out for? Any custom indicators from this forum that works great? 🙂

    2 users thanked author for this post.
    avatar bezieh, avatar GraHal
    05/04/2018 at 1:58 PM #69790
    bezieh

    Bump, hopefully you will get your answers jebus, you deserve it fam! ^^

    05/04/2018 at 8:19 PM #69797
    Despair

    Most of my strategies are mean reverting. Doesn’t have to be much more fancy than for example some bollinger bands.

    1 user thanked author for this post.
    avatar jebus89
    05/05/2018 at 6:33 PM #69866
    Nicolas

    You can study the codes of bounded oscillator like RSI or Stochastic. Mean reversing is like waiting for a boomerang to get back..sometimes it doesn’t 😆

    4 users thanked author for this post.
    avatar jebus89, avatar Inertia, avatar Despair, avatar GraHal
    05/06/2018 at 4:42 PM #69893
    Despair

    Haha Nicolas analogy describes the nature of mean reversion trading perfectly. 😀

    05/06/2018 at 5:20 PM #69894
    Nicolas

    You can also talk about a dog at the end of a leash while you walk. The master of the dog is walking (he’s the moving average or the center of the price = the mean), the dog is the price (it will never go very far because retained by the leash, sometimes it will return to see its master, etc ..).

    1 user thanked author for this post.
    avatar GraHal
    05/07/2018 at 7:54 AM #69910
    jebus89

    @Despair if u dont mind me asking bout the majority of ur strategies:

    1. Whats ur favorite kind of markets?
    2. Whats ur favorite timeframe?
    3. Long/short only, or both?

     

     

    05/07/2018 at 2:08 PM #69937
    AlgoAlex

    Try to keep it simple.

    First, you have to identify a market who is more mean reverting than trending.

    Most of indices work good with mean reverting strategies, but only in particular timeframes.

    Then, use a method that statistically works on it to identify potential turnaround zone.

    I use a moving averages couple. I found certain combinations which work better according to the instrument and tf.

    Here is an example:

    05/15/2018 at 7:43 PM #70522
    Despair

    @jebus89: Sorry for late reply. I was on vacation and tried to stay away a little from the computer. 😉

    1.) Mean reversion is usually stronger in stocks than in commodities. So stock indices would be a natural choice.

    2.) Daily works best for me. On intraday TFs there is so much noise.

    3.) Depends on the market. Some I run long only, others both long and short.

    05/15/2018 at 7:48 PM #70523
    jebus89

    Thx @Despair for answering!

    Is slippage a big problem when buying on next candle open and is daily?

    If u dont mind me asking, whats ur avg winrate like in the majority of the algos? 70%+ ?

    How many trades would u like to see on backtest to give u more confidence, in ur mean reversion algos?

    05/15/2018 at 8:04 PM #70525
    Despair

    I don’t know if you want to call it slippage but when you trade a market that has no extended session and closes a few hours everyday it is not uncommon that the market opens with a gap up or down. Well this can be in your favor or against it. It was never a problem for me.

    The bigger problem is that PRT makes a lot of mistakes when running on daily bars (see my other posts). This is a serious problem and therefore I will now transfer all my strategies to matlab and send my orders direkt via Its API to take PRT out of the equation. We will see which problems pop up there. 🙂

    Winrate varies of course from strategy to strategy but maybe 65-75%?

    Of course the more trades you have in your backtest the better. I’d be careful with strategies with less than maybe 250-300 trades. But this of course also depends largely on how much data you have available.  Anyway no matter what if you only have say 50 trades it is in my eyes pretty weak. With so few trades the impact of so called data mining bias is huge. You probably have fitted the strategy to noise or it is a “worthless” strategy that simply got lucky on the dataset you used to test it on. As usual you should use WFA.

     

  • Author
    Posts
Viewing 11 posts - 1 through 11 (of 11 total)

Find exclusive trading pro-tools on ProRealCode Marketplace

Similar topics:
ndq mean reversion 5 min strategy

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