Reversion to mean systems big project

    Posted by matt on 25th of Jun 2021 at 11:11 am

    I started working on a big project, not sure what it's going to look like in the end but started testing. We have the SPY reversion to mean systems as you know. So I'm taking those same systems and running them on 15 - 20 of the sector ETF's so in the end have a basket of reversion to mean systems on the market and sectors. I'm not sure how much overlap there will be, obviously there will be but maybe it won't be bad and everything will be pretty synergistic as a group. 

    This is a big project. I basically have to run back tests on say 15 ETF's with 22 sub systems each, that's a lot of testing. YES I run back tests for reversion to mean. I know that back testing is a bad word in the systems community, and used in the wrong way it is. That said, in my opinion reversion to the mean systems are all about finding rare and important statistical events of the past data, you can't get around that. So to me it's necessary, you have to run tests on the past 20 and 30 years to see how the indiviidual statistical variations of those ETF's, you can't just use one setting for everything it depends on the instrument (well you kind of can but then you deal with low profit factors and what's the point. Anyway  with say 15 - 20 ETF's we are looking at 400 backtests, and generally I will run refined tests on those many times - so this could be a 2 or 3 month project if everything is looking good and I decide to do the full tests. On Friday's like this and at night and weekend's I'll be running tests.

    That said, let's say in the end I'm able to get this to work. Logistically I will probably have to chance some of the alert system to be able to handle it. For the SPY systems right now I send alerts for sub systems spread across multiple charts. For these ETF's I won't be doing that, I'll have all sub systems on one chart, otherwise it's too much to follow. 

    Also currently I send out option alerts for the SPY, well for the ETF's, especially if multiple trigger in one day will be difficult for me to do that. So I'll probably send the alert with what my parameters are for the options - for 1st entries I target expiration dates 2 months out with a 0.6 - 0.7 delta, so  you can just use that info to pick the options, or just do the ETF's. 

    Lastly - the email system can be delayed - I think the best way to handle alerts in real time would be a BPT systems alert APP that you would install from the AAPL  app store or the Google Play store - then you simply get alerted through your phone on the app.

    anyway that's some things I'm working on, if this all works out it would be probably a couple months - and it's possible that after I get into this I find that the stats are not worth it, so it's very preliminary, just letting you know what I'm working on the side.

    Use search function. Options, Matt

    Posted by retirefire on 29th of Jun 2021 at 05:12 am

    Use search function. Options, Matt

    regarding the reversion to mean

    Posted by matt on 27th of Jun 2021 at 03:06 pm

    regarding the reversion to mean systems I said last week that I was looking to add these systems to a collective group of ETF's instead of just SPY. 

    As I stated this is a big project to do, adding all these sub systems to each of these ETF's will take a long time to test. Over the weekend I've tested the Bear Long sub system to most of these, but you can see how many more systems I have left, again it's going to take a month or two in order to complete this; for example it's taking since Friday through the weekend just to do the Bear Long tests with my computer running all day and night, big project. Then of course I have to see how much overlap there is, for example on a pullback in an uptrend I really don't want 10 or 15 of the ETF's to trigger long at once, ideally a few would at most, but we'll have to see how the data looks as I get more of it. 

    Anyway this is the list of ETF's I'm looking out. I also threw in EEM and IWM index ETF's. I may consider GLD and FXE and maybe TLT as well as those are pretty uncorrelated

    I would like to see

    Posted by skyfish on 27th of Jun 2021 at 11:53 pm

    I would like to see GLD in your R2M system.

    Thanks for the innovation and

    Posted by mstaples37 on 27th of Jun 2021 at 07:11 pm

    Thanks for the innovation and hard work, Matt.  Those should be very useful.   I know there are scope creep concerns, and this usually has high correlation with the S&P 500 (I believe), but DM ex-U.S. (e.g., VEA or SPDW) /or its two main constituents, Europe (e.g., FEZ) and Japan (e.g., EWJ) would be interesting, at least to me.   That, along with EEM, would cover most of the world.  I remember that rundown you did of all the country / regional ETFs and it was interesting.

    Here's an updated table, this

    Posted by matt on 28th of Jun 2021 at 09:25 am

    Here's an updated table, this is the ETF's I'm testing, at the bottom you can see 'other' which are non correlated to the general market

    been working on these, mostly

    Posted by matt on 28th of Jun 2021 at 03:19 pm

    been working on these, mostly done with one sub system LOL.  So much more to go

    anyway the bottom ETF's are pretty non correlated to the index and sector ETF's. 

    looks like there would have been a FXE Euro reversion to mean trade last week. 

    trying to get the bear long finished on GLD, tough to get good numbers on that ETF

    Thanks, I like some of

    Posted by matt on 27th of Jun 2021 at 07:14 pm

    Thanks, I like some of those like FEZ makes sense, SPDW and VEA interesting. Though I"m trying to stick with ETF's with data back as far as possible

    This will be awesome Matt,

    Posted by morton7 on 26th of Jun 2021 at 03:36 pm

    This will be awesome Matt, thanks!

    This would be incredibly valuable

    Posted by timebandit on 25th of Jun 2021 at 04:09 pm

    This would be incredibly valuable (assuming the backtesting validates it), and a significant enhancement to the BPT subscription. I would think you would want  some compensation for all of this work. Maybe you can charge a one-time price for the app to help offset your time investment. The app would of course still require a valid BPT subscription to operate.

    That will be fascinating, especially

    Posted by kevindeng0727 on 25th of Jun 2021 at 02:06 pm

    That will be fascinating, especially if the winning %  is close to the current SPY system.

    Sounds like much work but

    Posted by steverobin on 25th of Jun 2021 at 01:15 pm

    Sounds like much work but could have huge benefits for the Community.  

    Thanks Matt !

    Posted by iaranger on 25th of Jun 2021 at 12:04 pm

    Thanks Matt !



    Awesome concept, your always improving

    Posted by retirefire on 25th of Jun 2021 at 11:54 am

    Awesome concept, your always improving BPT. Best value out there by far. Have you considered a fund that trades on systems and members buy shares? Not sure the legality and headache on your end for that though ??

    I would put all my

    Posted by lapri24 on 25th of Jun 2021 at 12:15 pm

    I would put all my account money in that fund.

    i like the phone alerts

    Posted by RP on 25th of Jun 2021 at 11:36 am

    i like the phone alerts i have missed last five trades as i am working on a project at home or rentals or my church etc and some folks here still are working so that would bee great to get alerts on our phones. sounds like great plan! another thing i like about BP always trying to grow and move forward setting still is never good.

    RP

Newsletter

Subscribe to our email list for regular free market updates
as well as a chance to get coupons!