OmniTrader Professional Forum
OmniTrader Pro General Discussion
ATM Micro States Method
40 replies, 7382 viewings
|
OmniTrader Professional Forum
OmniTrader Pro General Discussion
ATM Micro States Method
|
Last Activity 2/23/2025 4:01 AM 40 replies, 7382 viewings |
|
|
| Printer friendly version |
mholstius Regular  Posts: 76 Joined: 6/15/2011  User Profile |
Nirvana originally included the excellent “Universal” method with ATM, and has since added methods that utilize updated Market State definitions and ARM5 strategies. I’m always working on ways to increase diversification and stability, so I thought I’d experiment with some new Market State definitions in ATM. I spent the past couple of weeks developing this new “Micro States” method. It adapts rather quickly to changes in the market and trades a list of 300 symbols using a cross section of 8 strategies. Like Nirvana’s Universal method, it has Bull, Bear, and Default states - but it changes the market state much more often. Nirvana’s Universal method recorded 79 changes of market state from 1/1/2003 to the present. This Micro States method had 269 changes of market state over the same time period. The snag below colors the Market State zones for both the Universal and Micro States methods since 2003 to illustrate the difference; The snag below magnifies the areas numbered 1-4 in the Micro States method above to illustrate when the market states change; Below are some statistics on the changes in Market States from 2003 forward; These are the Strategies used by the Micro States method (only XLS-19 V2 is used in more than one market state); Now, how does it perform in the Port Sim…??? This is a 15 year run from 1/1/2003 to present using 2X Margin and both Longs and Shorts; This is the same 15 year period from 1/1/2003 to present in an IRA using 1X and Longs Only; Here are the statistics for the two 15 year runs from 2003 to the present; The first half of this chart is the particularly challenging period in the market from 2014 to 2016; I’m posting this because a lot of us new owners of ATM are trying to learn how to make the most of this great new tool. This new Micro States method is just an example of what can be done - and I haven’t even utilized any of the Optimize capabilities available in ATM. ATM isn’t like anything we’ve had before (or that I’ve ever seen on the market). It allows us to test and implement ideas we used to only guess about, uncovering relationships we couldn’t quantify before. My hope is that if we share our discoveries, we can accelerate the learning curve. While a particular idea one person has may not be the Holy Grail, it could be the missing puzzle piece for someone else… That said, I believe this new Micro States method has potential - so that’s why I’m sharing it with everyone. It appears to have stability, consistency, and diversification over the past 15 years; 9,542 trades 8 different strategies trading 300 symbols 269 market state changes To make it easier for everyone who owns ATM, and in case its results and availability might have an influence on those of you still considering the purchase of ATM, I’m donating this Micro States method to the library that Nirvana is making available to all ATM owners. My hope is that everyone can save some development time, verify the method by duplicating my results, and let us know if you find ways to improve it. Maybe you’ll develop a better ranking system, find better strategies to use, or define a good dynamic list to pair with it. The possibilities are endless… If you don’t want to wait for the official release of Micro States from Nirvana, I’ve attached a pdf file that contains sequential screen shots showing all the details so that you can duplicate the Micro States ATM method today: “Details of Marks Micro States ATM Method.pdf” I’ve also attached the symbol list I used. It’s similar to the list I’ve posted previously, but with the recently extremely volatile SVXY removed: “Micro States Symbol List.csv” So, this is just free for the asking…??? Yup. There’s no $100/mo subscription - but I do have one small request. If you find that this Micro States method adds value to your investment in ATM, either by saving you development time, learning the concepts and ideas it uses, or improving your trading results… Please, donate whatever you feel that additional value is to the St. Jude Children’s Research Hospital; https://www.stjude.org/ They’re doing some amazing things for kids dealing with serious illnesses. Thank you! Mark [Edited by mholstius on 2/26/2018 10:48 AM]  Attached file : 00 Comapre Profiles.jpg (126KB - 2746 downloads) Attached file : 01 Various Periods.jpg (187KB - 2750 downloads) Attached file : 02 Statistics.jpg (102KB - 2732 downloads) Attached file : 03 Strategies Used.jpg (24KB - 2728 downloads) Attached file : 04 Margin.jpg (101KB - 2671 downloads) Attached file : 05 IRA.jpg (99KB - 2754 downloads) Attached file : 06 Port Sim Statistics.jpg (121KB - 2633 downloads) Attached file : xx 2014.jpg (77KB - 2654 downloads) Attached file : 07 St Jude.jpg (3KB - 2597 downloads) Attached file : Details of Marks Micro States ATM Method.pdf (1675KB - 623 downloads) Attached file : Micro States Symbol List.csv (1KB - 480 downloads) | |||
| gbarber Member Posts: 28 Joined: 2/26/2018 Location: PEARLAND, TX User Profile |
Thanks again for excellent work and also for your outstanding graphics that make things so readily clear. I am wondering about some things you didn't do as well as what you did. I hesitate to ask you because you are already doing so much for us. So, if you have some available time, here they are. 1. Why did you choose the path of definng moree granularity in the states we have been using (raging bull, bear, and bull (default)) instead of expanding the number of states defined. For example, Jim Dean has been using 9 states. These are from one of his posts "The "DeanMarket" algorithms presume that Market States are permutations of: Three Trend states: Bull, Bear, Flat ("flat" = short for "not Bull or Bear") ... and ... Three Volatility states: Wild, Calm, Norm ("norm" = not Wild or Calm) Thus there are Nine Market States defined by the permutations of those: BullCalm, BullNorm, BullWild, FlatCalm, FlatNorm, FlatWild, BearCalm, BearNorm, BearWild." That seems to pretty well cover all of the possibilities. The problem then is to identify a specific set of strategies that are particularly well suited to each of the 9 states. I don't have the skill set to come up with logical rationale to understand the characteristics of strategies that would make it unique to a particular market state when there are 9 of them. For just 3 states, it is easier to say for example that raging bull is a significant trend so trending strategies should work better. However, one could use the strategy optimization feature of ATM to let the computer figure that out. I have tried that a bit and found that it takes an enormous amount of compute time and crashed out once. I am interested in pursuing that path when I can find the time to do so. So just interested in knowing your thoughts regarding the path of investigation. 2. the second question is directly related to the first. I am wondering why you chose the strategies you did for each of the states. For example, you chose XLS-19 v2 for all states including raging bull even though that is not a trending strategy. | |||
| Jim Dean Sage    Posts: 3433 Joined: 3/13/2006 Location: L'ville, GA User Profile |
1. Relatively low volatility (ie MTV calm values) 2. A bullish trend that … 3. Is well established/ mature. (The latter two would show up as a high bullish MTV score, btw) [Edited by Jim Dean on 2/26/2018 2:00 PM] | |||
| mholstius Regular Posts: 76 Joined: 6/15/2011 User Profile |
To answer your questions… 1. Why did you choose the path of defining more granularity in the states we have been using It’s a question of choices and available time. I try to evaluate one thing at a time, and this looked like a good choice for investing the 50 or so hours it took to do. I’m glad it worked out - sometimes things end up being a dead end. Jim has invested a LOT of his time investigating market states. His work looks very promising, and I certainly hope he’ll develop something much better than mine. 2. the second question is directly related to the first. I am wondering why you chose the strategies you did for each of the states. Because of the statistics, I’ve been biased toward using RTMs for the past couple of years - that’s why all the strategies in this method are RTMs. Also, I chose to have this micro method close all trades when every state changes, so I don’t feel that would allow enough time for the trending trades to develop. They’d probably be closed out too soon in many cases. You may want to uncheck the “Close All Trades…” box on each state and try some trending strategies to see what happens…??? Thanks for your kind words - I hope this thread produces some additional insights for everyone. Mark | |||
| Jim Dean Sage Posts: 3433 Joined: 3/13/2006 Location: L'ville, GA User Profile |
My development work is complete. You’re welcome to use it any time! | |||
| Buffalo Elite  Posts: 603 Joined: 7/11/2007 Location: Braintree, MA User Profile |
Love the micro zones. Should work even better improving our ATM results. THANK YOU for scoring another big win. I hope to help out as soon as I get all these ARM strats retrained on our new 300 large list. I also built some extra ARM strats - like a CRT-3 NN version. But these cluster NNs take forever. But I promise to contribute! I noticed you did NOT incorporate your cutoff zones into this micro-state ATM. Is there a reason or just haven't gotten to trying that in it yet? Also it would be interesting to see the side-by-side apples-to-apples comparison of your new micro state against current Universal (w and w/o cutoff) | |||
| mholstius Regular Posts: 76 Joined: 6/15/2011 User Profile |
Actually, further testing with my FTM account and another test showed a negative effect in the real world using my original cutoff zones - so I’ve tabled that concept for the moment. The cutoff rules can probably be constructed so that they help, but that’s a future project… Another reason for not including the cutoff zones is that I try to limit any testing in ATM to one variable at a time. It’s an extremely powerful tool and can do so many interrelated things that you can easily confuse things if you try to test more than one thing at a time. Thanks for retraining the ARM strats! I wanted to do that, but I’m dealing with a lot of time constraints at the moment. I look forward to your results. Thanks for your input & help, Mark | |||
| Buffalo Elite Posts: 603 Joined: 7/11/2007 Location: Braintree, MA User Profile |
After I get them done and look at some test results I'll happily put the better ones up for people to download. | |||
| mgerber Regular Posts: 57 Joined: 3/30/2006 Location: Issaquah, WA User Profile |
A question, if I may: What is the nature of the 300 symbols? And how did you determine to use these? With that information, I might better understand the direction you traveled, and gain insight into how better to formulate a performance list. Thanks again, Mark G. | |||
| mholstius Regular Posts: 76 Joined: 6/15/2011 User Profile |
I didn't want to "bias" the list toward any particular group or industry, so I simply started with the "All Optionable U.S. Stocks" list and used the top 300 by money flow; SVXY was in that original list of 300, but I took it out of the list for these tests because it was having an unusual "outlier" effect on the statistics after it dropped 90% in 3 days in the drawdown at the beginning of Feb.; Good luck with your tests, Mark [Edited by mholstius on 2/27/2018 5:13 PM] Attached file : 00 Stocks List.jpg (61KB - 2331 downloads) Attached file : SVXY.jpg (103KB - 2317 downloads) | |||
| mgerber Regular Posts: 57 Joined: 3/30/2006 Location: Issaquah, WA User Profile |
--Mark G. | |||
| Barry Cohen Idol Posts: 2305 Joined: 1/1/1900 User Profile |
- It does not close trades when switching market states. The results were better with that disabled. - The Port Sim date range begins at 1/1/2014. If you want it to start at 1/1/2003, just modify it in the Port Sim Settings. - We changed the VBX-3 CB strategy to VBX-3, so that it wasn't limited to club members. The results are a little better with VBX-3 CB though. | |||
| THOMAS HELGET Elite Posts: 610 Joined: 3/22/2006 Location: BALDWINSVILLE, NEW YORK User Profile |
Thank you - Nirvana's effort to provide Mark's research to us is much appreciated! Tom Helget | |||
| Buffalo Elite Posts: 603 Joined: 7/11/2007 Location: Braintree, MA User Profile |
If I want to move a custom ATM method from one computer to another what files do I need to move? I tried the profile, the stuff in the ATM folder and the stuff in the VBA/ITS folder but it didn't move it successfully And thank you for the micro states .exe! | |||
| THOMAS HELGET Elite Posts: 610 Joined: 3/22/2006 Location: BALDWINSVILLE, NEW YORK User Profile |
I noted one little problem with the Chart Template provided for Mark's Micro States Method. Instead of providing the proper Color Chart setting for the SPY Micro State the Chart Template you supplied did so for each individual instrument. This can easily be modified by going to Chart Options...Color Charts and editing the Symbol Input to SPY for each of the three Background Color Charts rule as shown: Alternatively I have attached a modified Color Chart Template wherein that change (and a couple of other minor changes for style) has been made. Tom Helget [Edited by THOMAS HELGET on 3/2/2018 11:12 AM] Attached file : MicroState Chart Template Symbol Modification.PNG (41KB - 2149 downloads) Attached file : MicroStates.otd (128KB - 379 downloads) | |||
| Barry Cohen Idol Posts: 2305 Joined: 1/1/1900 User Profile |
If you move the profile & the method, that should be sufficient, as long as the destination machine has all the strategies & VBA code that was used. You do not need VBA/ITS. You do want to make sure both installations are on the latest update. If you save a method to your profile, which makes the "Custom [method]" that is profile specific, then you only need the profile. | |||
| Barry Cohen Idol Posts: 2305 Joined: 1/1/1900 User Profile |
| |||
| Jeff Brown Member Posts: 31 Joined: 1/29/2008 User Profile |
Very impressive! Thank you for your efforts and willingness to share over the many years. I've been doing a few tests and made a few tweaks -- ran a set of tests over different time periods and it has been hard to consistently beat or improve on the results across different time periods of back testing. Here's a couple that I've found that so far have stood up. Keeping everything the same as the original post and as described in your pdf I changed: 1) Removed $VIX from the symbol list. I don't think this is tradable, maybe I'm wrong, but I just removed it. 2) Activated: "Close all positions" for the Default state. This appears to let the trades from a Bull or Bear move longer. As Barry had mentioned they deactivated all the "close all positions" for all the market states because it produced overall better results. In my testing, I could reduce the average loss, improve the profit factor, calmar, and a few others without much deterioration of other statistics. It didn't produce as high of returns as fully deactivating the close all function, but for me, the decrease in return was acceptable given the improvement. 3) Changed the VTY_PRICE(10,5) to VTY_PRICE(4,1) for the Trade Ranking on all market states. This change had a large impact, but it also perplexes me. If you try to adjust the indicator on a chart, you can lower the smoothing factor all the way down to 1, only to 2. The formula doesn't restrict you and you can insert 4,1. I haven't yet looked at the formula behind this to see what's happening, but of all the indicators (and I tested over 20), the VTY_PRICE seemed to perform the most consistently. I'm not sure what OT is doing behind the scene with a 4,1 but the results were interesting. I also tested switching the Bear to the top so the flow of the market states would be: A) Check for Bear first B) Then check for Bull C) Use Default This produced worse results across the board and activating "Close All" didn't improve the results over the original Bull/Bear/Default order. I'm still exploring. Thanks, Jeff B | |||
| mholstius Regular Posts: 76 Joined: 6/15/2011 User Profile |
We’re all time limited, and passing on what you’ve discovered really helps. It can keep people from wasting time on things that don’t work, and can help to move things along in the right direction with things that do. I’ll have to try the change in the VTY setting... along with your "Close All Positions" suggestions. Thanks for sharing! Mark [Edited by mholstius on 3/5/2018 12:05 PM] | |||
| Buffalo Elite Posts: 603 Joined: 7/11/2007 Location: Braintree, MA User Profile |
FYI - $VIX and $SPX are not to trade but are in the focus list because some strategies use those symbols in filter blocks (or GAs/NNs for club strats). If you right click on them and look under enabled strategies they should both say "none". IOW NO strats will trade them (or generate signals) but if the symbols are used by a strat the symbols need to be there so the strat can get the info (C? H? EMA? etc). So be careful before you delete stuff - it may be there for a good reason FWIW | |||
| SteveJ New User Posts: 2 Joined: 2/2/2017 User Profile |
$VIX (as is $SPX) is required for some strategies so, I think, has to be in the focus list if those strategies are employed. On a different issue, do you know where the code for VTY_PRICE(x,y) can be found? Steve | |||
| Buffalo Elite Posts: 603 Joined: 7/11/2007 Location: Braintree, MA User Profile |
' * Volatility Price Ratio ' * by Matthew Greenslet ' * ' * Summary: ' * Indicator calculation from OT2009 ' * ' * Parameters: ' * Volatility Periods ' * Smoothing Periods ' ' Copyright 2009 Nirvana Systems, Inc. Austin, Tx, USA. ' *************************************************************************** #Indicator #PARAM "VolatilityPeriods", 14 #PARAM "SmoothingPeriods", 5 Dim fVPR as Single fVPR = 0 If C <> 0 Then fVPR = SMA(ATR(VolatilityPeriods)/C, SmoothingPeriods) End If Plot("VPR", fVPR) Return fVPR Attached file : OTindVolatilityPriceRatio.txt (0KB - 360 downloads) | |||
| Jim Dean Sage Posts: 3433 Joined: 3/13/2006 Location: L'ville, GA User Profile |
| |||
| Jeff Brown Member Posts: 31 Joined: 1/29/2008 User Profile |
I'll add $VIX back now that I know where to look to see that it won't be traded. None of the strategies in Mark's doc used it so it was unnecessary, but it makes sense given its use with some strategies -- just easier to switch and experiment. I haven't yet tried to adjust the Trade Ranking given the VTY_PRICE formula. I think it can be simplified to ATR(4)/C. I'll test it out later tonight after my system finishes its daily chores. I had run another test that came up with VTY_PRICE(1,1) with the same result as 4,1. Now that I see the formula, I'll see if simply ATR(x)/C is as effective. [Edited by Jeff Brown on 3/5/2018 4:13 PM] | |||
| Jim Dean Sage Posts: 3433 Joined: 3/13/2006 Location: L'ville, GA User Profile |
I advise you to include a divide by zero test in your simplification: iif( C = 0, 0, ATR(4)/C ) In some cases it is possible for C to have a zero value and that would cause the program to crash. This should prevent that. [Edited by Jim Dean on 3/5/2018 4:15 PM] |
|
|
|
|
Administrator
Forum Moderator |
Registered User
Unregistered User |
 E-Mail this thread to a friend |
Toggle e-mail notification |
