OmniTrader Upgrade Forums
ATM
Xsuite with ATM Micro Macro
46 replies, 10806 viewings
|
OmniTrader Upgrade Forums
ATM
Xsuite with ATM Micro Macro
|
Last Activity 4/21/2021 9:48 AM 46 replies, 10806 viewings |
|
|
| Printer friendly version |
LSJ Legend  Posts: 515 Joined: 8/17/2006 Location: Citrus Springs, FL  User Profile |
In the meantime I continue to test hypotheses. A little like the guy who took his chainsaw in because it was cutting slow and when the mechanic started it he said "what's that noise?" | ||
| EYEGUY Icon   Posts: 1543 Joined: 12/12/2003 Location: BALDWINSVILLE, NEW YORK User Profile |
I don't know how you performed your calculations for the Bear & Bear Market State. However I am attaching the same Excel Spreadsheet I presented originally with the addition of a summation for the four Strategies utilized in the Bear & Bear Market State showing that the figures presented in the Excel Pivot Table for that Market State are indeed correct. Remember that the Total for each Market State in the Pivot Table is given above it's various components and not below like we might perform conventional addition. Although I haven't done the same for the other eight Market States this is enough to convince me that the Pivot Table I originally presented is correct. Tom Helget [Edited by EYEGUY on 3/19/2018 11:20 AM]  Attached file : Trades Best Strategies.xlsx (669KB - 372 downloads) | ||
| EYEGUY Icon Posts: 1543 Joined: 12/12/2003 Location: BALDWINSVILLE, NEW YORK User Profile |
I think the problem here is that while the way Mark suggested testing each Strategy is a good idea since it is performed entirely within the ATM Micro Macro Market State paradigm (even allowing for trade Ranking) it doesn't allow for the full weight of trades the Strategy could have possibly assumed when a greater amount of capital from previous trades in precursor Market States has been made available to it. That is, only a handful of trades are being taken when we test each Strategy individually as Mark suggested. When all the Strategies are deployed they provide the Strategies going forward in each Market State with excess capital over and above that used in the Single Strategy Test which can be (unfortunately) deployed into loosing trades. More and more trades (or larger and larger trades) can thus be taken with this excess capital some of which might assume losses which swamp the gain the single tested Strategy would have implied. Therefore the only real way to supply the ATM Method with viable Strategies would appear to be to throw them into the mix and see what happens via Excel Pivot Table Analysis. And, yeah, that sounds like curve fitting. But the idea here is weather the storm in a variety of past market conditions with Strategies that have been proven to hold their own so that we might be hopeful of them doing the same in future market conditions. Tom Helget | ||
| mholstius Veteran  Posts: 175 Joined: 1/13/2017 User Profile |
I’m still not home, but would like to suggest that you set the trading % of equity for your testing to 1% ( and possibly also use the 2X margin to “take additional trades” ). That way you should never hit an available equity limit and you’ll see all the trades taken by every state / strategy combination, assuming they’re also set to let the trades complete (not close on change in market state). If you also set all the states to take the same # of trades (maybe 5L or 5S?), that can help to remove some bias from one strategy taking more trades than another. It might give a better picture of the relative performance of one strategy vs another in each state. Once you identify the good strategies, you can then decide how many trades to allow. Just a suggestion... Good luck in your testing, Mark | ||
| EYEGUY Icon Posts: 1543 Joined: 12/12/2003 Location: BALDWINSVILLE, NEW YORK User Profile |
I already did the 5 Long, 5 Short trade thing for my analysis - this was just to add balance and not give an edge to trading in any Market State. And I had 2X margin in play on $100,000. But my Trade Allocation was set to 10% so the 1% Trade Allocation is a real good idea to anticipate all the trades that likely could be taken at higher allocations. Thanks, Tom Helget | ||
| EYEGUY Icon Posts: 1543 Joined: 12/12/2003 Location: BALDWINSVILLE, NEW YORK User Profile |
OK, I followed your suggestions. As you might recall one of the bones of contention was the negative performance of the Swing 13 Strategy in the Bear and Bull Market State as denoted in the Excel Pivot Table: I only completed the first two Market States in my analysis but note that once again I would choose Swing 13 to be included in the top 4 Strategies for the Bear and Bull Market State: I know you will say, "but what did the Portfolio Simulation curve look like using 1% Trade Allocation?". Quite nice actually with a very small Maximum Drawdown and Average Annual Maximum Drawdown: So I am still stumped as to exactly how the very poor perfomace of Swing 13 in the Bear and Bull Market State could have been produced. Tom Helget [Edited by EYEGUY on 3/20/2018 11:59 AM] Attached file : Best Strategies With Negative Results.PNG (46KB - 1126 downloads) Attached file : Swing 13 Bear and Bull Final Equity 1% Allocation.PNG (90KB - 1167 downloads) Attached file : Swing 13 Bear and Bull PS Equity Curve 1% Allocation.PNG (353KB - 1177 downloads) | ||
| mholstius Veteran Posts: 175 Joined: 1/13/2017 User Profile |
I’m still away from home, but that equity curve you posted for Bear and Bull has a majority of its trades in states that are not Bear and Bull...??? Are all the states other than Bear and Bull set for 0% equity? Mark | ||
| EYEGUY Icon Posts: 1543 Joined: 12/12/2003 Location: BALDWINSVILLE, NEW YORK User Profile |
As always you have an eagle eye! And, no, ALL the Market States were set to 1%. However, I thought that no trade would be taken regardless of the % of Equity in the Market State unless that Macro & Micro Market State was "in play". What am I missing there? Is it maybe that your ATM hierarchy is not like that in ATM Universal? That is we are not "falling" from the most desired Market State to the least desired Market State - i.e. looking for Bull Market trades then Bear Market Trades and anything else is Default if it is not Bull or Bear? Thanks, Tom Helget | ||
| mholstius Veteran Posts: 175 Joined: 1/13/2017 User Profile |
The hierarchy is a bit different in that M+M starts with the Macro Bear instead of a Bull state, but ATM still starts at the top of the list of states and then stops at the first one in the order that satisfies the state formula. If you want to see how Swing 13 performs only in Bear + Bull, you have to set all the other states to 0% equity and then have ONLY Swing 13 selected in the Bear + Bull state (uncheck all the other strategies in Bear + Bull). In the equity curve, you’ll see that all the other states are “flat lined” (except for the trades that entered in the Bear + Bull state and closed later). The stats will also just reflect the performance of swing 13 in Bear + Bull. Try it - you’ll see what I mean. Then, do the same thing with whatever other strategy and state you want to see. You have to do them one at a time ( one state trading with 1% equity trading one strategy at a time ) Mark | ||
| EYEGUY Icon Posts: 1543 Joined: 12/12/2003 Location: BALDWINSVILLE, NEW YORK User Profile |
Ah, that makes good sense - I guess I am back to the drawing board once again! Thanks again for your continued assistance in understanding the Micro Macro ATM Method! Tom Helget | ||
| EYEGUY Icon Posts: 1543 Joined: 12/12/2003 Location: BALDWINSVILLE, NEW YORK User Profile |
So when we last left off we were looking for a solution to eliminate the Swing 13 Trading Strategy from consideration in the Bear & Bull Market State. To that end I set the Trade Percentage of Equity of Allocation in all Market States except for the Bear & Bull Market State to zero. I set the Trade Percentage of Equity Allocation in the Bear & Bull Market State to 1%. The Allocation Ratio in the Bear & Bull Market State was set to take 5 Long Trades and 5 Short Trades. Everything else was left alone in the Macro & Micro ATM Method you presented and I used a short Focus List of 81 Optionable Stocks. Like before Swing 13 put in an excellent show and could be considered in the top 4 Strategies for Final Equity in the Bear & Bull Market State: And here is the Portfolio Simulation Equity curve: Curiouser and curiouser! Tom Helget [Edited by EYEGUY on 3/20/2018 7:20 PM] Attached file : Portfolio Final Equity with 1% Trade Allocation in Market State Only.PNG (70KB - 1065 downloads) Attached file : Swing 13 in Bear & Bull Market State with 1% Trade Allocation ONLY.PNG (357KB - 1123 downloads) | ||
| mholstius Veteran Posts: 175 Joined: 1/13/2017 User Profile |
That equity curve and the associated stats looks much better... If you’re planning on trading just your 81 optinable stocks, that looks like a good choice. However, I’d suggest a larger list of symbols to minimize any survivorship bias or curve fitting...??? Good luck with your testing. I suspect some others may have learned from this discussion.too. Mark | ||
| EYEGUY Icon Posts: 1543 Joined: 12/12/2003 Location: BALDWINSVILLE, NEW YORK User Profile |
Respectfully, that's just it. What have we learned? The Swing 13 Strategy when isolated in the Bear & Bull Market State produces one of the best looking Equity Curves and Final Equity of the twelve Strategies examined. Yet when all the Strategies are operational and the full Micro Macro ATM method is in play it falls flat on it's face in the very same Market State one would think it would perform well in: This proves that Strategy isolation is not a suitable method to aid one in selecting Strategies to employ in the ATM Method under consideration. Tom Helget [Edited by EYEGUY on 3/20/2018 8:38 PM] Attached file : Best Strategies With Negative Results.PNG (46KB - 1064 downloads) | ||
| mholstius Veteran Posts: 175 Joined: 1/13/2017 User Profile |
I’d question what was being analyzed and presented by excel. We know that the equity curve and stats show what happens in the Bear + Bull state using Swing 13 using the choices made in the ATM settings. By using 1%, we can also see that an AVG of only 0.4% of equity is being used - so there’s probably very little interaction between the states. I’d go with the known stats and continue trying to identify additional good strategies using that process. It’s tedious, but we know that the stats are correct. Until you’re sure what’s being analyzed by excel, that’s an unknown... Mark | ||
| Buffalo Bill Legend Posts: 539 Joined: 10/3/2006 Location: Stafford, VA User Profile |
The first sheet has a box showing each M&M state and the Long, Short or Both strats that show promise noted for each. **NOTE: you'll see lots of strats that do well long or short but not both for a given M&M state hence my recommendation ATM be modified to select L or S or both for each start in each MS. If you agree pass it along to Barry and Ed** There is a "score" column on sheet 1 - it is a rough strategy ranking of how each did overall. I gave each strat 1 pt for looking tradeable in the L, S or Both for each MS. So a score of 18 - 9 MSs and Both L&S = 9x2 = 18pts. TLB2 RSI, WT3 ST, NSP-35 CB scored very well (all > 13). CRSI, CRT-3, VBX-3 NN, lots of X suite strats, MVX-15, ARM5 RTM17, WT3 LT all also scored well. Those are a good set of strats to include in your ATM strategy selection testing The second sheet is all the numbers for every strat broken down by M&M state The last sheet is where I did a PRELIMINARY test on using the MS filters on each symbol as well as SP500. I selected Macro Default+Micro Bear as the MS. Then I added a confirmer to only pass signals if the SYMBOL was in Macro Default as well. Then Symbol in Macro Def AND Micro Bull, and lastly no macro but just Micro Bull. Quick analysis tells me this is something we need to explore - the strat results IMPROVED - higher HRs and PPTs. This means we could risk more per trade potentially and min MDD. IDK yet - I did a quick look at ONE M&M and did NOT have a chance to try adding these symbol states in ATM. Note - IGNORE THE t3-sx stuff at the bottom of the last sheet. Something else I am heading to the airport for a trip to Beijing. I will look at this more but I hope someone smarter than me can play around with that idea. It could be really good for our ATM results I think. Also, this was just my first set of strat tests - I will start on another set of more strats, most with retrained ARM blocks (on SP100+NAS100 list) OR with newly added ARM blocks (my NNs) if none were available (like CRSI, CRT-3, TLB2 RSI for ex). When complete I will share my results I don't know how the .pdf version will look - gotta go but thought it might be good if someone doesn't have excel Attached file : MM Strat results.pdf (152KB - 380 downloads) Attached file : MM Strat results.xlsx (96KB - 375 downloads) | ||
| mholstius Veteran Posts: 175 Joined: 1/13/2017 User Profile |
That's a lot of work, and a lot of good data! I just got back home last night, and it'll take me a while to digest it all... Thanks for sharing it - hope you had a good trip over to Beijing. Mark | ||
| SteveL Veteran Posts: 262 Joined: 8/19/2005 Location: Boulder, CO User Profile |
Hope you had a good trip. I like your idea of using the Confirm Block with the two systems. But, ... Instead of a PortSim report, wouldn't the File->PrintReports->Strategies->PerformanceSummary be more useful/instructive about how well each of these strategies performed for each market states? And you could run all the strategies for a particular market state in the same analysis run. And then select the best for each market state for inclusion in the ATM market state. Of course, there is that ever present "curve fitting" problem because you'd be choosing strategies that you know worked well during each of those periods. But I think that problem exists with your current approach as well (if I'm correctly interpreting what you are doing). | ||
| Buffalo Bill Legend Posts: 539 Joined: 10/3/2006 Location: Stafford, VA User Profile |
That is actually the report I used - the one you suggested. I export it to excel then sort by HR to get candidates. Then I looked at the PS equity curve and stats for each strategy alone (as Mark suggested), in each M&M state, in port sim to decide on which to include in each M%M state. Then I ran a ton of comparative runs deleselecting each or multiple strats trying to find a better, and shorter, group for each. That was my process | ||
| SteveL Veteran Posts: 262 Joined: 8/19/2005 Location: Boulder, CO User Profile |
Thank you for the data. I'm going to go look at it again. | ||
| Keith Parsons Regular Posts: 76 Joined: 6/28/2009 Location: Durban, South Africa User Profile |
Thank you very much for sharing these Systems. I want to use your work in the Confirming Blocks in Mark's MacMic strategies. However I note the compiling issue with "End If" I would like to know if you have had this compiling problem. If not did yours work correctly in the Confirming Block ? NB: My motivation for this is that in MacMic Concurrent and running the ATM3 Method Settings I get unrealistic low curves. In fact it is lower than even that % Equity curve. Hoping that your Systems help rectify this. Apologies for any inconvenience Kind regards, Keith Parsons FileName: sysMSMicroBullL.txt; Line: 6; 'If' must end with a matching 'End If'. FileName: sysMSMicroDefL.txt; Line: 5; 'If' must end with a matching 'End If'. Build failed. Attached file : sysMSMicroEndIf Error.jpg (20KB - 262 downloads) | ||
| jpb Veteran Posts: 168 Joined: 5/11/2005 Location: Brown Deer, WI User Profile |
I haven't tried Buffalo's systems but in looking at the code, it looks like it just needs one more End If added. Jim educated me previously that within VB, IF can take multiple formats, one of them being a single line IF which doesn't need an End If. I suspect that was how OT processed one of the IF statements originally but perhaps something changed. I've attached a small change to the sysMSMicroBullL.txt to add in another End If and indented the last If statement for easier reading/grouping of the IF statements. I believe Buffalo's original intent was to wrap the multiple inner IF statements in one that checked for a close price greater than 0; a bad data check. If that works for you, you can apply the same extra End If to the remaining files. I spot checked a few and they all looked uniform in format and structure. [Edited by jpb on 2/12/2020 7:50 AM] Attached file : sysMSMicroBullL.txt (0KB - 254 downloads) | ||
| Keith Parsons Regular Posts: 76 Joined: 6/28/2009 Location: Durban, South Africa User Profile |
Really appreciate your responding to my post. Very kind of you. Most certainly I will post any positive aspects that I come across on Buffalo's code in the next few day. Many thanks, Keith |
|
|
|
|
Administrator
Forum Moderator |
Registered User
Unregistered User |
 E-Mail this thread to a friend |
Toggle e-mail notification |
