OmniTrader Professional Forum
OmniTrader Pro General Discussion
OT 2018 ATS With VBX3 And Large Cap Stocks
27 replies, 4047 viewings
|
OmniTrader Professional Forum
OmniTrader Pro General Discussion
OT 2018 ATS With VBX3 And Large Cap Stocks
|
Last Activity 2/23/2025 4:01 AM 27 replies, 4047 viewings |
|
|
| Printer friendly version |
mholstius Regular  Posts: 76 Joined: 6/15/2011  User Profile |
I hope the following info will allow you to duplicate my results and see for yourself how much ATS improves things… How I did it... Load in Nirvana’s Large Cap Stocks list, use only the VBX3 strategy, run the To Do list to update everything, and save the data for use in Port Sim; Running the To Do list will take a while, but when it’s done go to Port Sim and choose Large Cap Stocks for the Focus List; I used $100,000 and 2X Leverage for my example; It’s important that the parameters are the same for both the original and ATS runs, so I set both to use 5% per trade allocation; At the bottom left of Port Sim, click on ATS settings, click on these 3 choices, and then Close; In the top middle of the Port Sim, click on % of Equity and make sure the default settings are correct (you can modify them in the boxes if necessary); Click on Reversal and make sure those are correct (you can modify them in the boxes if necessary) Notice that the first 5 should be the same as the first 5 in % of Equity (so we’re comparing Apples to Apples) The last 2 have to do with ATS settings; Click on Run Analysis. It takes a while, but below are my results using BT 1/1/00-1/1/07 and FT 1/1/07-1/1/17, VBX3, and Large Cap Stocks; (Sorry, the image is a bit wide in order to make all the info visible) To see how ATS improves the results, click on the Reversal tab; The improvement is quite obvious. You can also click on the Summary tab to see both overlaid; I wasn’t in on the development of this new capability, but I’m certainly impressed. I was dubious after seeing such good results, so I exported the trades for both the original run and the ATS run and put them into excel for analysis. I can confirm the following; The original took 15,813 VBX3 trades. The ATS run took 9,182 of those trades and did not take 6,631 of the original trades. ATS didn’t add any trades, it just selected a much better subgroup of the original set. I want to learn more about how ATS works... I don’t know any more than you at this point, but these are truly impressive results, just by using the default settings published by Nirvana Maybe we can do even better...??? BTW - I intentionally left 2017 out of these tests, and plan to post those results in the near future. Hope this helps, Mark [Edited by mholstius on 12/4/2017 4:47 PM]  Attached file : 01 ATS.jpg (166KB - 1485 downloads) Attached file : 02 ATS.jpg (358KB - 1523 downloads) Attached file : 03 ATS.jpg (251KB - 1466 downloads) Attached file : 04 ATS.jpg (380KB - 1542 downloads) Attached file : 05 ATS.jpg (110KB - 1467 downloads) Attached file : 06 ATS.jpg (80KB - 1462 downloads) Attached file : 07 ATS.jpg (135KB - 1449 downloads) Attached file : 08 ATS.jpg (148KB - 1506 downloads) Attached file : 09 ATS.jpg (99KB - 1422 downloads) Attached file : 10 ATS.jpg (117KB - 1428 downloads) Attached file : 11 ATS.jpg (497KB - 1483 downloads) Attached file : 12 ATS.jpg (438KB - 1443 downloads) Attached file : 13 ATS.jpg (501KB - 1563 downloads) | ||
| Eric S. Member Posts: 5 Joined: 12/2/2017 User Profile |
>> Wondering ahead if this (ATS) will port to OV (when included in an OT Portfolio). ...Or if Nirvana can build an equivalent feature in OV. | ||
| THOMAS HELGET Elite   Posts: 610 Joined: 3/22/2006 Location: BALDWINSVILLE, NEW YORK User Profile |
I agree that the comparison results are truly amazing! But my concern is how the two ranking parameters in the Reversal ATS Method - RSI(3) (LONGS) and ADX(10) (SHORTS) for Rank 1 and TRII(60,40) for both LONGS and SHORTS for Rank 2 and then their subsequent weighting of 100 for Rank 1 and 10 for Rank2 could have ever been even conceived of initially. That is, have you tried to create your own ATS method? If so could you create one for, say, breakout stocks (we only have Reversal and Trend Reversal Methods currently)? What indicators and parameters would you suggest? What would be your weighting of those indicators? Just some thoughts - to my mind a little too "black boxy" (or maybe a lucky guess), especially with the weighting factor. Tom Helget [Edited by THOMAS HELGET on 12/4/2017 8:55 PM] | ||
| Eric S. Member Posts: 5 Joined: 12/2/2017 User Profile |
re: your comment: The ATS run took 9,182 of those trades and did not take 6,631 of the original trades. Does it tell us that (the actual # ATS took) anywhere (i.e. in Port Sim)? Thanks Eric | ||
| Mel Veteran Posts: 235 Joined: 3/18/2006 User Profile |
For breakouts try something like a volatility squeeze, i.e. ( BOL_UPPER(20,2)-KLT_UPPER(20))/KLT_UPPER(20) I notice that z-score works better than all the stuff showed so far for rtm, by a significant margin. Mel | ||
| THOMAS HELGET Elite Posts: 610 Joined: 3/22/2006 Location: BALDWINSVILLE, NEW YORK User Profile |
Thanks! Can you maybe elaborate on the z-score for Return to Mean with some sort of mathematical formula, please? Tom Helget | ||
| Mel Veteran Posts: 235 Joined: 3/18/2006 User Profile |
Attached file : zScoreInd.txt (0KB - 312 downloads) | ||
| THOMAS HELGET Elite Posts: 610 Joined: 3/22/2006 Location: BALDWINSVILLE, NEW YORK User Profile |
Thanks very much and Happy Holidays! Tom Helget | ||
| Mel Veteran Posts: 235 Joined: 3/18/2006 User Profile |
| ||
| THOMAS HELGET Elite Posts: 610 Joined: 3/22/2006 Location: BALDWINSVILLE, NEW YORK User Profile |
Yes, good reasoning there! Tom Helget | ||
| THOMAS HELGET Elite Posts: 610 Joined: 3/22/2006 Location: BALDWINSVILLE, NEW YORK User Profile |
One of your comments at the end of your post didn't make sense to me. You commented that: I can confirm the following; The original took 15,813 VBX3 trades. The ATS run took 9,182 of those trades and did not take 6,631 of the original trades. ATS didn’t add any trades, it just selected a much better subgroup of the original set. Yet your screen shots clearly show each Method (% of Equity and Reversal) taking around 15,000 trades each. And that makes sense to me as the VBX3 Strategy throws off a large number of possible trades on the huge Large Cap Stock list (2362 issues strong on today's OmniScan). As a matter of fact 46 possible Long and Short trades were available for the taking today alone. Certainly more than enough to choose from to get a full 20 trades at 5% allocation per trade even if no other Symbols were currently in trade. So, I would posit that what the ATS Reversal Method was doing was selecting the very best of those scadzillion trades whereas the % of Equity Method was just taking the trades alphabetically. And, on that matter, One thing you did not describe in your write-up is how your Trade Selection was done - I would assume you just left it the standard Symbol Ascending order. Now I always hated that since, what if, the "Z" symbols were the hot pistols currently? So I like to use something a bit more ranking in it's own right - Advisor Descending: Now when I did that I got roughly the same results you achieved with the Reversal Method but a bit better Equity Curve for the % of Equity Method: Now to me this would show that the ATS Reversal Method is selecting it's trades solely on the Ranking in the Method whereas the % of Equity Method could be bettered a bit by using the Advisor Descending sort in place of the customary Symbol Ascending order. Tom Helget [Edited by THOMAS HELGET on 12/6/2017 7:04 PM] Attached file : Trade Selection for VBX3 Large Cap Stocks.JPG (38KB - 1297 downloads) Attached file : VBX3 Large Cap Stocks Portfolio Simulation Ranked By Advisor.JPG (191KB - 1310 downloads) | ||
| mholstius Regular Posts: 76 Joined: 6/15/2011 User Profile |
And thanks for the detailed snags - I always like those. Unfortunately, I'm quite busy with some family things this week and won't be able to give this the attention it deserves right now. Just wanted to let you know I'm not ignoring you. I also don't like Alphabetic - and ranking by Advisor looks like a nice improvement. Thanks for your understanding, Mark | ||
| Mel Veteran Posts: 235 Joined: 3/18/2006 User Profile |
Report: Generated 19,490 trades, 51.28% hit rate, ppt .02 (combined long and short trades.) % of Equity PortSim: lost $5200 in 3000 bars, taking 15257 trades, 50.2% hit rate, .99 profit factor. z-score ATS sort on % of equity allocation: made a profit of $60,932,464. Took 15274 trades, 61.5 % hit rate, profit factor 2.02. Seem to good to be true. Attached file : portSimATS.PNG (74KB - 340 downloads) | ||
| THOMAS HELGET Elite Posts: 610 Joined: 3/22/2006 Location: BALDWINSVILLE, NEW YORK User Profile |
Yes, it does since it is a bit like using a filter to an already established system. Of course you can't weight multiple filters and so in that right the ATS is certainly superior. Tom Helget | ||
| Vinay Regular Posts: 70 Joined: 1/18/2012 Location: Planet Earth User Profile |
Originally written by Mel on 12/8/2017 10:20 AM Seem to good to be true. In the past I also got "too good to be true" results which turned out to be OT bugs. Here are the links for those cases: 1. https://www.omnitrader.com/currentclients/otforum/thread-view.asp?threadid=8864 2. https://www.omnitrader.com/currentclients/otforum/thread-view.asp?threadid=11223 After these "too good to be true" experiences, I never had complete faith in the results shown by OT. This may also turn out to be one of those cases. Who knows? BTW both the bugs mentioned above were reported several years ago still remain unfixed. [Edited by Vinay on 12/8/2017 8:31 AM] | ||
| LSJ Veteran Posts: 190 Joined: 7/23/2012 Location: Citrus Springs, FL User Profile |
zscore in red and will%r in orange. [Edited by LSJ on 12/8/2017 2:24 PM] Attached file : zwill.jpg (232KB - 391 downloads) | ||
| Mel Veteran Posts: 235 Joined: 3/18/2006 User Profile |
So I would expect a momentum indicator to improve a non-momentum system more that another momentum indicator would. I would expect unbounded indicators to do a better job of sorting than bounded ones (that can get pinned at 100) for example, and fail to make meaningful distinctions among the better trades. It takes a lot of trades to see the effects of ATS indicators clearly. But once we know the effects, we can choose the right ones for different kinds of systems, even those that do not generate lots of trades. Potential ATS indicators include, momentum, z-score, linreg slope, candle patterns, distance form last pivot, volatility changes, relative money flow, relative volume, implied volatility, distance to support/resistance, signal to noise ratio, trading range, and tradeability. My advice is to understand what you system is doing, at its root, widen your filters to let more trades through, and try those indicators that do something different from your system in your ATS experiments. Mel [Edited by Mel on 12/9/2017 8:22 AM] | ||
| Mel Veteran Posts: 235 Joined: 3/18/2006 User Profile |
Mel | ||
| mgerber Regular Posts: 61 Joined: 3/30/2006 Location: Issaquah, WA User Profile |
I'll begin with z-score to see if I can get something to work in ATS. --Mark G. | ||
| mgerber Regular Posts: 61 Joined: 3/30/2006 Location: Issaquah, WA User Profile |
Initial testing shows that the corrected ATS results are vastly different from the suspicious pie-in-the-sky results that have been quite properly viewed with a jaundiced eye. --Mark G. | ||
| mgerber Regular Posts: 61 Joined: 3/30/2006 Location: Issaquah, WA User Profile |
I began by trying my hand at making a zScore ranking formulation, but have failed at every attempt. Parsing Mel's OLang code for the zScore indicator, I thought that it would be something like this: ((C - SMA(4)) / STD(4)) But I get multiple character invalid compile errors. Any thoughts as to how this is properly done? Thanks, Mark G. | ||
| Jim Dean Sage   Posts: 3433 Joined: 3/13/2006 Location: L'ville, GA User Profile |
dim X, Y as single X = Std(4) if X > 0 then Y = (C-sma(4)) / X Else Y =0 End if | ||
| mgerber Regular Posts: 61 Joined: 3/30/2006 Location: Issaquah, WA User Profile |
Reversal used to be based on RSI(3), as depicted in Ed's webinar. NOW, it is based on VTY_PRICE(14,5) instead! Interesting. I replicated the RSI(3) coding, and it seems to perform nearly the same (and sometimes slightly under performs) the straight equity results. Even more interesting. This is all seemingly the probable result and reason for the "bug error" correction made in the Friday OT2018 update? We will not know for sure, of course, until/if/when Nirvana makes official comment. --Mark G. [Edited by mgerber on 12/10/2017 1:58 PM] | ||
| mgerber Regular Posts: 61 Joined: 3/30/2006 Location: Issaquah, WA User Profile |
Thanks for the comment. But what straight formula can I insert into ATS? I tried everything I could think of to massage what I deciphered from Mel's coding, and could not get anything to work without compile errors. Did I improperly decipher the code? #Indicator #param "Zlength", 4 #param "buylevel",-1.6 #param "shortLevel",1.6 dim av,st,zs as single if Bar>Zlength*2 then av = SMA(C,Zlength) st = STD(C,Zlength) zs = (c-av)/st plot("zScore",zs) plot("buyLevel",buyLevel,black) plot("shortLevel",shortLevel,black) end if Return zs ' Return the value calculated by the indicator --Mark G. [Edited by mgerber on 12/10/2017 2:03 PM] | ||
| Jim Dean Sage Posts: 3433 Joined: 3/13/2006 Location: L'ville, GA User Profile |
But anything looking for a script in OT can use a function call of an OLang indicator, which the code I showed could be a part of - just put #indicator at the top and return Y at the bottom. In theory, iif() could be used but sometimes I've found it to be problematic: Iif(std(4)=0,0,(C-sma(4))/std(4)) |
|
|
|
|
Administrator
Forum Moderator |
Registered User
Unregistered User |
 E-Mail this thread to a friend |
Toggle e-mail notification |
