OmniVest User Forums
General Discussion Topics
ATM: First Impressions
40 replies, 5309 viewings
|
OmniVest User Forums
General Discussion Topics
ATM: First Impressions
|
Last Activity 4/15/2019 7:43 PM 40 replies, 5309 viewings |
|
|
| Printer friendly version |
SteveL Veteran   Posts: 189 Joined: 10/11/2012 Location: Boulder, CO  User Profile |
I've only been playing with this for a few days. I haven't done any automated optimization, other than taking ideas from the "Universal method" (which may be optimized) and combining it with a set of strategies that I've been using. Here is a snapshot for comparison with the previous LONG ONLY with 1x leverage. The point is to show that we may be able to improve on the already impressive Universal method. NOTE: There are no commissions in any of these examples, which of course will degrade the results. But I believe it is a fair comparison to the previous snapshots. [Edited by SteveL on 1/28/2018 5:30 PM]  Attached file : 2018-01-28_16-16-39 Universal 2 Long-Only 1x leverage.png (153KB - 919 downloads) | ||
| MarkG Member Posts: 6 Joined: 7/19/2015 Location: New York User Profile |
WOW...That is a huge improvement for the long only, 1X leverage. All of these ATMS equity curves look great, I hope there is no curve fitting occurring. I know Mark H addressed this at the top of this thread. | ||
| SteveL Veteran Posts: 189 Joined: 10/11/2012 Location: Boulder, CO User Profile |
These are curve fitted. There have been tweaks of many things to make these curves. I've tweaked the strategies, allocations, and # of trades per day. I'm sure the market states formulas were tweaked to give the best results with the set of strategies being analyzed, and the choice of time period also allows for tweaking to fit the bear, bull and other market states during this 19 year period. BUT, these curves are the results of many thousands of trades. So, it's not quite as bad as curve fitting to a 100 trades. That said, there is always curve fitting. That is how these strategies, indicators, etc. were developed - to produce a good back test, and hope it produces reasonable results going forward. | ||
| SteveL Veteran Posts: 189 Joined: 10/11/2012 Location: Boulder, CO User Profile |
One more comment - I think Jim Dean put it well in his 1/21/2018 post #36967 earlier in this thread. Don't overtweak. That is, do things that have a logical reason for being done. [Edited by SteveL on 1/28/2018 6:20 PM] | ||
| Mark S Regular Posts: 56 Joined: 10/11/2012 Location: Barrington, IL User Profile |
In any case, I'm committed to the Elite trading approach, and trying to decide how ATM will work with Elite trading to improve results. Intuitively, it seems that ATM would front end sets of strategies that are tuned to the different market conditions, but I'm not sure. Any thoughts/ideas? | ||
| gbarber Veteran Posts: 282 Joined: 12/30/2012 Location: Pearland, TX User Profile |
I forgot how to paste pictures into the text of these posts so just attached a word document with the pictures showing results and 'method' used. Jim Dean taught me how to do that a couple years ago but I have lost that info. Anyway, the point is that a little experimenting with ATMS could bring some pretty outstanding results without much effort. However, we all know that even with the most thoughtful and best testing using historical data, the results achieved in the real world on the right edge do not necessarily mimic the testing results. The market characteristics change making all that testing less and less relevant as time passes. That's why adaptive is the key word. ATM gives us that on the large scale with use of different strategies per market state. It also allows some adapting on a large time scale via optimization of strategies used and their parameters. Perhaps doing that periodically would help. Need to test that. But I think we will need something to adapt to short scale change of market characteristics to conquer the right edge on a continuing basis. Perhaps periodic retraining of the AI strategies would help. Perhaps use of filters like Lain Bogle is developing for ATS would help. Is the seminar coming up this week going to teach us how to divide the historical data into one chunk of back test and a set of smaller chunks of forward test so we could see how different means of adapting to market change would work? Attached file : ATMS results.docx (290KB - 296 downloads) | ||
| Jim Dean Elite  Posts: 1059 Joined: 10/11/2012 Location: L'ville, GA User Profile |
1. ATM is not yet avail in OVest therefore Elite Trader is not yet applicable with respect to it. 2. Ed explained that when ATM becomes avail in OVest that it will not be “uploadable” like an ET strategy, but rather will be another tool in the OVest Family which can be used to enhance any OVest Strategy/Portfolio, whether ET or native 3. Ed suggested that an ET author might recommend a particular ATM config in OVest (once avail), much like a certain patterns of settings might be suggested 4. If you buy ATM now then when it’s avail in OVest you’ll get to use it for free. If someone who does not own an early copy wants to use it in OVest later, ed suggested that the monthly cost might be $300-$500. | ||
| gbarber Veteran Posts: 282 Joined: 12/30/2012 Location: Pearland, TX User Profile |
Attached file : ATMS results.doc (303KB - 274 downloads) | ||
| Mark S Regular Posts: 56 Joined: 10/11/2012 Location: Barrington, IL User Profile |
| ||
| Jim Dean Elite Posts: 1059 Joined: 10/11/2012 Location: L'ville, GA User Profile |
| ||
| Mark Holstius Elite  Posts: 744 Joined: 10/11/2012 Location: Sleepy Hollow, IL User Profile |
I added additional strategies to the Nirvana selections and then spent some time the past week focusing on the “robustness” question: whether ATM methods will maintain their edge when applied to various lists. Once again, I’ve been impressed with what we’re able to accomplish with ATM - and we’ve only scratched the surface. I’ll put up a couple of examples. This shows my ATM Method vs Nirvana’s Universal using a list of 300 stocks from 1/1/2002; Restricting mine to 1X and longs only using that 300 stocks list; Restricting mine to 1X and longs only using the Nirvana stocks list (fewer stocks = fewer trades / profits); I have other examples, but let’s just say that I’m impressed with the consistency of ATM. MarkG: thanks for bringing up the 1X Long question. If we question whether the majority of profits are coming from the long or short position, it’s really easy to answer now. ATM isn’t just useful for the end result. It allows us to quantify the contributions of different variables. If you think you have a good ranking formula, you can just toggle it on & off to verify its effect. Same for filters, allocations, and even different definitions of the market states themselves. I’m really looking forward to being able to use this in OV… This is a fantastic tool on so many levels! Mark [Edited by Mark Holstius on 1/29/2018 11:11 AM] Attached file : 00 3oo stocks.jpg (583KB - 827 downloads) Attached file : 01 1X Long 300 stocks.jpg (539KB - 825 downloads) Attached file : 02 1X Long Nirvana stocks.jpg (550KB - 810 downloads) | ||
| MarkG Member Posts: 6 Joined: 7/19/2015 Location: New York User Profile |
Thanks, MarkG | ||
| Tom Helget Member Posts: 42 Joined: 10/11/2012 Location: Baldwinsville, NY User Profile |
I don't believe they have been posted just as of yet. Tom Helget | ||
| Jim Dean Elite Posts: 1059 Joined: 10/11/2012 Location: L'ville, GA User Profile |
:-) | ||
| Buffalo Bill Member Posts: 28 Joined: 1/2/2015 Location: Stafford, VA User Profile |
ref 2 things you mentioned - 2- "Ed explained that when ATM becomes avail in OVest that it will not be “uploadable” like an ET strategy, but rather will be another tool in the OVest Family which can be used to enhance any OVest Strategy/Portfolio, whether ET or native 3- Ed suggested that an ET author might recommend a particular ATM config in OVest (once avail), much like a certain patterns of settings might be suggested" it seems to me to utilize ATM as its intended we have to re-think how we build OV portfolios. Currently they tend to try and capitalize on one aspect of the market, most of them volatile markets using RTM strats. But ATM is designed to work in all market states, maximizing perf and minimizing risk in each by detecting the market state and applying the right strats (long term trend vs RTM, etc), L+S mix (all L, 50-50, all S, whatever) and trade selection weighting factors. So in your example a STANDARD OV portfolio would NOT be the best portfolio to apply ATM. ATM could improve it, sure, but that improvement would most likely involve taking far less trades during a very bullish, non-volatile market (like now) instead of switching from the RTMs and going with a WT long term trending strat instead. The portfolio would have to have many, many strats of all types, dynamic lists designed for tradeability (lots of symbols) vs volatility (my guess here) and defined market states, weighting factors for trade selection, strat selection and L+S mixes. Luckily we can OPT these factors to let OT help build all of those but it is a new paradigm vs current OV. *The new ATM portfolio might not perform all that well w/o ATM* but is designed from the get go to use ATM anyway and only those who bought ATM should use that portfolio. Oh, and if ARM5 has some nice improvements making our strats even better ATM in OV with ARM5 strats - well, oh boy! Seminar this week on ARM5 and ATM... What do you think - will we need to re-think how we build an OV portfolio for ATM vs non-ATM? [Edited by Buffalo Bill on 2/4/2018 4:35 PM] | ||
| Jim Dean Elite Posts: 1059 Joined: 10/11/2012 Location: L'ville, GA User Profile |
I’m hoping that I quoted Ed correctly - I can’t recall if those comments were recorded or not. My guess is that they have “rough draft” plans which will be gradually refined. I do know that the ET interface took a couple of years longer than was originally anticipated (for several reasons) - so it’s possible that they are trying to avoid a potentially messy upgrade to the engine. Again, just guesses. As to your main point and questions - I think that “ATM-ing” OV will essentially be adding another layer to the top of it, rather than rearranging its current architecture in some big way. Since OV already has equity analysis thoroughly integrated (moreso than OT did), that might speed it up. The main thing (my guess) they need to do is add an “ATS” engine to gather information to “rank” things. Re current portfolios being usable - that’s likely a maybe / maybe not thing, depending on how those portfolios were constructed. I think that OV likely will do it’s “ranking” on portfolios rather than strategies (aka OV Systems), since the existing pro tools already exist To build portfolios from strategies. So - my guess is that *if* you’ve been working away at constructing portfolios that are intended for different Market States, then they won’t need much tweaking if any. The OV “ATS” capability will simply enhance how Signals within a given portfolio (the one for whatever the active Market State is) are prioritized. Also, since ATM includes segmentation of trade size allocations by market state, and since that would overlap with the OV “general settings” rules, I would expect that part of OV to be the main thing needing revamping - so that different settings can track with different portfolios, as Market states change. The most important missing ingredient in the OT implementation of ATM is that different lists cannot be used for different Market States, and that (from what I’ve been told) no one has as yet tested OT Dynamic Scan lists with OT ATM. I very much hope that missing piece will be integrated, first in OT so we can do appropriate devel work, and later in OV when the likely changes are made to provide different account setting to different Market States. That is - ability for user to specify different Focus Lists (Dynamic or Static) for each discretion Market State. To be clear - I strongly believe that Dynamic Scans offer a huge benefit which has only been partially realized thus far. Equally as important as ATS ranking. You may recall my Bash presentation 3-4 years ago about this - the analogy of fly-fishing vs a fish-farm. So, I have tried to strongly recommend that Dynamic Scans be made an integral part, in both OT and OV, of the Market-State segmentation. Lots of guesswork here. But maybe it will give folks some useful ideas. [Edited by Jim Dean on 2/4/2018 5:12 PM] |
|
|
|
|
Administrator
Forum Moderator |
Registered User
Unregistered User |
 E-Mail this thread to a friend(un)/Freeze thread | |
Toggle e-mail notification |
