Search | User Listing Vidsite | Calendars | Quotes
Home Page ->  Optimizing OmniVest -> Populating Portfolios -> View ThreadLogon (or Register, or Join TradeTight)

You are logged in as a limited-access Guest.To join TradeTight, first read the info in the Organization & Content room, then click the link above. 

John 6:27 (NIV) ... Do not work for food that spoils, but for food that endures to eternal life, which the Son of Man will give you. On him God the Father has placed his seal of approval.


Multi-Factor Portfolio Optimization
Jump to page : 1
Now viewing page 1 [50 msgs/pg]
Jump to room :
JimDean
Posted 3/28/2013 7:57 AM (#4445)
Subject: Multi-Factor Portfolio Optimization



Owner/Admin

Posts: 3835
2000100050010010010025
Location:
USA: GA, Lawrenceville
This was posted on the N forum (click here) also but I want to make sure it does not get buried:

Mark Hostius' analyses (click here) use CAR/MDD for $10k vs $100k accounts, with progressively-increasing numbers of strats in the portfolio. This approach has been THE most useful tool that I've seen thus far regarding the benefits offered by OmniVest. I truly hope that Nirvana will integrate the ability to do these comparisons automatically - formalizing the approach he has initiated.

In addition to this, several of us have proposed the need for and outlined an approach to enhanced comparative analysis for different market conditions - in a separate post (click here) I suggested six specific date ranges that represent five individual market types (stable bull, stable bear, volatile bull, volatile bear, and extended flat/intermediate/ranging) plus one range that includes the full spectrum. If statistics were run just ONCE for those periods for each Strat and two account sizes, they would not need to be run again, since the periods are static. Thus there would be NO additional demand on the servers. The statistics and equity curves from those runs would be a valuable addition to OmniVest's toolkit, allowing users to select strategies based upon how well they perform in particular market types - either to find a portfolio-basket which does "pretty well" no matter what the conditions are - OR (my preference) to find several alternative portfolios, each of which is optimal for a given market condition. This multi-portfolio approach would allow users to "trade their beliefs" regarding the current state of the market simply by switching focus or balance between pre-determined portfolios.

Also, it seems to me that there is another factor that should somehow be "folded in" - the Percent-Invested (PI). If you examine Mark's comparative snapshots with an eye to the variance in PI, it sheds additional light on the results. That is - it seems logical that there are TWO underlying and interactive factors that increase OmniVest's "mojo" - 1) the number of well-selected participating strategies, and 2) the degree and consistency of that combination for producing high PI levels. If this is so, it would be helpful is the PI could be factored in to the CAR/MDD "score".

CAR/MDD is better when it's a high number. But that ratio does not reflect how much "inertia" is behind the performance. That is, for a given CAR/MDD, if PI is low, neither the actual profit nor the efficiency of money-use committed to the account is not as likely to be as good as a portfolio with the same CAR/MDD and a significantly higher PI.

I've not yet really fiddled with alternative formulae for this metric - Mark might want to since he's got all the stats in useful form. Two possibilities come to mind:
1) CDP = CAR/MDD * PI… this would incorporate PI in a linear fashion
2) CDP = CAR/MDD * (PI^2) … this applies PI geometrically

Horse-sense tells me that the second approach, maybe with a different exponent, would be better, since higher PI's result in greatly improved COMPOUNDING, which is a geometric process. With adequate (exhaustive) study, comparing CDP/year to actual returns over the tested period normalized to a per year basis, would home in on an appropriate value for the exponent.

Comments appreciated.

Thread moved by JimDean on 8/4/2014 9:22 AM from Nirvana Club (members only) > OmniVest & Trade Processor > Multi-Factor Portfolio Optimization

Top of the page Bottom of the page
BillLeake
Posted 3/28/2013 6:30 PM (#4450 - in reply to #4445)
Subject: Multi-Factor Portfolio Optimization



Regular

Posts: 59
2525
Location:
USA: VA, Stafford
Jim

The 64k question is always how do you know which market you are in...

If you don't know the market type, or you change your strat set to the new market type "late", that approach loses some of its value and could actually be counter productive

A definitive way of determining market type, that isn't too late in it's call, would be required.

Thread moved by JimDean on 8/4/2014 9:22 AM from Nirvana Club (members only) > OmniVest & Trade Processor > Multi-Factor Portfolio Optimization

Top of the page Bottom of the page
JimDean
Posted 3/28/2013 7:12 PM (#4451 - in reply to #4450)
Subject: Multi-Factor Portfolio Optimization



Owner/Admin

Posts: 3835
2000100050010010010025
Location:
USA: GA, Lawrenceville
I agree - it's tricky to do a really good job of market-state prediction. BUT, it's easy to do a "so-so pretty good" job of it ... lots of rules and tools exist for that.

Re the "really good" approach ... I have such a beastie ... it's quite tricky to create. One of these days I will release it. I've offered repeatedly to GIVE it to Nirvana, if they would implement it without fiddling with it and messing it up. No response.

Most people however do it by eyeball+gut. Almost everyone has an opinion about it.

The point is ... if you DO have a strong trading belief and bias regarding market type and how to trade it, then the stat's that I've suggested will be very helpful. Many people fall into that category. For the people that don't agree, they can choose to ignore that information.

The idea here is to try to "make up for" the BLACK BOX nature of OVest's "canned" strategies. By seeing how they perform in these targeted time periods, we can better understand where their strengths and weaknesses are.

Thread moved by JimDean on 8/4/2014 9:22 AM from Nirvana Club (members only) > OmniVest & Trade Processor > Multi-Factor Portfolio Optimization

Top of the page Bottom of the page
Jump to page : 1
Now viewing page 1 [50 msgs/pg]
( E-mail a Link | Printer Version | Search Room )

Owner of site: Jim Dean -- Forum content is confidential, and may not be distributed without written permission.