Search | User Listing Vidsite | Calendars | Quotes
Home Page ->  Optimizing OmniVest -> Creating Conditions -> 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. 

James 1:4 (AMP) ... But let endurance and steadfastness and patience have full play and do a thorough work, so that you may be perfectly and fully developed, lacking in nothing.


Sticky Paradigm-Shift: Equity Curve Filters
Jump to page : 1
Now viewing page 1 [50 msgs/pg]
Jump to room :
JimDean
Posted 5/10/2014 6:04 AM (#5859)
Subject: Paradigm-Shift: Equity Curve Filters



Owner/Admin

Posts: 3888
20001000500100100100252525
Location:
USA: GA, Lawrenceville
I believe that when we write OScript conditions for the equity curve in OVest, we need to "wrassle our perspective around" considerably, compared to how we think about price data series.

Price data tends to react more quickly up and down, with lots of volatility-noise, bar to bar, in comparison to an equity curve - since ofttimes several symbols might be participating in the behind-scenes variation of that equity curve for a single Strat - and if were viewing the curve of the portfolio, then there often can be multiple strats combining their action together. The net effect is (should be anyways) that of "smoothing" the equity curve out in comparison to what we are accustomed to seeing on the tens of thousands of price charts that we've viewed and concentrated closely on over the years.

It's a major paradigm shift that this far has been unheralded. Maybe think of this post as a wakeup call. If OVest ever allows the OScript components (ie the EMA's, etc) to be plotted on top of the equity curve, on precisely the same scale (maybe Cartesian not semi log when zoomed in), I think this will become much more obvious.

So, equity curve conditions require different thinking. Steve Mayo recently had a cool idea that he asked me to code up, and gave me permission to share - a condition designed specifically for measuring cumulative +Rtn vs -Rtn of equity - click here to review it. Also, I created some OScripts for CAR, MDD and Calmar intended for equity curve eval (click here for those) … the "paradigm shift" is that they examine the Calmar, etc on a bar-by-bar basis, rather than just at the end manually for the entire chart.

It may be that those OScripts don't work out well (no time yet to test them) - but they well-illustrate the need for "thinking about" equity curve conditions from a different perspective than we would when designing conditions for a Symbol.

Thread moved by JimDean on 8/4/2014 8:45 AM from Markets & Methods > OmniVest, Money Mgmt & Risk Control > Paradigm-Shift: Equity Curve Filters

Top of the page Bottom of the page
JimDean
Posted 5/10/2014 6:25 AM (#5860 - in reply to #5859)
Subject: Paradigm-Shift: Equity Curve Filters



Owner/Admin

Posts: 3888
20001000500100100100252525
Location:
USA: GA, Lawrenceville
As I mentioned earlier, part of the OVest-revolution equity-curve paradigm shift needs to consider that portfolio and even individual strategy equity curves will usually be "smoother" than the price curves we are used to looking and, thinking about, and designing solutions for.

We are used to applying some kinds of indicators to price which are purposefully designed to "smooth" the response. EMA's are a good example - they are infinite series that tend to "dampen" sharp pops and drops in price data - helps prevent whipsaws when used as part of strategy logic. Wilder RSI (not Nirvana's) also uses very high attenuation (smoothing) - as does Wilder ATR and Wilder ADX. My gut tells me that excessive smoothing like that is NOT appropriate for equity curve analysis.

For example, a recently-posted (click here) equity condition looking at Ema7 vs Ema21 may well be inadequately reactive to prevent bad short trades (or to allow good ones) … since the market turns down pretty fast, and dead cats bounce quickly … posssibly ZMA (zero-lag exponential) would work better. This isn't a wild guess. I did some earlier tests with C vs EMA on equity compared to C vs ZMA, and ZMA definitely was better. That testing was for Long Only - I suspect the distinction will be more dramatic for shorts. Sorry about guesswork - just no time to do a bunch of tests right now.

We need to remember that the Equity curve has a LOT of lag in it - each Strat has it's own lag - and switching between them introduces more, and once we get portfolio switching, there will be even more! Even though I'm not an EE, my guess is that some of Ehlers work (DCMA? SSF?) might come in really handy for equity curve analysis. Nirvana's SMA-based RSI, used with short periods, could also be beneficial. Don't forget the old standby Stochastic, with a tight lookback window and little or no smoothing (like maybe 10,1,3).

New tools call for new thinking!


Thread moved by JimDean on 8/4/2014 8:45 AM from Markets & Methods > OmniVest, Money Mgmt & Risk Control > Paradigm-Shift: Equity Curve Filters

Top of the page Bottom of the page
JimDean
Posted 5/10/2014 8:03 AM (#5861 - in reply to #5860)
Subject: Paradigm-Shift: Equity Curve Filters



Owner/Admin

Posts: 3888
20001000500100100100252525
Location:
USA: GA, Lawrenceville
Mel made a very good suggestion about this:
This is an application where lag is very important. I recommend using the Nirvana OptiSmooth filter that is now in the ACT 2 plugin. It is the closest thing to the SSF, and by varying the period, they look very similar. A 5.5 period OptiSmooth is almost identical to s 10 period SSF, and it should be directly accessible to Omniscript.
I doubt we will be using the much longer periods the DeCycler works better on.


I have yet to dig into the SSF re OLang fiddling around, but from what I've seen of it, that seems to be a great substitute for short term MA's.

I'll check on whether the SSF can be modelled via OScript, to provide it for a direct use in OVest.

Thread moved by JimDean on 8/4/2014 8:45 AM from Markets & Methods > OmniVest, Money Mgmt & Risk Control > Paradigm-Shift: Equity Curve Filters

Top of the page Bottom of the page
RossKovacs
Posted 5/10/2014 9:38 AM (#5862 - in reply to #5861)
Subject: Paradigm-Shift: Equity Curve Filters



Veteran

Posts: 123
100
Location:
USA: GA, Snellville
I am not an OmniVest user, but I have compared Nirvana's OptiSmooth filter with the SSF programmed in OmniLanguage.
N's OptiSmooth filter apparently does not create initial values for the filter which creates weird starting values for the OptiSmooth filter.
If the Equity Curve Filter is sensitive to the starting values, then I suggest that you need your own code for the SSF.
For example, I use a 3 bar SMA for initial values (where Periods are the length of the SSF) :

If Bar < Periods Then ' Use 3 bar SMA for initial values of Super Smoother
SS2 = (Pr + Pr[1] + Pr[2])/3 ' Super Smoother code after Else
Else

Thread moved by JimDean on 8/4/2014 8:45 AM from Markets & Methods > OmniVest, Money Mgmt & Risk Control > Paradigm-Shift: Equity Curve Filters

Top of the page Bottom of the page
JimDean
Posted 5/10/2014 10:26 AM (#5863 - in reply to #5862)
Subject: Paradigm-Shift: Equity Curve Filters



Owner/Admin

Posts: 3888
20001000500100100100252525
Location:
USA: GA, Lawrenceville
Thanks Ross. There's been extensive discussion on the OVest forum re built-in "warmup" that occurs for Strategies. It's unclear thus far whether that warmup also applies to OScript Conditions that the user creates, and how long the warmup period lasts in that case. This is another good example of why that might be important. My guess is that Nirvana will solve this somehow so the user doesn't need to fret about it ... for example, a 200-period MA of any type definitely needs warmup, if used in a Condition. My guess is that the SSF values sort themselves out, without a "seed" such as you've described, after some reasonable mult of the periods.

Thread moved by JimDean on 8/4/2014 8:45 AM from Markets & Methods > OmniVest, Money Mgmt & Risk Control > Paradigm-Shift: Equity Curve Filters

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.