Search | User Listing Vidsite | Calendars | Quotes
Home Page ->  Markets & Methods -> General Methodology Discussions -> 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. 

Luke 12:33 (NKJV) ... Sell what you have and give alms; provide yourselves money bags which do not grow old, a treasure in the heavens that does not fail, where no thief approaches nor moth destroys.


Finding System-Suited Stocks
Jump to page : 1
Now viewing page 1 [50 msgs/pg]
Jump to room :
RoyWild
Posted 9/18/2012 1:16 AM (#3986)
Subject: Finding System-Suited Stocks



Veteran

Posts: 210
100100
Location:
USA: AZ, Phoenix
Simple logic tells me that I should only trade stocks that have shown good results with my System. Those stocks that have shown good results in the past then should be considered as well suited for future trading.

This raises the question of how to define good results so we can consider a stock to be "well suited"?

OT offers couple different methods but for simplicity, we might define them as anything with a BackTest Hit Rate above 60 and APR above 25%.

Then the only need would be to run a todo list that contains enough data for the stats to be meaningful. Elsewhere, Jim defined a long-term BT range since 1996, and a mid-term range from 2003, each of which covers a full spectrum of market types.

That way I can use a list of stocks that historically have done well with my System and ensure that I don't waste time (and lose money) on stocks that may not be suited for that trading System.

I would like to hear your thoughts and different ideas on how to find stocks well-suited to a System.

Thanks


Thread moved by JimDean on 9/18/2012 3:04 AM from OmniTrader HowTo's > Voteline, Advisor & FL Stat's > Vehicle selection

Top of the page Bottom of the page
JimDean
Posted 9/18/2012 3:34 AM (#3987 - in reply to #3986)
Subject: Finding System-Suited Stocks



Owner/Admin

Posts: 3925
2000100050010010010010025
Location:
USA: GA, Lawrenceville
This thread is has been moved to a different Section-Room, since it is an excellent Methodology-related topic. (I changed the title so that the topic was clearer.) Click here for the post mentioned above that discusses good BackTest periods to use.

OT has this kind of "suitability-search" built right into it, in three to five "layers". The first layer is OmniScan. Although OScan can't be back tested easily, it's the fastest (CPU-time) way of finding an appropriate shortlist of stocks to consider for HRE trading. You can duplicate its tech rules in a filter block if you want to do backtesting with it, btw.

I believe the best use of OScan is to find something very similar to what you describe, but not quite as specific. It *begins* the suitability-search, by narrowing down a huge starting list (like all US or all Optionable or Russ 3k) to a medium list (< 500 symbols, maybe a lot less) that have consistently NOT exhibited price-volume "personality-behavior" which would be "counter productive" to the type of System you are running. For example, low average Volume is usually not good. Or stocks that don't "go" anywhere. Or ones that bounce around in highly erratic ways. Set up OmniScan rules to reject the stocks that have these BAD personalities.

There also will be certain P+V characteristics that are GOOD for the success of a given System. Some of these are cut and dried yes/no traits - anything that can be defined the way usually should also be made into another OScan rule, to filter out Stocks that don't meet that "good" criteria. For example you might want to find stocks whose prices are outside of the H-L range of the past N months ... it's a yes/no test, so it fits the OScan "boolean" paradigm.

The specifics of these Boolean (true/false) rules of course depend on what kind of trades your System (really your full Strategy) is designed to handle. That is a long discussion, which I spend time on in the Friendly Focus Lists video training. Hopefully the idea is clear enough here - relegate filters that are "for-sure yes/no" rules to your OScans.

Now we move to the second layer of the search …

Thread moved by JimDean on 9/18/2012 3:04 AM from OmniTrader HowTo's > Voteline, Advisor & FL Stat's > Vehicle selection

Top of the page Bottom of the page
JimDean
Posted 9/18/2012 3:59 AM (#3988 - in reply to #3987)
Subject: Finding System-Suited Stocks



Owner/Admin

Posts: 3925
2000100050010010010010025
Location:
USA: GA, Lawrenceville
Another pair of OT features which complement OScan are the "EZ-Setups" focus-list array of colored squares - and the ability to create custom FL columns from most any Indicators in the OT arsenal (including your personal OLang indicators). These two tools, unlike OScan, provide the ability to measure and SORT the qualities of a list of stocks in comparison to one another. This should be applied to stocks that pass the Y/N "Boolean" OScan rules discussed earlier.

The custom FL columns should be set up with a few (no more than a half dozen) indicators (canned or homegrown OLang) that reveal "measurements" about something which is important to your System's success. That is, you don't have a sure, fixed threshold-value to say a consistent "yes/no" for these traits - but you know they are important and you want to be able to choose stocks that have more attractive numerical scores in these areas. Those are the kinds of biases that are best suited for a sorted Custom FL process.

EZ-Setups offer a lot of nifty capabilities - starting with a colorful visual display that you can evaluate quickly. EZS's, despite their name, actually take some careful forethought and visualization to lay out effectively and efficiently (done poorly they can eat up a lot of CPU time). But the results are worth it!

The MAIN thing that EZS brings to the table which OScan and CusFL can't do, is the ability to evaluate *Alternate Time Frames*. That is, part of your "stock-suitability" filtering might involve a check of some rule or measurement in a higher (weekly vs daily) or lower (hourly vs daily) time-frame. This may be a yes/no question - which OScan can't address since it's restricted to Daily - or may be a better-worse-scale that CustFL's can't address since the FL is locked into just the current timeframe.

A detailed discussion about this is way too much for now … that's part of the Friendly Focus Lists video training - but the key distinctions in the use of these filtering tools is important to "burn in to your brain" so you can use OT most effectively.

So, we've seen two "layers" now of suitability-searching … but there's more …

Thread moved by JimDean on 9/18/2012 3:04 AM from OmniTrader HowTo's > Voteline, Advisor & FL Stat's > Vehicle selection

Top of the page Bottom of the page
JimDean
Posted 9/18/2012 4:41 AM (#3989 - in reply to #3988)
Subject: Finding System-Suited Stocks



Owner/Admin

Posts: 3925
2000100050010010010010025
Location:
USA: GA, Lawrenceville
Everything discussed so far relates to general "personalities" of the stock - rejecting bad ones and homing in on the best ones. Now we turn a corner and check out a Third layer, to determine whether each stock is "in a good mood" right now, for your System.

With people, sometimes you can't tell how they might respond to a topic unless you ask a leading-question and watch their face and body-language. You might find out that they are happy or even excited about discussing it - or you might learn that it turns them off. You ask that "leading question" to test the waters, BEFORE fully committing yourself to stating your views on that topic.

Similarly, OT provides a way to "test the waters" with each stock, to see if it's partial to your System "topic" or not. This is done with the Performance and Vote Block features of your Strategy - and utilize some of OT's powerful optimization features in the process. I'm not going to discuss Optimization in general (or in detail) here - yeah, that too calls for in-depth video training - rather, let's consider just one subset of that here - filtering out stocks based on how they've performed in the past with the System you are working with.

The performance Block uses historical results of how well each stock has done with a particular System, to FILTER OUT the ones that don't meet particular minimum standards, such as the 60% HR and/or 25% APR mentioned at the top of this thread. Just that capability alone is an important "layer" of the suitability-search process. I suggest that you set relatively low standards here - maybe the HR limit should be 40 rather than 60 - due to the limited number of "samples" that this test has access to, and since if you use a lot of historical data to get more samples, it's less timely.

The second aspect of this "good mood" part of the search is the Vote Block's Advisor Rating feature. This is poorly-understood by most OT users and therefore is often ignored - but it's a *powerful* feature if applied properly. Like the perf block, it considers the historical performance of each symbol - the advisor rating is a proprietary combo of APR and HR … just think of it as how "consistent" the "mood" of the stock is.

The unique and cool thing that the Vote Block's Advisor rating does is to evaluate the "mood" of each symbol in light of ALL the symbols in the current Focus List, to "grade" each symbol's mood "on the bell curve". For example, if you're in Chicago and the Cubs win the pennant - hey - everybody's in a good mood and you have to set your discernment-standards pretty high. But in New York, since the Yanks were humiliated by it, even a glimmer of a smile might bode very well in any conversation.

The Advisor therefore helps you choose which symbols are the "relative best" in the FL-"city" you are working … for your System. It won't help as much for other different types of Systems … for instance, a science-geek might care less about baseball results but get really excited (or depressed) about a probe that recently landed (or crashed) on Mars. A different focus list is needed for that alternate System - go back to layer #one.)

Now that this third layer is in place, and we've found stocks that have personalities which fit our System-topic, and moods that seem to be conducive to it, there is still one more suitability-search feature OT offers …

Thread moved by JimDean on 9/18/2012 3:04 AM from OmniTrader HowTo's > Voteline, Advisor & FL Stat's > Vehicle selection

Top of the page Bottom of the page
JimDean
Posted 9/18/2012 5:11 AM (#3990 - in reply to #3986)
Subject: Finding System-Suited Stocks



Owner/Admin

Posts: 3925
2000100050010010010010025
Location:
USA: GA, Lawrenceville
The prior three methods (OScan, EZ-Setups + Custom FL's, and Perf-Block + Vote-Block-Advisor filters) are applicable to virtually any strategy or profile. However, OT does not limit you to just ONE System in a given Strategy (or just ONE Strategy in a given Profile). That's where the fourth (and fifth) "suitabilty-search" layer comes into play.

Expanding on our prior analogy of personalities and moods … imagine being at a party and being able to find SEVERAL people whose personalities and moods fit well with a general topic. The resulting discussion becomes more lively and interesting, and might continue through the entire evening. This is what can happen when you combine several Systems together in a Strategy, which all are generally seeking the same "kind" of trade. Typically we might classify these systems as Trending or Breakout or Trading-Range, etc.

Once you've identified and polished a few Systems that are similar in this way, you can activate all of them in one Strategy. This means that on any given bar, the focus list is more likely to produce candidates to consider - and also - some bars might have MULTIPLE System-hits for a given Symbol. It's similar to the tendency for a large group of similarly-minded people to bring up more nuances in a discussion than if only two people are chatting together - and the excitement level rises when lots of folks agree.

OT manages this using Performance-Block Ranking in conjunction with Vote-Block lookback and consensus-gathering. The Perf-Block "grades" the "opinions" of each System for a given Symbol, and passes that info on to bias the Voting. The Vote Block decides if the collected weight of opinion is sufficient to decide in favor of trading that symbol on each bar. Also, the Vote-Block's lookback input allows conclusions drawn on prior bars (from outside info sources, in our analogy) to be "voted" along with current info.

So, OT can use multiple Systems of a similar nature (important!) to gather extra opinions and produce a more "refined" conclusion. That's a powerful capability that other Analysis tools don't have. Keep in mind that the Systems you combine here should all have a similar "topic" (trading-focus), using stocks with similar personalities (from OScan) that have a unified "good mood" for that focus (found via Custom FL's and EZ-Setups).

There is yet another layer, similar to this one, which handles really robust and involved trading methods which require a dedicated profile …

Thread moved by JimDean on 9/18/2012 3:04 AM from OmniTrader HowTo's > Voteline, Advisor & FL Stat's > Vehicle selection

Top of the page Bottom of the page
JimDean
Posted 9/18/2012 5:58 AM (#3991 - in reply to #3986)
Subject: Finding System-Suited Stocks



Owner/Admin

Posts: 3925
2000100050010010010010025
Location:
USA: GA, Lawrenceville
Sometimes we just can't leave good enough alone - there's another cool idea or nuance that appears, and we just HAVE TO fold it in ... that's how much of my work progresses - and it's definitely the way that OT has grown over the years. This fifth layer (discounting Dynamic Optimization and ARM4 artificial intelligence tools) adds that extra pizzaz.

Our analogy about conversation about a topic with group of people who enjoy the topic can be expanded once more - sometimes folks in that group decide to meet regularly for lunch, or form a club, because there just seems to be no "fixed end" to the interesting discussions that they have about that topic. It's fun and informative - and it can even grow into a group of clubs, with chapters across the country.

The layer of OT that models the "discussion club" idea utilizes multiple Strategies of a similar nature in a Profile dedicated to that nature of trading. Don't forget that the EZ-Setups and the FL column arrangements that comprise the "second layer" are locked in and stored by and universal to a particular Profile.

Why bother with this extra layer? What NEED is there for multiple Strategies using similar Systems for similar trading-styles in similar markets? The analogy above holds the clue - the expansion into "clubs" occured because the people could not find an "end" to their discussions ... or because circumstances dictate different "ends" as situations change.

Systems are all about "beginnings" - getting INTO the trade. OT has always majored on this "trade prospecting" theme. But experienced traders know that good refined EXIT rules are even more valuable than good entries. And just as there can be many Systems which address a particular trading-style / market-condition, so also there are many good exit-method candidates to consider.

You can utilize more than one exit-algorithm in a given Strategy, but usually when you include more than just two or three, they "collide" and result in worse performance, since any one of them can end the trade. Sort of like having too many highly-opinionated and vociferous people in a discussion group. It gets too intense and argumentative, which ends the discussion too early and makes people want to avoid the group in the future.

The solution to dealing with a variety of "endings" - ie alternative Exit methods - is to create a family of Strategies that are very similar or even identical UP TO the Trade Plan or Orders Block, but which implement different Trade Management and Exit approaches. Thus the difference in the performance of those Strategies will be measures of the effectiveness of the different Exit methods.

OT manages this through the profile's Strategy Voting settings. You activate multiple Strategies (similar trading style, mind you) in a profile, and use the Ranking and Voting rules in the All Strategy Voting panel (Ctrl-S > Strategies > Strategy Voting button) to make the decision as to which Strategy should CONTROL the trade, when more than one fires on a given Symbol on the same bar. Refer to discussions above about Performance filtering and ranking, and the related voting process - it's very similar for Strategy selection versus System selection (but no "Advisor" filter option). The idea is to wisely select the EXIT (or trade management) rules that best fit your goals, and work well with that Symbol.

This thinking is further refined in Portfolio Simulator, which deals with the problem of "real money" equity limitations. Since PS is currently going thru a big overhaul and expansion, I'll save discussions about that for later.


Thread moved by JimDean on 9/18/2012 3:04 AM from OmniTrader HowTo's > Voteline, Advisor & FL Stat's > Vehicle selection

Top of the page Bottom of the page
JimDean
Posted 9/18/2012 6:04 AM (#3992 - in reply to #3986)
Subject: Finding System-Suited Stocks



Owner/Admin

Posts: 3925
2000100050010010010010025
Location:
USA: GA, Lawrenceville
This was a very good question to raise - I hope the general discussion in the five long prior posts (3 hours of writing!) has provided a lot of food for thot - and a basis for productive effort - and that the analogy helps y'all understand how all the pieces work together in a really nifty fashion.

I'll expand on all of these areas via video training, since each has nuances and techniques worth talking about. ANY ONE of the "layers" discussed above provides an answer to the original question, in part. As a whole, they do a superb job.

This kind of thing is why I bought OT in the first place, many years ago - and Nirvana has continued to improve it along these lines, ever since.

Thread moved by JimDean on 9/18/2012 3:04 AM from OmniTrader HowTo's > Voteline, Advisor & FL Stat's > Vehicle selection

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.