consistency of entry method AND symbol

How do you know when a trend has started? Ended? This forum is for discussions about trend indicators and signals.
Post Reply
zoopy12
Contributing Member
Contributing Member
Posts: 5
Joined: Mon Apr 05, 2010 7:50 pm

consistency of entry method AND symbol

Post by zoopy12 » Thu Jan 12, 2012 7:12 pm

Hello fellow Bloxsters, I have a question and I would appreciate some input on it from others. My question is this: Let's say you have a mechanical entry signal method and you use that method on a pool of several, let's say 20 instruments. You have a constraint however that allows you to only have a trade on in, at most, 2 of the instruments at any one time.
You monitor 20 only because you want more signal(trade) opportunities than you would have if you monitored 5 instruments. My question is: Does this type of trading framework lead to taking trades inconsistently, because of necessity there will arise situations where you take losing trades in instrument A and B, you are out of those, then you enter trades in instruments C and D but while you are still in C and D either A or B or both A and B signal you again, and now you can't take those next trades in A and B because you are "full" with C and D, thus you are trading A and B inconsistently. At least that's what it looks like.
Next, even if this were true does it really matter over series of many many trades in the entire "pool" of instruments. I guess it's akin to asking the question in a different way. If you have 1,000 ES daytrading entry
signals and you took every other one, instead of taking all 1,000 signals would it make difference to your profitability or rather expectancy ??
I ask this question because of something I read in the real Turtles rule book, that said part of the system insisted that they take every signal in a given symbol if they decided to trade that symbol at all. Any help on this ??

Moto moto
Roundtable Knight
Roundtable Knight
Posts: 427
Joined: Mon Jun 01, 2009 4:12 am
Location: once again in the UK

Post by Moto moto » Fri Jan 13, 2012 3:37 am

"Does this type of trading framework lead to taking trades inconsistently"

Not if you test for it, figure that the past does in fact represent a fairly accurate picture of what might occur in the future and then decide to systematically apply the testing.
If you use discretion then probably yes.
Many testers take portfolio management to the next level and test for questions such as you ask - "what happens if I test this idea rather than that"
and in a nutshell if i could be so bold to summarise what at least I read out of a lot of other results - diversification is important - so you are better to do so.
.....and then ultimately a lot depends on how many resources you have to be able to efficiently apply such a process both in terms of money, time and effort and of course desire.
Ask yourself....what is the process for determining how, why and when to reject trades.

babelproofreader
Roundtable Knight
Roundtable Knight
Posts: 138
Joined: Wed Nov 10, 2004 4:36 pm

Post by babelproofreader » Fri Jan 13, 2012 9:37 am

By trading in the way you describe you are effectively randomly trading your system as it is applied to any single instrument, which of course will drastically alter its results in an unpredictable way. This very issue was the subject of the NAAIM 2011 Wagner Award winning paper, downloadable from the link.

Paul King
Roundtable Knight
Roundtable Knight
Posts: 207
Joined: Mon Feb 23, 2004 9:13 am
Location: Vermont, USA
Contact:

Post by Paul King » Fri Jan 13, 2012 1:31 pm

As far as I can tell, what that article is saying (in 43 pages no less, but hey, it is rather a large font) is that if your historical testing ignores significant real-world implementation constraints (margin rules in this case) then actual trading results may differ considerably from the simulated ones due to the fact that what you have tested looks dissimilar to what your are actually implementing.

My advice; wherever possible don't ignore quantifiable significant real-world implementation constraints in your historical system testing if you want representative results.

Paul

Greyzy
Contributor
Contributor
Posts: 2
Joined: Thu May 26, 2011 3:32 am
Location: Germany

Post by Greyzy » Fri Jan 13, 2012 5:49 pm

babelproofreader wrote:By trading in the way you describe you are effectively randomly trading your system as it is applied to any single instrument, which of course will drastically alter its results in an unpredictable way. This very issue was the subject of the NAAIM 2011 Wagner Award winning paper, downloadable from the link.
Unfortunately NAAIM didn't put my EXCEL tools on their website. But you can download it here http://www.iitm.com/Weekly_update/Weekl ... 3_2011.htm or send me a PM and I will email it to you.

Feel free to ask for help with the tools or discuss your ideas.

Greyzy

zoopy12
Contributing Member
Contributing Member
Posts: 5
Joined: Mon Apr 05, 2010 7:50 pm

Post by zoopy12 » Sat Jan 14, 2012 9:30 pm

Hello, First of all let me give a big thanks for all the comments and insights. I try hard to find the potential flaws in my trading approach. I only trade my own funds. The whole reason for my question arises because traders are always facing constraints. You maximize component X of your total method because you think it will create this effect and then of course it has it's chain of "side effects" usually unwanted, that you must then cope with.
So much of the way the markets and therefore trading venues/platforms are structured can easily lead a person to never even consider this "problem" that we are talking about. Indeed, the notion that it isn't necessary to wait for your particular symbols to meet your entry signal definition because software can instantly find you as many symbols as you would like that do meet your criteria right now contributes to overlooking the problem completely. It creates a virtually unlimited symbol pool instead of the trader defining the size of the symbol pool.
The direct solution of course is to always have a totally matched dedicated capital and risk allocation to trade each symbol in one's universe. Then each and every signal can be taken without fail, all wins all losses. Thanks Z

Greyzy
Contributor
Contributor
Posts: 2
Joined: Thu May 26, 2011 3:32 am
Location: Germany

Post by Greyzy » Sun Jan 15, 2012 10:10 am

zoopy12 wrote:The direct solution of course is to always have a totally matched dedicated capital and risk allocation to trade each symbol in one's universe. Then each and every signal can be taken without fail, all wins all losses.
There is actually some middle ground where you intentionally accept the risk of skipping trades, but on the other hand have the advantage of trading bigger position sizes or a wider universe of symbols (or both).

This depends on the characteristics of your system.

In simple terms the question is how much your system depends on a few big winners, because then you need to reduce the risk of skipping a trade (it might be that rare big winner). On the other hand if you have a high win-rate and typically get out at somewhere between 1x to 2x your initial risk (low standard deviation of your individual trade results) then it might actually be worth risking to skip a winner every once in a while, but to be able to have a higher trading frequency (due to more symbols in your universe) or to trade bigger position sizes (more capital that's deployed and thus earning profits).

Greyzy

Robwynge
Senior Member
Senior Member
Posts: 42
Joined: Wed May 25, 2011 10:33 pm

Post by Robwynge » Sun Jan 15, 2012 4:59 pm

I read a study once, which I'll link later if I can find it, that said skipping valid trades due to capital limitations (or other reasons) can kill a system with a profitable expectancy because it adds a random element to the system (the ability to take the trade) that degrades your edge. Greyzy gives a good example of what can go wrong.

Paul King
Roundtable Knight
Roundtable Knight
Posts: 207
Joined: Mon Feb 23, 2004 9:13 am
Location: Vermont, USA
Contact:

Post by Paul King » Sun Jan 15, 2012 6:09 pm

Something to ponder:

In roulette, the betting strategy of playing "red" (or "black") every time and doubling bet size after a loss is a positive expectation "system". However, this "system" also requires infinite capital, and a casino with no table limits - these are unrealistic implementation requirements.

Saying that this "positive expectation system" has been "killed" by being forced to "skip" trades isn't really a useful analysis of the real-world situation and is a waste of time IMO.

Paul

p.s. I know I'm probably opening a can of worms here because there are obviously some strong/ingrained beliefs at work regarding the effect of "randomly" removing some percentage of a trade sample, and how that changes expectation.

zoopy12
Contributing Member
Contributing Member
Posts: 5
Joined: Mon Apr 05, 2010 7:50 pm

Post by zoopy12 » Sun Jan 15, 2012 7:18 pm

I submitted this question originally because the biggest test I ever ask myself in trading/speculative methods is what parts matter long term and what is arbitrary. Great points have been made here so far. At this point I am kind of in the camp of thinking that I am not sure what the truth is on this matter, but if it is going to rough me up emotionally knowing that I bypassed winners after taking some losers and the only way to avoid that discomfort is to structure my trading so that I never miss any valid signal I am willing to do it, sort of like wearing my lucky green derby when I go to the casino. But again I am not sure that it truly does matter which way one approaches the issue, at least from a math standpoint long term. I once worked part time in a plastics factory and the quality control people would come by to check the parts that had just been made. Most pallets of finished product would have about 5,000 to 15,000 parts on them. When I saw QC person randomly pull just 2 boxes out of a hundred or so, checking them, replacing them, and then filling out their report I was shocked. Then I asked.... sampling

Robwynge
Senior Member
Senior Member
Posts: 42
Joined: Wed May 25, 2011 10:33 pm

Post by Robwynge » Sun Jan 15, 2012 9:18 pm

Paul King wrote:
Saying that this "positive expectation system" has been "killed" by being forced to "skip" trades isn't really a useful analysis of the real-world situation and is a waste of time IMO.
Just providing information. People are free to value it as they shall.

The paper is attached. Turns out it was the 2011 NAAIM Wagner Award winner "for advances in active investment management." Whether that means much to anyone, I don't know.
Attachments
G2011_Buying Power.pdf
(1.04 MiB) Downloaded 625 times

Post Reply