Profitable Strategies being impacted by the #'s using them.

[darkhorse]

Originally posted by murray t turtle
I would want to know the time of the day + time of the month and such.

IMHO I think that because the market is a chaotic system it defeats most (i changed this from 'all' don't want to sound dogmatic) attempts to profit from overly precise backtesting.

Tradestation is almost a joke in the way people abuse it by finding false correlations that only exist in the past. Say that over the past eight years the market has been up on Tuesday more than Wednesday 47% of the time (this is just a hypothetical example). Who cares?

A chaotic system- that is, a system with defined rules but too many variables to reliably predict outcome- will always move in similar looking patterns but never produce exactly similar results. So being overprecise in your paramaters is sort of like picking patterns from a random number generator in the false belief that they will still be there tomorrow instead of just in hindsight.

This is why I've always been skeptical of backtesting. My method was built painstakingly on general principles of price behavior and observations regarding the overall structure of the market as a self reflexive system. Every rule I have is based on a foundational, logical and permanent observation rather than a possibly temporary correlation with what went on last week or last month or last year.

Not that this is the only way to go, I'm sure there are many profitable systems built on short term correlations- but they all have limited lifespans- some of them too limited to be of much use because you never know when they will go bad.

As usual just my .02, not gospel truth

 

[limitdown]

let me weigh in with my $.02, but you might want to ask for change.

We, as traders, scalpers and professionals (market makers, institutionals and the like) are in want of heart. We essentially have nothing to go upon, and are jumping on what ever train even acts like its ready to leave the station.

we are in want of heart, and the markets and the larger market, namely the economy, just doesn't have any heart.

there was another commentary within a thread regarding why the TRIN (Richard Arms Index) doesn't work, and someone wisely posted a website that went into even more detailed fundamental data to support his conclusions. Namely that the Fed having lowered interest rates over 11 times, and other essential factors, not to mention the War that's on, and the substantial shock to the senses that the WTC incident has had upon the economy. Add to these the Enron'ization of America, and the A. Anderson collapse, then what we as a country have touted for ages, namely that we:

1) fully disclose
2) America is the model for (democratic) capitalism
3) other countries should enforce proper accounting and disclosure of their corporate financials

and you have the malaise that we're in.

they say that time heals,
well here's to understanding the bigger picture, before looking into the technical indicators to get another opinion.

 

[Ryan Alder]

Darkhorse et al,

I agree & disagree regarding your stance on backtesting -- I have a plethora of issues with tradestation; I think that Globalserver is unfriendly as an interface, I absolutely despise the fact that you can't backtest baskets (only one instrument at a time), and I think that output/input features in easy language are redementary. Besides that I haven't been able to keep it stable when employing scale (real-time; 10 MKTs) except for on a Dual Xeon machine. I also don't like that its not built on SMT technology.
However, for the purpose of testing your hypothesis in a quick fashion; not the extreme of using 20 variables to optimize historical return... its a fair platform. I personally use Matlab, SPSS & Maple & Mathcad to perform most optimizations.
Based on your "theory of reflexivity" view of the markets... backtesting has its place. For instance to optimize for an variable (i.e. time) to employ capital in different markets.. I found that cyclical relative strength (normalized & ranked) is optimal for identifying trends in markets. For Soros it would of been REITS -> Conglomerates -> semiconductors; if my memory serves me correctly.
I use tradestation as a labratory to test my hypothesis and Optimize variables -- based on the results I attempt to understand why x variable was optimal.

 

[darkhorse]

Just to clarify, I have never used tradestation and never will.

I only referred to it because it is such a popular interface for backtesting, and its capabilities in that area are pasted all over their advertising.

As for the intrinsic value of backtesting itself, I know my opinion is very much in the minority when I say it's way overrated. And I'm sure it has some value for other more quantitative oriented strategies rather than simple, uncomplicated approaches such as mine. So I don't find it all surprising for you to disagree w/ me, many others do also.

Test on

 

[metooxx]

Originally posted by darkhorse


As for the intrinsic value of backtesting itself, I know my opinion is very much in the minority when I say it's way overrated. And I'm sure it has some value for other more quantitative oriented strategies rather than simple, uncomplicated approaches such as mine.

I agree to a large extent; the real test is with money ...

 

[DarynC]

I would agree with the theory that a strategy becomes either more difficult to implement or dissipates over time as more people start using the strategy.

I trade pairs and would have to say that although I don't think the strategy will ever become obsolete (because the relationship between the pair is so elastic), I do notice that it can become much more difficult to put on good spread prices. If you have two stocks trading 1 million shares per day each and they work well as a pair, the specialist won't notice a few people trying to put on positions but when you add 20-100 people all trying to put on 200-1000 share positions at certain spread prices, you can't tell me the specialist doesn't notice the added pressure. In the end I think that he who works the hardest and moves the fastest will achieve the best results.

Daryn

 

[nitro]

...

I know highly profitable systems that are right less than 50% of the time. Yes your read that right.!! They involve catching longer term trends. I tend to only be right about 40% of the time when I find these but when I am right my winners are 30 times what I risked while my losers are the amount I risked or maybe 1.5 times (giving it some slippage or a gap past my stop)

There is nothing contradictory or surprising in this. _Frequency_ of wins has "nothing" (except the psychological) to do with trading results, _EXPENTANCY_ does.

The longer term strategies are harder to water down from what I've seen especially with having to follow a system that you must be wrong more than right. This is extremely hard for the majority to comprehend.

Computerized system testers that have tested hundred of systems and traded the system(s) know this - though I agree, when one has their training wheels still on, it is very difficult to let go of the "winning percentage" pacifier.

Now on the opposite end of the spectrum I know quite a few strategies that are very short term and have winning trades that seem to be more than 60% of the time (some even 80%). The problem with these is they are very short term. The profits are limited and if too many are doing them than when I'm trying to stop out for .10 I end up losing .35 due to too many others using the same stop.

My experience as well trading on high frequency data.


nitro

 

[Rigel]

The best and kindest thing a small independent trader can do for his beloved hardworking fellow traders is to NOT share his successful strategies. Successful strategies come from hard work and diligence which produces "luck". If a trader finds something that works well he can be sure that there are several others already benefiting from it. As soon as the cat is let out of the bag all the freeloaders will come in like a swarm of locusts and kill, cook, and eat the goose that laid the golden eggs. The scavengers then come in, pick over the bones, and write a book about it.
I suggest a motto. IF YOU CARE, DON'T SHARE.

 

[rfoulk]

> Depending on the strategy being shared, you're right up to a point.
> But because trading is a zero-sum game - well actually a negative sum
> game, once you factor in commissions - there are simply always going
> to be losers. Not everyone can win - unlike in your driving analogy,
> where we can ALL become better drivers.

The Earth is a closed system too.

It would take a mighty small market for the effects of a closed system to matter enough to be noticeable. There's probably one out there somewhere, but I can't think of it.

I'm just amazed every time someone claims trading is a zero-sum game. I guess it helps some trade better thinking of it as some mortal struggle.

 

[daniel_m]

Originally posted by rfoulk


The Earth is a closed system too.

It would take a mighty small market for the effects of a closed system to matter enough to be noticeable. There's probably one out there somewhere, but I can't think of it.

I'm just amazed every time someone claims trading is a zero-sum game. I guess it helps some trade better thinking of it as some mortal struggle.

The topic of this thread is whether sharing a strategy nullifies that strategies effectiveness. Someone then said something about if we all drive better it's better for all of us, to which I replied it's a bad analogy to trading, as we can't ALL become better traders - otherwise who's gonna take the other side of our trades.

I thought trading being a zero sum game was a pretty simple, obvious concept. If we're all wrong on this, I really would love to hear another explanation. And if you're gonna be introducing "closed systems", thermodynamics or other more complex theories, please, for me, and others, who are bit slow on the uptake, help us out with some clear reasoning.

Daniel