Which strategies work, and which don't

Quote from jcl:

Well, I don't think that Daniel Kahneman does trade. At least not since he won the Nobel prize of economics in 2002.

It might sound like a strange concept, but reading a book can sometimes be useful. Of course only when the author knows more than the reader. But for you this is certainly not a problem.
More...

Wow, a Nobel prize in economics in 2002. Very prestigious award.

This means he could charge more for his books and speaking fees.

So, $1,000/book sounds about right.
 
Quote from ocean5:

I wonder,what did he mean by 'a professional trading',the average ET or prop shop trading?
More...
He meant employed traders in trading firms. I think the brokerage accounts of private traders or prop shops are not easily available to studies.

His book is not only about trading, but traders are a prime example of "illusion of skill and validity".
 
Quote from jcl:

He meant employed traders in trading firms. I think the brokerage accounts of private traders or prop shops are not easily available to studies.

His book is not only about trading, but traders are a prime example of "illusion of skill and validity".
More...

Illusion of skill and validity. Sounds like the random walk theory.

Now it's getting interesting.

I gotta a feeling that his work is along the lines of Jack Hershey's posts. I say this with no disrespect.

His message is like saying that Michael Jordan's ability to play basketball is an illusion.

Or Magic Johnson's ability to play basketball for that matter. Maybe that's why Ervin Johnson is called Magic. He puts on a good magic show. Loved his "show"time basketball.

The author could go the philosophical route and be right. He could argue that life is an illusion.

At the end of the day, it just becomes a Jack Hershey post. Lol.
 
Before the book even begins, the introduction discusses the issues that impair the discretionary trader.

This discussion involves a belief system and this is a major reason why it's difficult for a trader to switch from discretionary trading to algorithmic / automated trading.

From MarkBrown.com in Pursuit of the Holy Grail
If one item were to be singled out as the most detrimental to the project, it would have to be that there were preconceived ideas of what a successful trading system should look like. The computer was assumed to be in error, time and time again no matter how successful the per trade profit looked. This assumption of error was perpetuated by the resistance to investigate what the results were showing. The computer modeling did not look and behave as was anticipated by the research team. Following in the footsteps of one of the early pioneers of computer testing and trading. It was realized that the project was closer to completion, than what had been previously thought.

The project completion was being held up because the team refused to allow the computer model to trade the methodology that it found most profitable. In spite of what the research team thought, one area was left not completely researched. That area was the research team themselves ability to allow a computer to do the actual trading. In other words the team needed to ensure that they were profit motivated and not motivated by the need to be traders. Thus the intervention of the pioneer mentioned earlier. He having been in the very same situation years ago when developing his computer model, was gracious enough to share the road map for the human part of the systems evolution.

The main element that was provided to the team was the experience that once you have a methodology that works, you have to let the system do its job. It must become the trader and you a servant of it. After this is accomplished trading actually becomes a rather mundane experience.
More...

I have read many times from discretionary traders that there is no way a program could give you insights to how the market works. I guess these guys have no idea what data-mining is. They also do not understand that many successful automated traders use a scientific top-down approach to developing trading systems and prove trading ideas.

Rule induced data-mining is a way of proving if a trading method is valid. This is not AI, but a method that allows a trader to see market tendencies and determine by computer simulation if the tendency is tradable. If you read a few other books, such as The Talent Code or Talent is Overrated it has been proven that practice creates talent. A computer simulator can practice years of trading in minutes that would take a trader years to complete manually. The market is changing and the new breed of day-traders will be automated.
 
Quote from keeptradin':

Funny you should ask, that "revelation" is what led me to this:

http://www.elitetrader.com/vb/showthread.php?s=&threadid=240708

Far from perfect, but certainly simple. Or "stupidly simplistic" as some might say...LOL!! :p
More...
Interesting. If I understand your system right, this is the code:

Code: 
	asset("SPX500");
	BarPeriod = 60;
	Stop = 10; 
	Profit = 10;

	if(lhour(ET) == 9) // NYSE opening hour
		enterLong();
	if(lhour(ET) == 15)  // NYSE closing hour 
		exitLong();

It has 52% winners, but is losing in the long run, with a 0.92 profit factor. With detrended prices the profit factor goes down to 0.88. But the system has long winning periods, almost 1 year, f.i. from mid-2009 to mid-2010.

Such seasonal systems were the second-most profitable from what I tested. Usually, to make such a system really profitable, you let the computer generate a filter based on some data from the previous day.
 
Quote from jcl:

Interesting. If I understand your system right, this is the code:

Code: 
	asset("SPX500");
	BarPeriod = 60;
	Stop = 10; 
	Profit = 10;

	if(lhour(ET) == 9) // NYSE opening hour
		enterLong();
	if(lhour(ET) == 15)  // NYSE closing hour 
		exitLong();

It has 52% winners, but is losing in the long run, with a 0.92 profit factor. With detrended prices the profit factor goes down to 0.88. But the system has long winning periods, almost 1 year, f.i. from mid-2009 to mid-2010.

Such seasonal systems were the second-most profitable from what I tested. Usually, to make such a system really profitable, you let the computer generate a filter based on some data from the previous day.
More...

A few other simple ideas:

Base the Profit and Stop on the volatility of the instrument. My research shows that this is the Single Most Important factor in regards to intra-day profit target / stop-loss placements. This is because the volatility is the most important change in price-action that occurs from day to day - Contractions and Expansions.

While there are several different ways to calculate intra-day volatility, all you need is ONE.

Also, the location of the OPEN is extremely helpful in determining what direction the instrument may be headed for the trading day.

Remember, not all markets possess the same tradable tendencies and Market Based Indices are MUCH different than individual stocks or Sector-Based ETFs.
 
Quote from jcl:
While working on a couple different trade strategies, I found an interesting trend. The more time I spent with implementing an algorithm, the less profitable it turned out to be. Quite frustrating. The most profitable algorithms were also the simplest.

Have you made similar experiences?
More...
TRADING TRUTH #1:
Even with a great strategy, the outcome of any single trade is uncertain (“anything can happen”).

TRADING TRUTH #2:
Losses cannot be fully eliminated (“losses are the ‘costs’ of the trading ‘business’”).

TRADING TRUTH #3, #4, #5 …. Etc:
[I know, I know …. there are many more!]

IMO, a “complex” strategy can be a reflection of the strategy developer’s subconscious denial of Truth #1 and Truth #2 above (i.e. an attempt by the strategist to achieve near “100%” certainty with each trade, or to accept only the smallest of losses, or even to eliminate losses completely).

Or alternatively, a “complex” strategy may result from a belief by the strategy developer that the best(*) trades always reveal themselves in advance, in which case strategy design is just a process of refining the trade filter until eventually all trades can be certain winners.

However, because the best trades don’t always reveal themselves in advance (Truth #1?) filtering trades too much reduces opportunities for the set up to play itself out, and can actually filter out the best trades disproportionately versus the less good/bad ones. dom993 stated it here… http://www.elitetrader.com/vb/showthread.php?s=&postid=3488552#post3488552
Quote from dom993:
... the "randomness" of the large winners ... IMO, the tendency of big winners to "prefer" marginal setups comes from the surprise factor ... when the majority is caught on the wrong side ... you never get that if you are taking the most obvious or cleanest setups...
More...
So, it is not that “simple is better than complex” per se (because additional complexity can turn a more simple system into a better system); it’s that sometimes “complexity” is pursued for the wrong reasons and reflects a bias that is detrimental to successful systematic trading. In which case, these complex systems will perform less well than simpler, “uncontaminated” ones.

= = = = = = = = = = = =
(*) BTW, I‘m using “best trade” here in the sense that it can be seen (after it’s over) that it was a big winner; I’m not using “best trade” in the sense that it looked (before the fact) like it had an excellent risk/reward ratio, and so was a risk worth taking, even though we don’t know what the eventual outcome will be.
 
Quote from abattia:

TRADING TRUTH #1:
Even with a great strategy, the outcome of any single trade is uncertain (“anything can happen”).

TRADING TRUTH #2:
Losses cannot be fully eliminated (“losses are the ‘costs’ of the trading ‘business’”).
More...

Indeed, eliminating losses seems in fact to make strategies less profitable. The most profitable strategies I found had a win rate in the 40% range.

Eliminating losses is easy, just set a tight profit target, but all tested strategies so far got worse when a profit target was used at all. Which confirms that one should let the winners run.
 
You must log in or register to reply here.
Back
Top