Comparing all TA methods statistically

[Gringotts Bank]

Please see my post in the ASF feedback thread.

I wanted to know if traders who are very familiar with one or more techniques would be interested/willing to take part in a "trade off", one method against others.

Joe is ok with it. If there are people who want to take part, please indicate here to get an idea of numbers. We might need an excel person to keep track of trades and tabulate data. Gains and losses would be compared with 'method' chosen. The best method would be one where most % gains have been made across all traders and all trades. It's not a competition where a person wins, but a method.

Hopefully this is of interest to everyone. It's the best way I could think to compare TA methods against each other.

How it would work:

from the list of 10 stocks, analyze each, find one you like using a method you like. If nothing appeals, make no trades. Submit your trade to the ASF thread. Buy and sell as you wish within the set time frame.

Obviously there are times when a certain stock may appeal on a number of different criteria. eg. a 'breakout' stock may also be a 'momentum' stock may also be a 'candle pattern' set up. In this case you can nominate more than one method for each entry and exit you make.

Anyone?

 

[tech/a]

Please see my post in the ASF feedback thread.

I wanted to know if traders who are very familiar with one or more techniques would be interested/willing to take part in a "trade off", one method against others.

Joe is ok with it. If there are people who want to take part, please indicate here to get an idea of numbers. We might need an excel person to keep track of trades and tabulate data. Gains and losses would be compared with 'method' chosen. The best method would be one where most % gains have been made across all traders and all trades. It's not a competition where a person wins, but a method.

Hopefully this is of interest to everyone. It's the best way I could think to compare TA methods against each other.

How it would work:

from the list of 10 stocks, analyze each, find one you like using a method you like. If nothing appeals, make no trades. Submit your trade to the ASF thread. Buy and sell as you wish within the set time frame.

Obviously there are times when a certain stock may appeal on a number of different criteria. eg. a 'breakout' stock may also be a 'momentum' stock may also be a 'candle pattern' set up. In this case you can nominate more than one method for each entry and exit you make.

Anyone?

Well
Youd never follow my discretionary trading.
Id be for ever explaining why I did this or that.
Its a combination of 18 yrs experience not just VSA or Chart Patterns.

I think youll find with most of the experienced guys here theyd be the same.

Systems trading has a set rule base so when Im trading that again thats easy.

We dont care whats "Best" and my best wont be yours if you attempt to emulate it.
You wont be able to.

What you need to do is learn HOW to trade.

 

[Gringotts Bank]

This is not about comparing systems. Systems are easily compared. What's not easy to compare is TA methods that can't be sytematized, eg. trendline breaks, pattern set ups, elliot wave analysis, vsa etc etc.

I really need to know:

1. Can any of these non-systematizable methods stand on its own?
2. What can be thrown out all together?
3. What methods involve a high degree of subjectivity/user experience?

Don't you think that's useful info?

The 'how' is a different ball game, and I agree that's where the money is made.

 

[Trembling Hand]

I really need to know:

1. Can any of these non-systematizable methods stand on its own?
2. What can be thrown out all together?
3. What methods involve a high degree of subjectivity/user experience?

Don't you think that's useful info?

I got a better idea. Why don't we pick 10 different athletic events and do the same.

I really need to know which one I can make a living out of.

 

[Gringotts Bank]

You're saying a good depth trader (for example) will not necessarily be a good elliot wave trader, which is a fair comment.

But allowing for that, don't you want to know if a good depth trader makes more or less than an elliot wave trader? Maybe a "good" elliot wave trader makes 10% pa, whereas a "good" depth trader makes 100%.

See what I'm saying? I'm wanting to isolate the method on its own.

 

[Trembling Hand]

See what I'm saying? I'm wanting to isolate the method on its own.

How are you first going to isolate "good" which is by far the more important trait rather than the method.

 

[Gringotts Bank]

That's probably not possible.

But if we had 100 participants, the skill factor across various methods would even out to some degree and become less influential in the measurements.

 

[tech/a]

This is not about comparing systems. Systems are easily compared. What's not easy to compare is TA methods that can't be sytematized, eg. trendline breaks, pattern set ups, elliot wave analysis, vsa etc etc.

I really need to know:

1. Can any of these non-systematizable methods stand on its own?

Of course.

2. What can be thrown out all together?

Lots of stuff.All that you end up not using in YOUR discretionary method.

3. What methods involve a high degree of subjectivity/user experience?

All methods which involve discretionary trading---its discretionary.

Don't you think that's useful info?

No

The 'how' is a different ball game, and I agree that's where the money is made.

So what are you doing with ideas like the one above?

 

[tech/a]

That's probably not possible.

But if we had 100 participants, the skill factor across various methods would even out to some degree and become less influential in the measurements.

I doubt youll get 1 let alone 100

 

[McLovin]

That's probably not possible.

But if we had 100 participants, the skill factor across various methods would even out to some degree and become less influential in the measurements.

Are there even 100 regular participants on this forum? I would have thought it was around 50.

 

[skc]

Don't you think you should at least quote what you suggested on the other thread?

I had an idea for a one-off competition.

Let's choose say 10 speccy stocks (chosen randomly - so that no one is advantaged).
Then a group of practitioners of each of the following methods* can analyze and submit trades (long only) on these ten stocks.... and see which comes out ahead over a set period of time.

*some common methods
-depth reading
-(horizontal) support/resistance
-trendline support/resistance
-swing trading
-breakout trading
-pattern trading (other than breakout)
-wkykoff/vsa
-momentum
-indicator trading (hey, why not?!)
-candle/candle patterns
-gap trading
-your own method/other

You nominate your method, stock, entry time and price, sell time and price.

If "the skill of the practitioner" is your objection and reason for not participating, then how useful is the method? See? Can the method stand alone - that's what I want to know, and so do you! My bet is no method can stand alone. And if I'm right about that, why use a system at all? Why not go the dartboard method?

Position sizing must be a fixed amount and the same for everyone and every trade, for obvious reasons.
Stops and targets can be made with different methods if desired. eg. you can enter on a 'breakout' method and sell on massive momentum spike.

Joe?

There are many problems with the above competition and really won't yield anything meaningful (even if you can get 100 people to participate - which you won't).

First of all, the profitability of any method over a period of time is directly related to the number of opportunities available and the effectiveness in identifying such opportunities. Limiting the competition to 10 randomly selected stocks will not test the above, so any results are by and large useless.

Secondly, unless you have a very large sample size in both number of participants and number of trades taken, random variations in statistical distribution of outcomes will make the results from your small sample competition statistical insignficant.

You better off having a simple survey of practitioners and ask them what is the average return of their method from their experience and average that over all the respondents for that method. You'd probably get an answer no worse than this competition while saving yourself a fair bit of time.

 

[Gringotts Bank]

You better off having a simple survey of practitioners and ask them what is the average return of their method from their experience and average that over all the respondents for that method. You'd probably get an answer no worse than this competition while saving yourself a fair bit of time.

I like that idea apart from the fact that people tend to lie. Perhaps I'll do it anyway, since it's easy to ask.

 

[skyQuake]

Here's a better idea:

1. Work for the ATO.
2. Find the most successful/profitable traders
3. Audit them and claim they're doing inside trading.
4. Learn their methodology as they attempt to explain their legitimate strategy
5. ???
6. Profit!

BONUS:
3b. Turns out they were doing inside trading! Cut a deal. $$$

 

[sinner]

Why don't you do this yourself?

State of Trend Following report does something similar.

You would need to replicate their model IMHO across

* Trend following strategies
* MR or S/R strategies
* Volatility breakout strategies

http://www.automated-trading-system.com/state-of-trend-following-in-february/ for an idea of what I mean.

Needs to be diversified across a large basket of futures. SOTF uses 50 instruments from many exchanges.

My guess is the results would not be very useful, and only confirm what most already know (those who have done the research already) about market regimes and discretionary trading.

 

[peter2]

1. Yes absolutely.
2. Anything that is not useful for each individual trader.
3. All of them.

eg MA crossover systems back test badly, but a competent trader can use the MA crossover and be very profitable.

The "how" you mention is the simple process of skewing the numbers (trade stats) to ensure a profitable outcome over time. Every profitable trader does the same things even though they may use completely different methods.
The methods don't matter.
I repeat. The methods don't matter and again, the methods don't matter.

The methods aren't responsible for their profitable edge. Their methods provide them with trading opportunities. Proper trade management skews their results so that they are profitable over time.

 

[Gringotts Bank]

Here's a better idea:

1. Work for the ATO.
2. Find the most successful/profitable traders
3. Audit them and claim they're doing inside trading.
4. Learn their methodology as they attempt to explain their legitimate strategy
5. ???
6. Profit!

BONUS:
3b. Turns out they were doing inside trading! Cut a deal. $$$

LOL. I have done something similar in that I have asked a few accountants about client traders (no names of course).

 

[Gringotts Bank]

Why don't you do this yourself?

State of Trend Following report does something similar.

You would need to replicate their model IMHO across

* Trend following strategies
* MR or S/R strategies
* Volatility breakout strategies

http://www.automated-trading-system.com/state-of-trend-following-in-february/ for an idea of what I mean.

Needs to be diversified across a large basket of futures. SOTF uses 50 instruments from many exchanges.

My guess is the results would not be very useful, and only confirm what most already know (those who have done the research already) about market regimes and discretionary trading.

Thanks for the link, but I'm more wanting to know about hard-to-test or impossible-to-test TA methods. Anyone with a profitable and purely mechanical system won't share it (nor should they).

My initial hypothesis on the other thread was that method would not matter... then maybe try to disprove that. I tend to agree with you peter, but not totally convinced without numbers. I have a feeling that certain methods are slightly easier to use and more robust than others, even if the difference is small.

 

[tech/a]

Thanks for the link, but I'm more wanting to know about hard-to-test or impossible-to-test TA methods. Anyone with a profitable and purely mechanical system won't share it (nor should they).

Have done --- its now in a book.(Techtrader)
As for the impossible to test discretionary methods---how on earth are you going to follow something that cant be tested---worse how on earth am I or anyone going to put it down on paper and answer the myriad of questions generated by those who dont see what I see?

My initial hypothesis on the other thread was that method would not matter... then maybe try to disprove that. I tend to agree with you peter, but not totally convinced without numbers.

RE READ

The "how" you mention is the simple process of skewing the numbers (trade stats) to ensure a profitable outcome over time. Every profitable trader does the same things even though they may use completely different methods.
The methods don't matter.
I repeat. The methods don't matter and again, the methods don't matter.

The methods aren't responsible for their profitable edge. Their methods provide them with trading opportunities. Proper trade management skews their results so that they are profitable over time

Now
RE READ.

Now thats ALL you need know its in a nut shell.
Everything else is UP TO YOU!

 

[Gringotts Bank]

Have done --- its now in a book.(Techtrader)
As for the impossible to test discretionary methods---how on earth are you going to follow something that cant be tested---worse how on earth am I or anyone going to put it down on paper and answer the myriad of questions generated by those who dont see what I see?



RE READ

Now
RE READ.

Now thats ALL you need know its in a nut shell.
Everything else is UP TO YOU!

A head-and-shoulders pattern can easily be identified and traded. Backtesting is near impossible. That's the sort of thing I'm talking about.

So why do you spend so much time and effort on VSA as opposed to just using a simple MA to find opportunities?

 

[sinner]

A head-and-shoulders pattern can easily be identified and traded. Backtesting is near impossible. That's the sort of thing I'm talking about.

So why do you spend so much time and effort on VSA as opposed to just using a simple MA to find opportunities?

I have no idea how you could consistently 'easily' identify a H&S but not be able to back test it? They are the same thing!

Anyway, Bulkowski has (as usual) done the hard work.

http://www.thepatternsite.com/hst.html

Google 'bulkowski head and shoulders' to get a better listing of bottoms/tops/complex stuff.