How do you know if your method is right? How do you know if TA is valid?

[FishSauce]

I am here in my new workplace as our team is unleashing the final version of a time series analysis software for interest rate derivatives hedging and MBS arbitrage. But I want to relay what happened on Tuesday.....

A particular incidence. I was trading the YM (secretly of course) last Tuesday and a coworker happens to come by and remarked offhandedly, "you know you cant win in the long run, all that technical analysis bullshitting." What struck me was not just his arrogance but also the blatant ignorance. He runs the equity trading portfolio where all day he just search for simplistic autocorrelation - a fancy work for trend and reversal ,i.e. another form of technical analysis! This is like saying that the fastball is horrible pitch and come out throwing the four-seamer. (P.S. Rich Harden is a stud, this year or next year ). When I approached him with this, he casually wave it off ....

This incidence struck me with the fact that quants hide behind "quantitative analysis" or "time series analysis" so as to NOT be associated with technical analysis. TA the dirty faux pas word in this industry. <b>The irony of which is that pretty much ALL these Quant technique are just another form of technical analysis. Imagine the hypocrisy in the fact that Quant are very anti-Technical analysis.</b>


I bring this up because of a peculiar circumstance regarding technical analysis. While "time series analysis" (TA close cousin) has some validity because its backed by <i> Theories</i>, Technical analysis is not! Why is that!!!????

I also bring this up to ask traders HOW and WHY do they know that their method works? You can backtest all you want and give me the Sharpe ratio.... but that's doesnt suffice. That might be a result of data mining. You can show me your Profit/Loss of a steady equity curve for ten year and I can counter with the word, CHANCE. What I am more interested in is the underlying ASSUMPTIONS or THEORIES.


I ran into an issue of Northfield's article (Dan diBartolomeo) that alluded to the fact that technical analysis might work simply because of random walk trendiness. Here's a couple of articles that also mention this phenomen. A MUST READ.
http://satyrican.tripod.com OR http://satyrican.tripod.com/Vol1No1.pdf
ANd
http://www.northinfo.com/papers/index.html ... (A bit more technical)


Another popular assumption is that of psychological self-prophecy:
http://www.sugaronline.com/editorials/editorials/FAQ/technical_analysis.pdf

What's your take?

 

[OddTrader]

Quote from FishSauce:


This incidence struck me with the fact that quants hide behind "quantitative analysis" or "time series analysis" so as to NOT be associated with technical analysis. TA the dirty faux pas word in this industry. <b>The irony of which is that pretty much ALL these Quant technique are just another form of technical analysis. Imagine the hypocrisy in the fact that Quant are very anti-Technical analysis.</b>


I bring this up because of a peculiar circumstance regarding technical analysis. While "time series analysis" (TA close cousin) has some validity because its backed by <i> Theories</i>, Technical analysis is not! Why is that!!!????

Cause and Effect!

:eek:


From Cause (company performance, weather, interest rate, etc.) to Effect (upcoming price, upcoming volume, etc.) ----> Quant Analysis


From Cause (historical price, historical volume, etc.) to Effect (upcoming price, upcoming volume, etc.) ---> TA

 

[OddTrader]

Quote from FishSauce:

I also bring this up to ask traders HOW and WHY do they know that their method works?

What I am more interested in is the underlying ASSUMPTIONS or THEORIES.

imo:

1. From External Cause (company performance, weather, interest rate, etc.) to Effect (upcoming price, upcoming volume, etc.) ----> Quant Analysis. This is a quantitative approach of fundamental analysis. Players are capital-intensive. They are market manipulators based on solid facts. They as well have to react to any market reality, which may be very much against their actions, for any adjustments.

2. From Internal Cause (historical price, historical volume, etc.) to Effect (upcoming price, upcoming volume, etc.) ---> TA. Players are capital-stressful. They are market followers based on trends (or breakouts. Actually they are the same thing if viewed from a distribution perspective). They completelky rely upon market reality (too many books talking about psychology of TA already).

3. Obviously these two approaches are closely related. The causal relationship betweeen them is Cause (QA) to Effect (TA) in the long-run.

 

[damir00]

Quote from FishSauce:

I also bring this up to ask traders HOW and WHY do they know that their method works?

the question is backwards. given that any given (index) tick has an extremely high probability of producing both a profitable long and a profitable short, the better question is why are so many trading system - and by extension traders - UNsuccessful.

imo there is way too much emphasis - at least on ET - on finding entry points, not nearly enough on finding decent exit points, and virtually none on combining both with some kind of rational money management.

and thanks for the links, interesting stuff.

 

[plumlazy]

To find out if your method is any good or not, trade against it in realtime. If it kicks your ass, then it's a goodun! If it doesn't then find another.

 

[FishSauce]

Quote from plumlazy:

To find out if your method is any good or not, trade against it in realtime. If it kicks your ass, then it's a goodun! If it doesn't then find another.

But how do you know if its due to CHANCE or not. What are you underlying assumptions?

There are plethora of traders who hit it big for a year, five year, ten year, etc and then go broke.....

 

[FishSauce]

Quote from damir00:



the question is backwards. given that any given (index) tick has an extremely high probability of producing both a profitable long and a profitable short, the better question is why are so many trading system - and by extension traders - UNsuccessful.

imo there is way too much emphasis - at least on ET - on finding entry points, not nearly enough on finding decent exit points, and virtually none on combining both with some kind of rational money management.

and thanks for the links, interesting stuff.

I couldnt have agree more on the Entry argument.

 

[FishSauce]

Quote from OddTrader:



Cause and Effect!

:eek:


From Cause (company performance, weather, interest rate, etc.) to Effect (upcoming price, upcoming volume, etc.) ----> Quant Analysis


From Cause (historical price, historical volume, etc.) to Effect (upcoming price, upcoming volume, etc.) ---> TA

I think you have it all wrong.

Cause (company performance,etc) -> Effet (upcoming price) = Money Managers (Fundamental analysis)

Cause (Autocorellation, Theta, Volatility, ARIMA) -> Effect (price) = Quant

Autcorealltion, Theta, vol, Arima = advanced technical analysis term

Thus logically Quant = Advanced Technical analysis

 

[Banjo]

Bottom line is that there is always the element of chance . The markets are about people, their fear and greed. The prices at any given moment's closing bar is all that's real. Circumstances, or percieved circumstances control individual behavior. What set of percieved circumstances would change that nice guy next door into a murderer? You don't know but you could probably come up with a set of parameters and indicators if you had your personal financial well being riding on it. The risk is always that the apparent inertia of the last few closing bars will not be interupted by unforseen events before we extricate ourselves thereby ending the risk. We are playing an insane game and looking for reasons to make it not seem that way. You may be just coming to this plateau of understanding.

 

[OddTrader]

Quote from FishSauce:



I think you have it all wrong.

Cause (company performance,etc) -> Effet (upcoming price) = Money Managers (Fundamental analysis)

Cause (Autocorellation, Theta, Volatility, ARIMA) -> Effect (price) = Quant

Autcorealltion, Theta, vol, Arima = advanced technical analysis term

Thus logically Quant = Advanced Technical analysis

I would think you have it 50% wrong , simply because

Advanced technical analysis =\= Technical analysis