Replying to a Message

Microsoft HPC

Forum Navigation:

login

FORUMS > General Forum

Replying to Topic: Quantitative Analysis is a higher form of Technical Analysis
Created On Thu Feb 08, 07 03:05 AM by jomni

jomni
Senior Member

Posts: 989
Joined: Jan 2005

Thu Feb 08, 07 03:05 AM

User is offline

Here's something that I posted in my blog:
Quantitative Analysis = "Highly" Technical Analysis (?)

Ok quants, I'm ready for your bashing.

-------------------------
visity my (dead) blog:

rmquant.blogspot.com

Edited: Thu Feb 08, 07 at 08:26 AM by jomni

Ceres629
Member

Posts: 111
Joined: Jul 2006

Thu Feb 08, 07 03:33 AM

User is offline

Interesting read. Having just started learning about trading methodologies and I could never really understand the differentiation between quantitative analysis and technical analysis. I would really like to hear some opinions on how they differ...

I especially like the analogy with astrology and astronomy. Although i'm sure technical analysis has more merit than astrology.

jomni
Senior Member

Posts: 989
Joined: Jan 2005

Thu Feb 08, 07 06:19 AM

User is offline

Well the main difference i can think of is that technical analysis does not think price movements are random (there is an inherent pattern in them)... while quantitative analysis models this randomness in a certain way. If you're familiar with Taleb and all this talk about being "Fooled by Randomness". It looks like he's talking about focusing on behavioral finance and market dynamics which is Technical Analysis instead of Quantitative Analysis.

-------------------------
visity my (dead) blog:

rmquant.blogspot.com

Edited: Thu Feb 08, 07 at 06:21 AM by jomni

ZmeiGorynych
Senior Member

Posts: 875
Joined: Jul 2005

Thu Feb 08, 07 10:31 AM

User is offline

Technical analysis is a fairly well defined tradition with a finite, known set of techniques - in that way similar to 'system dynamics' in general (non-finance) modeling. 'Quantitative analysis' in finance is any kind of number crunching that a trader can't do in his head.

Thus clearly there is a lot of overlap. I myself am a quant, and have often drawn inspiration from particular technical analysis tricks. Of course, the tricks in their raw form are usually useless because so many people look at them. The kind of questions that T.A. methods attempt to answer can actually teach you more than the specific methods they use.

-------------------------
You give a hungry man a fish, he owes you one fish. You teach him to fish, you lose your monopoly on fishing.

yabbadabba
Senior Member

Posts: 261
Joined: Jul 2006

Thu Feb 08, 07 11:14 AM

User is offline

I'd be highly interested in a thorough discussion on this topic.

First off, what you mean by quantative analysis? Science is quantitative analysis. I think you are refering to standard financial theory which has its own assumptions and paradigms.

Now the next question is, what is Technical Analysis? Due to it's lacking scientific background most of TA is home-brewed stuff which very often suffers from the representativeness bias and gambler's fallacy. Most technical analysists would say that TA can't be measured which really means that they don't have the mathematical tools for measuring properly.

I would like to argue that both approaches have one problem and that is they don't explain human behaviour or only through a very skewed lens.

So to answer the question, no, technical analysis is not quantitative. Both QA and TA make the same falsy assumption, namely that prices are agents. Prices don't act, they are the result of action.

-------------------------
"Our criticism of the accepted classical theory of economics has consisted not so much in finding logical flaws in its analysis as in pointing out that its tacit assumptions are seldom or never satisfied, with the result that it cannot solve the economic problems of the actual world.�

Keynes, The General Theory of Employment, Interest and Money

jomni
Senior Member

Posts: 989
Joined: Jan 2005

Thu Feb 08, 07 11:46 AM

User is offline

Quote
Originally posted by: yabbadabba
I'd be highly interested in a thorough discussion on this topic.

First off, what you mean by quantative analysis? Science is quantitative analysis. I think you are refering to standard financial theory which has its own assumptions and paradigms.

Now the next question is, what is Technical Analysis? Due to it's lacking scientific background most of TA is home-brewed stuff which very often suffers from the representativeness bias and gambler's fallacy. Most technical analysists would say that TA can't be measured which really means that they don't have the mathematical tools for measuring properly.

I would like to argue that both approaches have one problem and that is they don't explain human behaviour or only through a very skewed lens.

So to answer the question, no, technical analysis is not quantitative. Both QA and TA make the same falsy assumption, namely that prices are agents. Prices don't act, they are the result of action.

I agree to your last statement. In an ever-changing environment, past price movements may be meaningless.

-------------------------
visity my (dead) blog:

rmquant.blogspot.com

jbott
Junior Member

Posts: 8
Joined: Jan 2007

Thu Feb 08, 07 03:14 PM

User is offline

Doesn't TA basically assume that prices have memories (granted, this excludes whereas quantitative analysis would seek to explain relationships between prices (or other variables)? Also, I would contend that prices ARE agents (in response to yabbadabba) as prices reflect human actions (viz., the decision to buy or sell).

bskilton81
Member

Posts: 158
Joined: Dec 2004

Thu Feb 08, 07 04:30 PM

User is offline

I would basically say that technical analysis is often an ignorant form of quantitative (or statistical) analysis. I think you'd be hard pressed to find a quant who doesn't believe, for instance, that FX rates display persistent momentum effects (this phenomenon is well documented in academic research). The difference is that if a quant is going to model momentum, he or she does it in a more mathematically rigorous way, doing things like correcting for differences in volatility that would make random noise look like momentum.

Usually a technical analyst looks at a charts, makes something up, and then trades on it, without being cognizant of the statistical errors he or she might be making. The best example is the tendency for a chartist to look at equity prices in a graph, while a quant (who probably won't even use a graph), would look at persistence of percentage or log returns (especially if you're looking at a long data set), understanding that this method better captures the statistical dynamics of the market. In fixed income markets, technical analysts may look at bond futures prices, while quants will model yields.

Or technical analysts may do a regression on the levels of two equity prices. A quant would be unlikely to do this, because prices are not covariance stationary. Instead, a quant would use log returns. Consequently, a technical analyst is more likely to find a spurious correlation.

Edit: I would add that in my limited experience, technical analysts are less cognitive of the covariance matrix of their holdings (they look at a bunch of charts and pick different things based on what they see, and the portfolio tends to be constituted on an ad hoc, spontaneous basis). To a quant making investment decisions, the covariance matrix is the main point of interest, and everything else is just commentary.

Edited: Thu Feb 08, 07 at 04:34 PM by bskilton81

Tadragh1
Member

Posts: 114
Joined: Jul 2006

Thu Feb 08, 07 05:26 PM

User is offline

I consider Technical Analysis to be a term which intersects with Quantitative Analysis. Bskilton81, You seem to associate TA with "chartism", which is not accurate in my opinion. I would define TA as any way of creating trading rules while basing on the past. With this definition, ARIMA should be qualified as TA also. What about neural networks, pattern recognition etc. In my opinion they all belong to TA as well as QA.
There are quite a few authors trying to "quantifise" TA and a number of papers testing the forecasting power of various TA tools. For example, there is a nice paper by Neftci about how a moving average (widely used in TA) seems to capture nonlinear nuances which remain undetected by some classical models.

N
Senior Member

Posts: 2816
Joined: May 2003

Thu Feb 08, 07 06:40 PM

User is offline

Quote
Originally posted by: jomni
Here's something that I posted in my blog:
Quantitative Analysis = "Highly" Technical Analysis (?)

Ok quants, I'm ready for your bashing.

I'd almost agree, if you said "highly mathematical technical analysis" (meaning non-stochastic analysis) and threw away all that junk you mentioned in your blog.

Pannini
Member

Posts: 147
Joined: Mar 2005

Thu Feb 08, 07 07:34 PM

User is offline

I think it is (somewhat) incorrect to equate Technical Analysis with "chart reading." This is because Technical Analysts distinguish between "reading the tape" and "reading charts," but you only covered chart reading. There are many similarities between opportunistic trading algos and "tape reading."

yabbadabba
Senior Member

Posts: 261
Joined: Jul 2006

Thu Feb 08, 07 08:34 PM

User is offline

Quote
Originally posted by: bskilton81

Usually a technical analyst looks at a charts, makes something up, and then trades on it, without being cognizant of the statistical errors he or she might be making.

On the other hand the quant is not cognizant of the underlying effect of prices, namely the behaviour of agents in the financial market. That is the idea of technical analysis in the first place.

Quote
Also, I would contend that prices ARE agents (in response to yabbadabba) as prices reflect human actions (viz., the decision to buy or sell).

But both QA and TA are in my view wrong in that they believe that prices act. That is a major categorical error! Prices are not agents, they are the result of cooperation of agents. Of course it's very hard to 'reengineer' prices to extract information about the original cause, but in my view it is the only way to go (from a scientific viewpoint).

Let me give an analogy. Say you want to explain the structure of cities. One method is to model the morphology in an abstract way without refering to architecture or society in general. There is no way this model can explain the difference between modern new york and ancient rome in detail because it leaves out the things that matter (transport systems, sewerage, style of living, in essence human behaviour).

-------------------------
"Our criticism of the accepted classical theory of economics has consisted not so much in finding logical flaws in its analysis as in pointing out that its tacit assumptions are seldom or never satisfied, with the result that it cannot solve the economic problems of the actual world.�

Keynes, The General Theory of Employment, Interest and Money

lewishortthemall
Member

Posts: 47
Joined: Jan 2007

Thu Feb 08, 07 08:45 PM

User is offline

Technical analysis ,which is a very broad field , is everything that has to do with the motto : "history repeats himself" ,so any technique that uses past data to forecast future prices can be linked to technical analysis no matter how complex it is .Now do the quants want to be assimilated with the average amateur chartist who looks at charts with his bunch of "prop indicators"(e.g. the "ultimate" oscillator v3.14 or other lagging grails) ?

yabbadabba
Senior Member

Posts: 261
Joined: Jul 2006

Thu Feb 08, 07 08:53 PM

User is offline

Quote
Originally posted by: lewishortthemall
Technical analysis ,which is a very broad field , is everything that has to do with the motto : "history repeats himself" ,so any technique that uses past data to forecast future prices can be linked to technical analysis no matter how complex it is

Sorry, but no! That thing, where we extrapolate from history to future in a mannered way, we call science. TA is often just bad science, because people don't have the mathematical tools to map their often nonlinear observations to a structured model, so they stick to case-based reasoning.

-------------------------
"Our criticism of the accepted classical theory of economics has consisted not so much in finding logical flaws in its analysis as in pointing out that its tacit assumptions are seldom or never satisfied, with the result that it cannot solve the economic problems of the actual world.�

Keynes, The General Theory of Employment, Interest and Money

StatTrader
Member

Posts: 167
Joined: Jan 2007

Thu Feb 08, 07 09:30 PM

User is offline

Quote
Originally posted by: bskilton81
I would basically say that technical analysis is often an ignorant form of quantitative (or statistical) analysis. I think you'd be hard pressed to find a quant who doesn't believe, for instance, that FX rates display persistent momentum effects (this phenomenon is well documented in academic research). The difference is that if a quant is going to model momentum, he or she does it in a more mathematically rigorous way, doing things like correcting for differences in volatility that would make random noise look like momentum.

Usually a technical analyst looks at a charts, makes something up, and then trades on it, without being cognizant of the statistical errors he or she might be making. The best example is the tendency for a chartist to look at equity prices in a graph, while a quant (who probably won't even use a graph), would look at persistence of percentage or log returns (especially if you're looking at a long data set), understanding that this method better captures the statistical dynamics of the market. In fixed income markets, technical analysts may look at bond futures prices, while quants will model yields.

Or technical analysts may do a regression on the levels of two equity prices. A quant would be unlikely to do this, because prices are not covariance stationary. Instead, a quant would use log returns. Consequently, a technical analyst is more likely to find a spurious correlation.

Edit: I would add that in my limited experience, technical analysts are less cognitive of the covariance matrix of their holdings (they look at a bunch of charts and pick different things based on what they see, and the portfolio tends to be constituted on an ad hoc, spontaneous basis). To a quant making investment decisions, the covariance matrix is the main point of interest, and everything else is just commentary.

Nicely put. This pretty much sums up my view on the subject too.

Having said that, it is often the case that "ignorant" quant analysis + market intuition > quant analysis + lack of market intuition. Very often, if the idea is sound, a trading strategy will still work even if there implementation is less than perfect. The same cannot be said of the reverse.

lewishortthemall
Member

Posts: 47
Joined: Jan 2007

Fri Feb 09, 07 02:07 AM

User is offline

Quote
Originally posted by: yabbadabba
Quote
Originally posted by: lewishortthemall
Technical analysis ,which is a very broad field , is everything that has to do with the motto : "history repeats himself" ,so any technique that uses past data to forecast future prices can be linked to technical analysis no matter how complex it is

Sorry, but no! That thing, where we extrapolate from history to future in a mannered way, we call science. TA is often just bad science, because people don't have the mathematical tools to map their often nonlinear observations to a structured model, so they stick to case-based reasoning.

You're right ,Technical Analysis is a lower form of Quantitative Analysis actually !

Ceres629
Member

Posts: 111
Joined: Jul 2006

Fri Feb 09, 07 06:06 AM

User is offline

What a great reply bskilton81, you've certainly cleared up a lfew of misconceptions I had regarding quantitative analysis.

KackToodles
Senior Member

Posts: 3971
Joined: Aug 2005

Fri Feb 09, 07 09:09 AM

User is offline

Quote
Originally posted by: Pannini
I think it is (somewhat) incorrect to equate Technical Analysis with "chart reading." This is because Technical Analysts distinguish between "reading the tape" and "reading charts," but you only covered chart reading. There are many similarities between opportunistic trading algos and "tape reading."

what is the difference between chart reading and using back-fitted time series to predict the future? nothing.

yabbadabba
Senior Member

Posts: 261
Joined: Jul 2006

Fri Feb 09, 07 11:07 AM

User is offline

Quote
Originally posted by: KackToodles
what is the difference between chart reading and using back-fitted time series to predict the future?

There are many differences. The methods, the number of observations, the approach, the time for modelling and execution, the degree of formalism.

-------------------------
"Our criticism of the accepted classical theory of economics has consisted not so much in finding logical flaws in its analysis as in pointing out that its tacit assumptions are seldom or never satisfied, with the result that it cannot solve the economic problems of the actual world.�

Keynes, The General Theory of Employment, Interest and Money

N
Senior Member

Posts: 2816
Joined: May 2003

Fri Feb 09, 07 01:20 PM

User is offline

Quote
Originally posted by: bskilton81
I would basically say that technical analysis is often an ignorant form of quantitative (or statistical) analysis.

bskilton,

That's right. But lets take your statement further... Isn't statistical analysis also just an ignorant form of quanitative analysis?

For a simple example, random number have increasing entropy, but those who simulate random numbers use random number generators that do not increase entropy (computer generated random numbers obviously have fixed Kolmogorov complexity, ie the 20 lines of code that generate the sequence). So using MC simulation in quantitative finance is just as ignorant to me (and many others) as those using charts are to you.

Ignorance is relative!

N

FORUMS > General Forum

Forum Navigation:

© All material, including contents and design, copyright Wilmott Electronic Media Limited - FuseTalk 4.01 © 1999-2010 FuseTalk Inc.