What is Quant Trading?

[NoVoodooHere]

I have no intention of getting into a spit fest, but does this "volpunter" ever NOT have an opinion? I see that he's all over ET, on each and every thread, handing out his unwelcomed condescending remarks. Should anyone disagree with his pompous remarks, he pisses everyone off with his half-ass invectives.

He speaks down on TA with authority like as if he ever used it but just when does he ever trade when he's spending all his waking hours posting here on ET? I can't believe this crap is being tolerated. Seriously! :eek:

Yo Skitzo, volpunter is an ass#ole. It's not his fault he was just born that way, or so he thinks. We try to keep the dirtbag busy with janitorial activities most of the time, but he refuses to take his pills.

Block the douche bag like I did

 

[xandman]

If you mean wanting to generate profits is the common theme then I agree. But you are basically saying playing lotto or gambling and an ordinary income generating job are essentially the same. Not sure I can agree with that just because they both may aim at generating a return. Sure, you can put into either one mediocre people or extremely smart people but the odds are most always stacked against the one on gambling/lottery side. How is applying finance with a scientific and proven approach the same than applying some esoteric lines and scribbles on charts and believing that the market bends to support and resistance rather than new incoming fundamental information?

One can apply the scientific method to come up with a strategy based on technical analysis.

TA not simply visualization support and resistance. It also encompasses statistics. Bollinger Bands is a good example.

However, what's wrong with support and resistance? A quant referencing calculus would call it a local minimum and a local maximum. Can the creation of a new local min/max prior to a news release possibly foretell a move further in the direction? Do you think there will be pauses at previous min/max depending on the magnitude of the move?

 

[dartmus]

One can apply the scientific method to come up with a strategy based on technical analysis.

TA not simply visualization support and resistance. It also encompasses statistics. Bollinger Bands is a good example.

However, what's wrong with support and resistance? A quant referencing calculus would call it a local minimum and a local maximum. Can the creation of a new local min/max prior to a news release possibly foretell a move further in the direction?

You misunderstand a basic principle. The clients of an investment fund salesman want to know they're getting QA for their money ...not TA or daytrading. This 'conflict of interest', where income is dependent on promoting QA and shunning both TA and daytrading ... this is what needs to be mitigated before you can debate facts.

in the meantime URF isn't going to stop bashing TA and daytrading, and calling us morons.

 

[xandman]

You misunderstand a basic principle. The clients of an investment fund salesman want to know they're getting QA for their money ...not TA or daytrading. This 'conflict of interest', where income is dependent on promoting QA and shunning both TA and daytrading ... this is what needs to be mitigated before you can debate facts.

in the meantime URF isn't going to stop bashing TA and daytrading, and calling us morons.

Roger. My term for Quant as "rebranding" has a cynical tone and may have struck a nerve with people who call themselves quants for a living. Poor choice of words. What I simply wanted to express is that the math has always been around. Though admittedly, TA has it's tangents. Much like Astronomy in the physical sciences. ( If I get flamed by an Astronomer, I'll just stop posting.)

 

[volpunter]

No there won't be pauses and there won't be any foretelling because what happened to prices two hours ago has zero bearing on whether price now will remain below or exceeds the previous high. What happens to prices now is a function of the information disseminated now and decisions market participants make NOW and not the reaction by market participants NOW to prices two hours ago. Maybe 5% or so of past prices affect prices now but some TA apostles make it apoear as if past prices explain current and future prices to the degree of almost 90%. Uncountable studies have disproved such theory.

Ask yourself and rigorously test whether previous price highs have a bearing on future prices. Let's say our hypothesis is that past highs form s resistance level and we want to test whether future prices stay below such resistance levels. You will find out that you must reject such hypothesis on the grounds of poor statistical results. That is the honest way to do it. The dishonest way is to bark like a demon possessed monkey from mom's basement and swear at everyone who dares to question the validity of Technical Analysis.

One can apply the scientific method to come up with a strategy based on technical analysis.

TA not simply visualization support and resistance. It also encompasses statistics. Bollinger Bands is a good example.

However, what's wrong with support and resistance? A quant referencing calculus would call it a local minimum and a local maximum. Can the creation of a new local min/max prior to a news release possibly foretell a move further in the direction? Do you think there will be pauses at previous min/max depending on the magnitude of the move?

 

[d08]

If you mean wanting to generate profits is the common theme then I agree. But you are basically saying playing lotto or gambling and an ordinary income generating job are essentially the same. Not sure I can agree with that just because they both may aim at generating a return. Sure, you can put into either one mediocre people or extremely smart people but the odds are most always stacked against the one on gambling/lottery side. How is applying finance with a scientific and proven approach the same than applying some esoteric lines and scribbles on charts and believing that the market bends to support and resistance rather than new incoming fundamental information?

When dealing with economics, nothing is completely scientific in this field because there is quite a bit of randomness.
By TA I mean methods that are quantified, people who draw lines and use discretionary methods aren't comparable because these methods can't be objectively compared to quant methods. But comparing something simple like RSI with co-integration is perfectly valid since both are based on same data series.
You go into extremes when you compare the worst of TA to the best of quants. I'm sure there is a term for this king of argument. Plenty of quants build models that are based on false assumptions, not unlike TA followers.
Case and point: LTCM. Pretty much everyone thought their methods are solid and they were applying scientific methods, yet...

 

[volpunter]

I guess TA has never been properly defined, and maybe that is what worries me a bit that it simply is not clearly definable. When I speak of TA I am talking about deriving prediction and current state of price levels purely from past prices and maybe volume as well. Quantitative Analysis does not put such constraint on its definition. In fact a quantitative trading strategy can derive its metrics purely from fundamental factors or news or economic release data. TA as it is defined and used on this site seems to limit itself to past price analysis and I do not believe there is validity in such approach alone because there are countless studies in academic and practitioner realm that disprove that past prices alone have any predictive power. Why do you think investment banks and hedge funds plow tens if not hundreds of millions of dollars into architectures and analytical approaches that go way beyond price analysis? Why do they hire very expensive engineers to dig way deeper than just historical price time series? It is because pure price analysis is without merit.

What shocks me is that some die-hards insist on believing looking at prices alone has merit and that everyone in the professional trading and investment industry must be dumb and must have missed something such individuals on this site are seeing. Do I have proof that tweaking past prices to squeeze out some indicator has no merit? No I do not have such proof but I believe empirical evidence puts the chances of someone, who focuses on prices alone succeeding, at almost zero. Having said that Bollinger bands, static support/resistance levels, fib retracement, and virtually anything that relies exclusively on past prices is virtually worthless and proven to fail.

Re your LTCM argument I disagree with you: You pick one example of someone who did not manage risk prudently. Their analytical approach may have been sound it may not have been. Fact why LTCM failed is that they over leveraged and did not allow a margin of error. Hundreds of other funds approach trade idea generation in a very similar way than LTCM did and in conjunction with applying sound money management they succeed day in day out. They may have lackluster years but they do not have to top up their bankrupt bank accounts over and over and over like most TA apostles on this site.

When dealing with economics, nothing is completely scientific in this field because there is quite a bit of randomness.
By TA I mean methods that are quantified, people who draw lines and use discretionary methods aren't comparable because these methods can't be objectively compared to quant methods. But comparing something simple like RSI with co-integration is perfectly valid since both are based on same data series.
You go into extremes when you compare the worst of TA to the best of quants. I'm sure there is a term for this king of argument. Plenty of quants build models that are based on false assumptions, not unlike TA followers.
Case and point: LTCM. Pretty much everyone thought their methods are solid and they were applying scientific methods, yet...

 

[dartmus]

When dealing with economics, nothing is completely scientific in this field because there is quite a bit of randomness.
By TA I mean methods that are quantified, people who draw lines and use discretionary methods aren't comparable because these methods can't be objectively compared to quant methods. But comparing something simple like RSI with co-integration is perfectly valid since both are based on same data series.
You go into extremes when you compare the worst of TA to the best of quants. I'm sure there is a term for this king of argument. Plenty of quants build models that are based on false assumptions, not unlike TA followers.
Case and point: LTCM. Pretty much everyone thought their methods are solid and they were applying scientific methods, yet...

Almost everyone with either common sense or advanced technical knowledge agrees with you. However you're mistaken about line drawing methods. Tho scribbles aren't as easy to compare to QA as RSI is it's still possible to use a consistent process to draw lines. I'm not saying everyone drawing lines has tried to make their line drawing non-discretionary. I'm saying line drawing can be quantified and automated.

I Like your post. It illustrates clear thinking.

 

[learner2007]

I'm saying line drawing can be quantified and automated.

I've yet to see software that can draw trend lines correctly. Really hit and miss, and with few hits. Makes me feel good that I can do something better than a machine.

 

[dartmus]

I've yet to see software that can draw trend lines correctly. Really hit and miss, and with few hits.

You missed the point of my message. Software can only do what you tell it to do but it doesn't have any difficulty doing things consistently. This is in contrast to humans who are manually drawing lines. Manually drawn lines often deviate from the rules, for instance skipping past a condition which the observer didn't notice or worse applied filters which weren't part of the rules.

My point was lines can be just as objective as any thing else. This line of thought isn't considering the potential profitability of any method at the moment. It only clarifies lines as a method that are just as quantifiable and thus backtestable as other items like RSI and QA.