Short Time Frames Losing Their Edges?

[inandlong]

Quote from vladiator:




Wouldn't you like what you find to be <b>statistically</b> reliable? Unless the pattern or whatever you wanna call it is reliable statistically, how the heck do you know it's stable?
"Occurs again and again" is alternative to being statistically significant.
Yes, indeed statistics can be misused if need be, but that is usually done to mislead others, not the researcher.

Vladiator as always you make an excellent point, as did daniel_m regarding the same post. I am just too lame to remember where I am posting. :eek:

I agree with both of you.... for the most part, with the reservation that to apply statistical analysis the method must be mechanical. And as such, having very little to no faith in 100% mechanical methods, or perhaps better said, prefering to maintain some control, statistical analysis will not be accurate, and might dissuade me from pursuing further a method that empirically seems doable. Dang I am good at run-on sentences.!!!

Your thoughts guys!

What am I missing here, and "the boat" won't suffice as an answer?

 

[vladiator]

Quote from inandlong:


Vladiator as always you make an excellent point, as did daniel_m regarding the same post. I am just too lame to remember where I am posting. :eek:

I agree with both of you.... for the most part, with the reservation that to apply statistical analysis the method must be mechanical. And as such, having very little to no faith in 100% mechanical methods, or perhaps better said, prefering to maintain some control, statistical analysis will not be accurate, and might dissuade me from pursuing further a method that empirically seems doable. Dang I am good at run-on sentences.!!!

Your thoughts guys!

What am I missing here, and "the boat" won't suffice as an answer?

Not quite sure I understood you correctly, but yes, to apply econometric/statistical analysis, what you are looking at has to be mechanical/objective. With the subjective approaches, you are as good as your judgement

 

[jem]

Also Vlad it is you who take emh theory and say investors can only get market returns. My comment that you quoted questions the validity of the theory. If agents are acting to arb away abnormal returns, why would they stop there, should not these same agents work so as to make sure investors get nothing in the long run. In fact I think that my premise may have the makings of a paper.

After commissions and insiders, institutions, brokers and liars, in the long run investors can make nothing.

A Theory by JEM
Elitetrader dept of thinking outside the box

I will go to dinner now and gladly accept your pen.

 

[vladiator]

Quote from jem:

Also Vlad it is you who take emh theory and say investors can only get market returns. My comment that you quoted questions the validity of the theory. If agents are acting to arb away abnormal returns, why would they stop there, should not these same agents work so as to make sure investors get nothing in the long run. In fact I think that my premise may have the makings of a paper.

After commissions and insiders, institutions, brokers and liars, in the long run investors can make nothing.

A Theory by JEM
Elitetrader dept of thinking outside the box

I will go to dinner now and gladly accept your pen.

Those agents will sure try. But the benchmark (normal return given the level of risk etc) is harder to identify for longer horizons, as I indicated in my earlier post.

"After commissions and insiders..." an average investor should expect to get a normal risk adjusted return less transaction costs.

 

[daniel_m]

Quote from vladiator:
No it's not child's play and yes, some of them do have contracts open while they teach. I know of at least two at this university that trade actively. I'm not a professor yet, but I often give my students a quiz and check the action in the meantime, add/close a position here and there etc. So I don't see your point.
Besides, you misunderstood mine. Although this task might not be easy, the theories provide the tools to better understand and use the action.

wait a minute vlad, you yourself used the words "easily" and "predict".

to mean that means that market action is easily predictable according to the theories these professors have come up with.

well, if they can actually tell what is going to happen, then hey, they must be making millions right? if not, they need their brains checked.

Unless what you are using for trading is grounded in sound economic theory, it has a high probability of being spurious. At least you can't tell if it is. I'm not sure about you, but I'd rather trade something that is not just a result of mindless data mining but is rather based in solid economic principles and thus is more likely to persist.
Cheers.

now stuff like this just kills me.

if it's ME that's doing the analysis and research, i'm just 'data mining' and my conclusions are likely to be spurious.

but if it's a professor with 16 letters after his name -- who is looking at the same data, mind you -- his conclusions are based in 'solid economic principles'.

well, i just have to take you up on that last point.

before that said professor sat down to do his latest research, the 'solid economic principles' in existence until then had failed (obviously) to make any consistent predictions about market action, right? so that researcher is then coming up with new hypotheses and theories -- or, in other words, looking at some set of data, forming some hypotheses, testing them out, (exactly what i do, btw) and then adding his conclusions to the body of 'solid economic principles'. man, i wish i had a can't lose proposition like economic theory of financial markets -- you know, the kind where no matter how bad/wrong it is, it's infinitely updatable...

 

[daniel_m]

Quote from vladiator:



Those agents will sure try. But the benchmark (normal return given the level of risk etc) is harder to identify for longer horizons, as I indicated in my earlier post.

"After commissions and insiders..." an average investor should expect to get a normal risk adjusted return less transaction costs.

if you've got a moment vlad, help me out here a little:

how do you boys (academics) define the "level of risk" for a long term buy and hold?

 

[vladiator]

Quote from daniel_m:



wait a minute vlad, you yourself used the words "easily" and "predict".

to mean that means that market action is easily predictable according to the theories these professors have come up with.

well, if they can actually tell what is going to happen, then hey, they must be making millions right? if not, they need their brains checked.





now stuff like this just kills me.

if it's ME that's doing the analysis and research, i'm just 'data mining' and my conclusions are likely to be spurious.

but if it's a professor with 16 letters after his name -- who is looking at the same data, mind you -- his conclusions are based in 'solid economic principles'.

well, i just have to take you up on that last point.

before that said professor sat down to do his latest research, the 'solid economic principles' in existence until then had failed (obviously) to make any consistent predictions about market action, right? so that researcher is then coming up with new hypotheses and theories -- or, in other words, looking at some set of data, forming some hypotheses, testing them out, (exactly what i do, btw) and then adding his conclusions to the body of 'solid economic principles'. man, i wish i had a can't lose proposition like economic theory of financial markets -- you know, the kind where no matter how bad/wrong it is, it's infinitely updatable...

Are you doing this on purpose to annoy me or do you actually believe what you say?
The word easily meant that the models are easily applicable, not that the action is easy to predict.
On the second frontier of your argument, you seem to have a very vague idea of how academics do research.
They <b>FIRST</b> come up with a sound prediction based on the theoretical model(s) and <b>THEN</b> look at the data to (dis)prove it. Otherwise, it's data-snooping. You can find all kinds of crap that will seem significant given enough looking and a big enough sample and then you dont' have to be really smart to come up with a plausible theory to explain it. That's not research, that's BS. Do some people do it? Yes, unfortunately they do. But it's always clearly apparent and never gets much attention. That's all I was saying.
Here's an example, I happen to have a theory that predicts what will happen under certain conditions as the large institutions move on way or the other, as new information gets disseminated to different classes of investor etc etc. I then went and analyzed about 200GB of data and the theory worked remarkably and was robust in lots of ways. Is that the kind of approach you use? Then more power to you, you are on the right track.
If you are just looking to find patterns regardless of why there are there, most likely there are spurious. See the difference?

 

[vladiator]

Quote from daniel_m:



if you've got a moment vlad, help me out here a little:

how do you boys (academics) define the "level of risk" for a long term buy and hold?

I don't do long term studies (for the reasons I indicated), but the model that is typically used to adjust for normal (risk adjusted returns) is either

Return_on_securityJ - Riskfreerate = Alpha + beta (MarketReturn-RiskFreerate) + error.

Or the Fama-French three-factor model that adds book to market and size factors. Another variant that's now frequently used is the Fama-French model plus a 4th Momentum factor.

Let me know if you are interested, I'll point you to some papers.
I only do very short term studies b/c the normal return over say 2 minutes is pretty much indistinguishable from zero and I don't care that I don't have a good model to proxy for it.

Edit: thus, risk would be in the loadings on the factors.

 

[inandlong]

Quote from vladiator:
...to apply econometric/statistical analysis, what you are looking at has to be mechanical/objective. With the subjective approaches, you are as good as your judgement

Excellent... thanks for the reply. I was hoping you would say that. It is the 'am I right or wrong' that makes for the fun of trading.

PS - Corso, you get my vote for best new thread starter. Every time you start one, it seems it takes off. Good job, and thanks.

 

[bone]

I'm getting sleepy... very sleepy.