Can linear regression analysis really predict the future?

[good one]

Quote from cfd_trader:

As for me my conviction may be summarized this way:

the Market is not predictable, except when it is.

This is the flaw though of academic studies on TA..."predictability" implies win rate better than random...
Win rate is totally meaningless without taking into account how much you lose vs how much you win on any given trade...Otherwise you can't even start to formulate the "E"xpected "V"alue of a trade you put on..(EV+/-..not VE as I assume the poster above is trying to say)
Things are further compounded by that the mean will be almost meaningless, the outliers are what will matter..This is why cutting losses quick and letting winners run, although "old school" stuff everyone knows, is the correct way to trade.
The reason no one has figured out the "distribution" of the market from t0 to t1 is that it is unknowable. Exactly because to know that distribution, would alter that distribution because of liquidity.

 

[good one]

Quote from Syprik:

Rather interesting you believe that is the "only" reliable way of making money in the markets. This is an increasingly common "quantway or the highway" attitude that is both arrogant and misplaced. Tell that to the 1000's of successful traders before and after the dawn of ARCH mingling with GARCH @ the CoIntegrated Surface Distribution party.

I don't totally disagree with you...However its pretty lazy to disgard quant tools entirely, because some idiots use them wrong.

 

[good one]

Quote from sosueme:

What has this laboratory outcome got to do with the reality that only a very small number of flippers control and manipulate 85%+ of the market.

I can easily grasp the isolated math of your probability example, but I think you are in need of a greater level of thought and direction when applying it to an asymmetrical object such as the market.

Anyway, as always, I will follow this thread with great interest.

You should start studying some game theory..this is great text to start with..
http://www.amazon.com/Playing-Real-...=sr_1_1?ie=UTF8&s=books&qid=1254784634&sr=8-1

In a game with a small number of participants coin flipping who control 85% of the flips probability, if you were a betting person..you would not then care about the distribution of such a game's coin flips, you would simply try to deduce the bias/direction of the small number of participants and bet with them..
Too bad this is a much easier game to "win" betting wise than the markets...

 

[sosueme]

Help me out with this example please.

It is clear from the postings that there are some very clever maths people onboard

Here is a breakdown of strikes (orders) placed during a single day and it is representative of most days........

0-10 ... 80,000
11-20 ...10,000
21-50 ...5,000
51-100 ...8,000
101+ ...1,000

In total they represent 1,400,000 contracts traded.

What can be derived from this information.

 

[Martinghoul]

Quote from Syprik:
Professionals who tend to be highly invested in their intelligence, particularly self-esteem wise, tend to be the most rebellious. Many come to realize that basic time-proven trend following behavior completely annuls their need and ability to understand complicated econometrics, including the emotional need to develop esoteric models that attempt to show why equities/commodities go up,down or revert to a mean. That typically doesn't feel good as there is an inherently strong desire for appreciation.

I would agree with this. And that is yet another 'fad'/herding/crowding bias that can potentially be exploited.

I am not sure whether my argument coincides with MAESTRO's, but all I was trying to say was that a lot of my problems with TA are not about the premise per se, but with the quality/robustness of the analysis.

 

[sosueme]

Quote from Martinghoul:

I would agree with this. And that is yet another 'fad'/herding/crowding bias that can potentially be exploited.------

I would be interested to know how you can exploit this group.

 

[Martinghoul]

Quote from sosueme:

I would be interested to know how you can exploit this group.

Well, you'd do it the same way as you would exploit the knowledge that there's a crowded trade out there. Except in this case it's not a crowded trade, but rather a crowded methodology that forces certain synchronized behavior.

One of the best, most dramatic examples sometimes offered is the role of CPPI-driven strategies in the Black Monday crash.

 

[sosueme]

Quote from Martinghoul:

Well, you'd do it the same way as you would exploit the knowledge that there's a crowded trade out there. Except in this case it's not a crowded trade, but rather a crowded methodology that forces certain synchronized behavior.

One of the best, most dramatic examples sometimes offered is the role of CPPI-driven strategies in the Black Monday crash.

I gather you want to position yourself in front of the herd.

Can you be more specific about how you would use LRA (or anything else for that matter) to achieve this goal.

 

[Syprik]

Quote from jdeezero05:

However its pretty lazy to disgard quant tools entirely, because some idiots use them wrong.

Of course, agreed. Just stating that there are other herds of traders that will continue to remain highly successful before and after the ARCH/GARCH scene.

For those interested, an introductory paper on garch/arch:

http://www-stat.wharton.upenn.edu/~steele/Courses/434/434Context/GARCH/garch101(ENGLE).pdf

 

[sosueme]

It seems a pity that such an interesting thread as this has run out of steam.

Perhaps we need to wait until "the LRA traders" get home after a busy day and can log onto ET.

I remain as enthusiastic as ever that someone can push forward and show how to translate LRA into profits.

I even provided some simple inputs a few posts back, in the hope that some learned soul could wield their mathematical magic and lead me out of the wilderness.

Still, there is always this evening to come.