Forums  > Trading  > Better Trading Decisions  
     
Page 6 of 7Goto to page: 1, 2, 3 ... 5, 6, 7 Prev Next
Display using:  

Nonius
Founding Member
Nonius Unbound
Total Posts: 12805
Joined: Mar 2004
 
Posted: 2005-05-03 15:55
getting back to DNA and languages, and, I understand that predicting sign is sooo six months ago...did you ever predict sign on a skewed basis?  that is, predict whether X is above K where K varies over a wide range of possible values?  it is a collection of binary predictions, but, in the limit its answer is a prediction of the distribution of the outcome..now, the intereseting thing is, that for a discrete partition of the real line, the prediction is a sequence of 1s and -1s....ie, a sort of language..;when everyting is 1, that is an indication of a directional bet...when everyting is -1 same thing..;when they are mixed....I think that is a sign of volatility pending....anyway, its all about signs I tell you, like Crop circles.

Chiral is Tyler Durden

kr
Founding Member
NP Raider
Total Posts: 3561
Joined: Apr 2004
 
Posted: 2005-05-03 16:50
cf. Campbell/Lo/MacKinlay on 1/-1...

something I'm sort of musing about though would be the sector-vector version of this... i.e.
- take some sector indices
- look at the vectors of sector-specific ups/downs
- is there a kind of rotation pattern that has any predictability at all?

Qualitative reasoning is sectoral flight-to-quality.  We know that you can do some of this with PCA (i.e. Bouchaud/Potters), but just make it a little more black+white - which sector combinations tend to trade together?

my bank got pwnd

Nonius
Founding Member
Nonius Unbound
Total Posts: 12805
Joined: Mar 2004
 
Posted: 2005-05-03 17:28
as a matter of fact, I have been exploring ratio of ups to downs.....it is interesting......if there is predictability there, then, it can tell you if your huge vat of tiny bets that are "evenly distributed" between long and short positions will pan out on the neutrality side...

Chiral is Tyler Durden

opmtrader
Founding Member

Total Posts: 1333
Joined: Mar 2004
 
Posted: 2005-05-03 17:49
NIP247, regarding the Crabel comment, that would be so mindbogglingly awesome. You just gave me inspiration to keep going today. Thanks dude!

This thread is awesome. There is still so much to talk about, sectors, system deactivation, portfolio selection, genetic algos, etc. I'll be sure to be back to ask you geniuses a bunch more questions later.

Nonius
Founding Member
Nonius Unbound
Total Posts: 12805
Joined: Mar 2004
 
Posted: 2005-05-03 17:58
see, OPM, everyong is laughing at my appeal to dumb machines, and I blame another sector of dumb machines-television- for this.  but, I am not proposing to black box trade with dumb machines...phuck that.  I know better....I've visited enough explosive trend folloiwng HFs to be dissuaded from following in their footsteps like a tiny television watching lemming into the sea...no NO NO!  my idea is simply to build a humongous parallel distributed network of machines, possibly distributed over the interneet and shit, using some fancy shared resource version of Naptster, and then build little Nonian Mining machines that will randomly mutate and phuck each other, all in the simple pursuit of predicting distributions of outcomes....tat is all I want.

Chiral is Tyler Durden

FDAXHunter
Founding Member

Total Posts: 8372
Joined: Mar 2004
 
Posted: 2005-05-03 18:01
That too, has been done already. Pretty much exactly 10 years ago.

The Figs Protocol.

Nonius
Founding Member
Nonius Unbound
Total Posts: 12805
Joined: Mar 2004
 
Posted: 2005-05-03 18:04
Everythng has been thought of...unfortunately....

Chiral is Tyler Durden

goldorak


Total Posts: 1091
Joined: Nov 2004
 
Posted: 2005-05-03 18:09
" Everythng has been thought of...unfortunately.... "

.... until the next revolution ! Evil Smile

If you are not living on the edge you are taking up too much space.

IAmEric
Phorgy Phynance
Banned
Total Posts: 2961
Joined: Oct 2004
 
Posted: 2005-05-03 20:20
Speaking of mind boggling...

I was just thinking of how one might use genetic algorithms to construct a population of automated "dumb" traders. Traders live and die, reproduce/mixing chromosomes, mutate, etc. You could make the universe as complex as you wanted.

What started boggling my mind was when I started thinking of bumping everything up a notch (like a 2-category). You construct a GA where the population is a bunch of different GAs. The GA rules themselves reproduce, mutate, etc. The evolution of evolution. It's a GA^2!

Ok. I'm losing it Big Smile

FDAXHunter
Founding Member

Total Posts: 8372
Joined: Mar 2004
 
Posted: 2005-05-03 20:37

Actually, that's not as easy at it sounds, as when you are mutating the "group" what will be changed withing the group?

Are the inhabitants of a group simply randomly evolved? In that case you can only pass on the mutation and so you get mostly a bit of grouping going on, but if your ranking algorithm is properly designed.. I don't see the point as the ranking can effectifly group as well.

Now, if the inhabitants are actually governed by a factory algorithm and you evolve that, well then you have something I think. But ultimately, aren't you just traversing the entire solution space in a more concentrated, more directed manner?


The Figs Protocol.

temnik


Total Posts: 204
Joined: Dec 2004
 
Posted: 2005-05-03 20:46

http://www.eweek.com/article2/0,1759,1238217,00.asp

And then the man who started MIT AI lab declares that AI is and has been dead since the 1970's...

Mon métier et mon art, c'est vivre.

James
NP High Priest

Total Posts: 2024
Joined: May 2004
 
Posted: 2005-05-03 21:33
"k ok ok guys, if you show me your yacht ("Dumb Luck II"), I'll show ya mine ("Chinese Junk")"

I plan to have a corporate jet called the "Indefensible II" since Warren owned the "Indefensible" for a time.

Until he bought a jet fractional ownership company.

Hmmmmm. Some of this crappy ABS is secured with airplanes. Now, if I just get enough paper with a high enough p= default, I'd end up likely owning a plane. If I own the paper through a shell company that is in the airplane leasing business, but is availble to company execs when the airplane is not leased, all I have to do is charge enough when it is leased to cover what I want to do with it when it isn't leased. Which is basically fly my friends and me from NP around to see each other, cooking up biz deals. Which also is expensible (deductible).

I'm converging on a model for a privately controlled jet plane that nevertheless is paid for by others.

Prior to the publication of the Black-Scholes model in 1973, the quest for a valuation formula that would describe option prices reflected one of the most elusive goals in financial economics. Though much work was done in the 1960s, many of the insights and techniques used to solve the problem were presented or anticipated at the beginning of the twentieth century by Louis Bachelier, an obscure French mathematician. These innovations include the first graphical representation of option pricing, a mathematical description of stock prices utilizing Brownian motion and anticipating the efficient market hypothesis, and the first formal option pricing formula.

IAmEric
Phorgy Phynance
Banned
Total Posts: 2961
Joined: Oct 2004
 
Posted: 2005-05-03 21:42
Now, if the inhabitants are actually governed by a factory algorithm and you evolve that, well then you have something I think.


Hi FDAXHunter,

I'm just thinking aloud, but hey, if you have ideas to make the thing more interesting, I'm open to suggestions Smiley

I like temnik's article. It's definitely relevant to the evolution^2 idea. In fact, casting it as an "emotion machine" makes sense. In a typical GA, you evolve the population, but once the environment becomes unfriendly, you can have a kind of extinction. But why not, store the "survivors" (similar to learned "emotional states") for later use. Just because one agent gets killed off when things go awry, there is no reason not to bring it back when the universe retraces back to a familiar state. You don't need to relearn how to be "angry" after a period of being "sad" or "happy" if you've already learned how to be "angry" once before. The rules of the game should evolve as you go from one "emotional state" to the next.

I think this has a chance of being quantified into relevant chromosomes.


kr
Founding Member
NP Raider
Total Posts: 3561
Joined: Apr 2004
 
Posted: 2005-05-03 22:35
well james, that just takes you back to Buffett, right?

probably this reference is too obscure for most people here to figure out, but you know these things

my bank got pwnd

apine


Total Posts: 1009
Joined: Jun 2004
 
Posted: 2005-05-04 02:36
temnik, that was a very interesting article. i like the idea of flashing through 10 emotions to shake up the thought process.

Too many people make decisions based on outcomes rather than process. -- Paul DePodesta

NIP247


Total Posts: 554
Joined: Feb 2005
 
Posted: 2005-05-04 08:41

kr >> about that sector-specific ups/downs etc. Just noticed that Brummer is launching a new hf called RacePoint that will be managed by their ex-head of research, ex-jp morgan guy Douglas Cliggott, trading mainly ETF's (about 65) using 400 (!) "fundamental" industry factors like "employment in the banking sector", "pricedevelopment in telecom services", "capacity utilization in semis" etc (freely translated from Swedish)- I thought all those points were lagging...

 

 


On your straddle, done on the puts, working the calls...

M1kz


Total Posts: 121
Joined: Sep 2004
 
Posted: 2005-05-04 08:51
NIP247: Where did you read/hear that Brummer is starting a new hf called RacePoint? A quick Google search / Brummer homepage visit yields nil...

The man who lets a leader prescribe his course is a wreck being towed to the scrap heap.

NIP247


Total Posts: 554
Joined: Feb 2005
 
Posted: 2005-05-04 08:58

client snail mail...

 

 


On your straddle, done on the puts, working the calls...

M1kz


Total Posts: 121
Joined: Sep 2004
 
Posted: 2005-05-04 10:44
Ok, I heard throught the grapevine that B&P got bored with Douglas bear attitude through the not-so-bear 2003-2004 that made Zenit return -2.83 % and -1.54% respectively and that he is setting up his own fund (not in collaboration with B&P) now.

The man who lets a leader prescribe his course is a wreck being towed to the scrap heap.

NIP247


Total Posts: 554
Joined: Feb 2005
 
Posted: 2005-05-04 11:03

Brummer & Partners will own 49% and Cliggott & Van Leuven will own 51%, i.e. a typical Brummer deal- B&P Fund Services will do the admin in Bermuda, and something tells me that their FoF Helios will soon include quarterly investments in the fund...

 

 

 


On your straddle, done on the puts, working the calls...

kr
Founding Member
NP Raider
Total Posts: 3561
Joined: Apr 2004
 
Posted: 2005-05-04 13:31

I was thinking more of a technical strategy... i.e. looking for trends (not even trends really, just short-term directonal prediction) in multidimensional sector-space rather than just up vs. down.  Doesn't even really have to be sectors, just any particular split.  For instance, you should be able to find a flight-to-quality portfolio that is still just long equities, even though ftq is traditionally short equities long bonds.  I'd guess it would be bluechips vs. smallcaps for instance.  This kind of relation is interesting to me.

can't remember now - maggot was 'economic strategy', something like that? 


my bank got pwnd

temnik


Total Posts: 204
Joined: Dec 2004
 
Posted: 2005-05-05 02:23
kr - I've got a strategy just for you. Big Smile

http://www.j-chart.com


The philosophy of J-Chart:

One day when I was sitting at the beach of Hawaii looking at the sunset, I suddenly realized that the concept of using fixed time-sequence intervals was the major "fraud" of thinking while doing market analysis. Time is an irreversible vector which has no meaning without events. There is actually no image point or cycle if nothing significant happens. People use one hour, one day, one week or even one month to display and observe the market's movement which is a wrong idea. For example, we would not remember 911 if there had not been a terrorist attack. It would just be another day in the calendar. Against this background, here are the conclusions:

Only the "event" has its image.
The event happened for the purpose of obtaining equilibrium.
The outcome of equilibrium is chaos and between them is the endless cycle.
Price is the event.


How does J-Chart forecast price?

During the process of obtaining equilibrium, some price will have its image, which is the boundary of balance.
This corresponding phenomenon of market price is the same as the particle and anti particle in physics, or day and night in the solar system, or even birth and death in human life.

J-Chart uses the 1st Point of Origin with the 2nd Point of Balance to figure out the 3rd Image Price i



Mon métier et mon art, c'est vivre.

Baltazar


Total Posts: 1782
Joined: Jul 2004
 
Posted: 2005-05-09 15:53
regarding GA

I think FDAX is rigth, you basically search the entire parameter space in some 'clever' way.
twopoints: if you write your problem as a maximization (let'say of the total wealth generated by your population of traders), what tells you that the breeding of two good strategies will give a better one? (or two traders)

implicitely GA hopes that the solution function is pretty smooth, if not it boosts the mutation rate (to create new strategies/trader from nil) but it is just an 'unclever' way of walking the solution space no?

and this assumes that the solution is stationary, otherwise you migth want to keep old not-so-good traders as, when the time'll change, they may turn to be good ones.


Qui fait le malin tombe dans le ravin

Nonius
Founding Member
Nonius Unbound
Total Posts: 12805
Joined: Mar 2004
 
Posted: 2005-05-09 16:02

my latest idea.....spawned by talkin with fund managers....

quant driven (via machines that combine random with intelligent searching), fundamental negative gate follow up, trade construction optimization.

so, on the middle part, the array of futuristic machines dump "viable opportunities" into a db on a certain frequent basis...then, a small team of relatively intelligent fundamental monkeys can say NO, but they cant introduce other strategies/positions, they are there simply to provide some color on why something maybe out of whack and why it should be dismissed...then, the optimization computes the best instruments to put on in order to maximize SPLUG, which, in turn is some risk reward measure...and voila...done.


Chiral is Tyler Durden

NIP247


Total Posts: 554
Joined: Feb 2005
 
Posted: 2005-05-09 16:04

otherwise you migth want to keep old not-so-good traders as, when the time'll change, they may turn to be good ones.

a little like those COBOL programmers that were taken out from the closet and cleared from dust a few months before y2k? Smiley

 

 

 


On your straddle, done on the puts, working the calls...
Previous Thread :: Next Thread 
Page 6 of 7Goto to page: 1, 2, 3 ... 5, 6, 7 Prev Next