Join Early Retirement Today
Reply
 
Thread Tools Search this Thread Display Modes
Re: 10M USD Reference Portfolio
Old 03-20-2006, 12:21 PM   #41
Thinks s/he gets paid by the post
 
Join Date: Jun 2005
Posts: 4,005
Re: 10M USD Reference Portfolio

I, for one, enjoy checking in on it from time to time. I am skeptical that the automated trading system will be able to produce a higher risk-adjusted return than any old slice and dice buy-n-hold type of portfolio over a meaningful length of time. But it is an interesting experiment nonetheless.

Advice to those who find this thread annoying or a waste of time when it gets updated once a week: DON'T CLICK ON THE LINK!!!! Or if you click it by accident, and decide "Oh, this is that stupid thread about stupid pointless stuff", then hit "back" on your browser, and continue surfing.

__________________

__________________
justin is offline   Reply With Quote
Join the #1 Early Retirement and Financial Independence Forum Today - It's Totally Free!

Are you planning to be financially independent as early as possible so you can live life on your own terms? Discuss successful investing strategies, asset allocation models, tax strategies and other related topics in our online forum community. Our members range from young folks just starting their journey to financial independence, military retirees and even multimillionaires. No matter where you fit in you'll find that Early-Retirement.org is a great community to join. Best of all it's totally FREE!

You are currently viewing our boards as a guest so you have limited access to our community. Please take the time to register and you will gain a lot of great new features including; the ability to participate in discussions, network with our members, see fewer ads, upload photographs, create a retirement blog, send private messages and so much, much more!

Re: 10M USD Reference Portfolio
Old 03-20-2006, 12:26 PM   #42
Give me a museum and I'll fill it. (Picasso)
Give me a forum ...
cute fuzzy bunny's Avatar
 
Join Date: Dec 2003
Location: Losing my whump
Posts: 22,697
Re: 10M USD Reference Portfolio

I hadnt read the thread in a while as I had put it in the bucket of "another thread about automated trading that will be posted to until the system craters because those dont work over long periods of time".

I did wish from time to time that I could 'unsubscribe' from the thread and not see any 'new' updates when I looked at 'see threads with new stuff in them', but as justin says, its pretty easy to ignore.

Heck, if *I* wont complain about it, its pretty low level.

However the petty squabbling about it is kinda fun to read, so uhhh...thanks for that...I think
__________________

__________________
Be fearful when others are greedy, and greedy when others are fearful. Just another form of "buy low, sell high" for those who have trouble with things. This rule is not universal. Do not buy a 1973 Pinto because everyone else is afraid of it.
cute fuzzy bunny is offline   Reply With Quote
Re: 10M USD Reference Portfolio
Old 03-20-2006, 12:57 PM   #43
Recycles dryer sheets
 
Join Date: May 2005
Posts: 228
Re: 10M USD Reference Portfolio

I had a personal exchange with Nords about what the system does, but not going into any details as it is not what is requested or addressed by the two previous posters and staying on the surface at the psychological level, it's funny to see how people are generally skeptical about trading automation.

I guess the primary frustration comes from the fact that nobody can envisage that a "stupid machine" could do better something that we humans have so many difficulties achieving. So people state that it blows over time (my experience is that it happens but to human traders more than to machines !), or that it does not work, or... etc. Everybody knows that trading is difficult so how could a machine be more clever than us It is both our failures (let me know of anyone who has traded and has never failed) and our competitive positioning with the system that makes us defiant.

But, acknowledge the truth. Experiments have been made with groups and PhDs do not succeed better than others (I know that for sure !). Successful trading has nothing to do with intelligence per se, and there are many psychological factors that make most human being fail with trading (many books address the issue). Many traders when you tell them "but the heck why did you do that !" say "yes I know, I knew I should not have done it, but it was compulsive, I did it !", kind of self punitive.

So the rules are well known. They can be put black on white. Not blowing an account on margin is just an equation, etc. A specification can be written of a system which does not infringe the rules that he knows it has to follow. And though not being the best trader on earth, consistency over time makes that automated trader beat most (99% ?) human traders over long period of time as it is - if manually done - a lot of work everyday, and we all have good and bad days, which the machine has not !

It's funny to see computers and AI programmes embarked in sophisticated systems and controlling many of our daily lives experiences, like trusting the aeroplane navigation system, the surgeon electronic devices, etc. We are all confident, but as soon as it is trading, everybody expects the system to crash !!!!

Planes take off, fly and land on their own if necessary. And nobody would expect them to crash ! Trading is not that complicated but is so much emotionally loaded as we talk about money !!!

Trading automation is like printing notes. Cannot work !

Well two cents to foster a bit ideas.
__________________
gnoti seauton
poyet is offline   Reply With Quote
Re: 10M USD Reference Portfolio
Old 03-20-2006, 01:19 PM   #44
Give me a museum and I'll fill it. (Picasso)
Give me a forum ...
cute fuzzy bunny's Avatar
 
Join Date: Dec 2003
Location: Losing my whump
Posts: 22,697
Re: 10M USD Reference Portfolio

Unfortunately short to intermediate term trading is influenced far more by individual and group psychology than by any numbers or measurements. Certainly the numbers and measurements influence the groupthink that runs the markets day to day. But those reactions are unpredictable.

Over long periods of time, the markets do march to the tune of earnings and growth. But modest or frequent trading based on trying to determine or predict the results of individual investors due to certain levels or events simply doesnt work or doesnt work over long periods of time unless your name is Buffett and you buy whole companies at a pop.

At least not up until now. Maybe this is different. Maybe you can divine the directions of the trading public.

Do keep reporting in, and be sure to give the full post-mortem if it crashes and burns.

By the way, what algorithm divines the upside down tendency for people to want to buy a stock when the market is up and sell it when the market is down? Or sell when numbers are good and buy when they're bad. Or the reverse for no apparent reason?
__________________
Be fearful when others are greedy, and greedy when others are fearful. Just another form of "buy low, sell high" for those who have trouble with things. This rule is not universal. Do not buy a 1973 Pinto because everyone else is afraid of it.
cute fuzzy bunny is offline   Reply With Quote
Re: 10M USD Reference Portfolio
Old 03-20-2006, 01:45 PM   #45
Recycles dryer sheets
 
Join Date: May 2005
Posts: 228
Re: 10M USD Reference Portfolio

Quote:
Originally Posted by Cute 'n' Fuzzy Bunny

At least not up until now. Maybe this is different. Maybe you can divine the directions of the trading public.

Do keep reporting in, and be sure to give the full post-mortem if it crashes and burns.

By the way, what algorithm divines the upside down tendency for people to want to buy a stock when the market is up and sell it when the market is down? Or sell when numbers are good and buy when they're bad. Or the reverse for no apparent reason?
The tape is all and the system does not divine anything. In fact, a chart (or its internal representation) of a stock, index, whatever is though of by the system in physical terms, say as if we were making some kind of target motion analysis. Of course, we need to smooth the trajectory in different ways and at different time scale, before computing on these representations rate of change, derivates, etc. But in the end, it either goes up or down ! And the system tries to be as simple as that. If the index chosen (either custom to the trading universe or standard like NDX) goes up and most stocks in the trading universe do so, the system has a long stance and reversely. Then, if long and if the market accelerates, it tries to find stocks which motion is the fastest, etc. Then it goes to money management and risk to check how much it can go shopping for, etc. Given the speed of the trajectory (and volatility) the system will protect by slower or faster stops... and the NAV of the account is reflexively looked at as a security (and its trajectory monitored) with stops against the NAV...

An extremely short summary...
__________________
gnoti seauton
poyet is offline   Reply With Quote
Re: 10M USD Reference Portfolio
Old 03-20-2006, 01:48 PM   #46
Give me a museum and I'll fill it. (Picasso)
Give me a forum ...
cute fuzzy bunny's Avatar
 
Join Date: Dec 2003
Location: Losing my whump
Posts: 22,697
Re: 10M USD Reference Portfolio

Ah...in short laymans terms, momentum trading.

Which works great until the momentum stops. Or (I believe it was Alec's term) the markets get drunk and do stupid things.
__________________
Be fearful when others are greedy, and greedy when others are fearful. Just another form of "buy low, sell high" for those who have trouble with things. This rule is not universal. Do not buy a 1973 Pinto because everyone else is afraid of it.
cute fuzzy bunny is offline   Reply With Quote
Re: 10M USD Reference Portfolio
Old 03-20-2006, 07:27 PM   #47
Moderator Emeritus
Nords's Avatar
 
Join Date: Dec 2002
Location: Oahu
Posts: 26,617
Re: 10M USD Reference Portfolio

Quote:
Originally Posted by Cute 'n' Fuzzy Bunny
Ah...in short laymans terms, momentum trading.
Well, the computer's too smart to read analyst's reports. What else could it work with?

The biggest flaw in momentum trading is... the human doing the trading.

Besides Poyet is the only other guy I've met on this board who knows LISP.
__________________
*
*

The book written on E-R.org, "The Military Guide to Financial Independence and Retirement", on sale now! For more info see "About Me" in my profile.
I don't spend much time here anymore, so please send me a PM. Thanks.
Nords is offline   Reply With Quote
Re: 10M USD Reference Portfolio
Old 03-20-2006, 07:48 PM   #48
Thinks s/he gets paid by the post
Spanky's Avatar
 
Join Date: Dec 2004
Location: Minneapolis
Posts: 4,046
Re: 10M USD Reference Portfolio

Quote:
Besides Poyet is the only other guy I've met on this board who knows LISP.
I used to develop expert systems in Lisp (back in the 80s).
__________________
May we live in peace and harmony and be free from all human sufferings.
Spanky is offline   Reply With Quote
Re: 10M USD Reference Portfolio
Old 03-20-2006, 07:54 PM   #49
Give me a museum and I'll fill it. (Picasso)
Give me a forum ...
cute fuzzy bunny's Avatar
 
Join Date: Dec 2003
Location: Losing my whump
Posts: 22,697
Re: 10M USD Reference Portfolio

I used to sell Alpha Microsystems running AMOS and LISP back in the 70's...
__________________
Be fearful when others are greedy, and greedy when others are fearful. Just another form of "buy low, sell high" for those who have trouble with things. This rule is not universal. Do not buy a 1973 Pinto because everyone else is afraid of it.
cute fuzzy bunny is offline   Reply With Quote
Re: 10M USD Reference Portfolio
Old 03-20-2006, 08:16 PM   #50
Moderator Emeritus
Nords's Avatar
 
Join Date: Dec 2002
Location: Oahu
Posts: 26,617
Re: 10M USD Reference Portfolio

Quote:
Originally Posted by Spanky
I used to develop expert systems in Lisp (back in the 80s).
Ever notice that no one ever develops incompetent systems? Not intentionally, anyway...

Quote:
Originally Posted by Cute 'n' Fuzzy Bunny
I used to sell Alpha Microsystems running AMOS and LISP back in the 70's...
I guess we should be thankful that the military didn't adopt it as a MILSPEC.
__________________
*
*

The book written on E-R.org, "The Military Guide to Financial Independence and Retirement", on sale now! For more info see "About Me" in my profile.
I don't spend much time here anymore, so please send me a PM. Thanks.
Nords is offline   Reply With Quote
Re: 10M USD Reference Portfolio
Old 03-21-2006, 02:42 AM   #51
Recycles dryer sheets
 
Join Date: May 2005
Posts: 228
Re: 10M USD Reference Portfolio

Quote:
Originally Posted by Cute 'n' Fuzzy Bunny
Ah...in short laymans terms, momentum trading.

Which works great until the momentum stops.
I'll try to address some points then I'll be back in the post to Nord's comments.

Well I did not use special terms to make the system look complex or more intelligent or competent. I used some analogies to try to make people grasp what the system does. It mainly uses TA (or derived) data but it could also use FA data (P/E whatever, anything which is represented with numerals is just natural to a computer not even talking of symbolic programming). And as most (if not all TA and FA) data are numerals we have a field very computer processable here !

Back to the wording:

I would refer to *trend trading* as the general category in which it falls (long/short and leveraged), adding that yes momentum trading is also correct if you wish, in the phases where either the long or the short side of the market accelerates and that we leverage on positions.

And yes until the trend stops. It is correct to say that performances are kind of flat when there is no trend (and that was a lot of work to manage to make the system smart enough not to loose money in these conditions). But if anyone teaches me of a good trading method working on flat markets (or without clear tendencies and not with exotic straddles) I'm ready to learn and to try to program it. No joke. But, we face the harsh reality, if there is no trend in one side or the other, there is no perspective of leverage and hypersized returns. If we handle trading accounts (with more risk than investments) it is to generate big returns when the markets trends, though they kind of be flat if the market goes nowhere. It's just a matter not to be slowly eroded with a negative return until the trend appears !

Then the more important part of the system is not "understanding" the market which is quite easy. Come on, it's not even 3D trajectory analysis, we're in 2D with a bit of volat ! No, what's more difficult is risk management (at various levels from the position, to the portfolio as w whole) and money management. Again, I've invented nothing. I've just put in music what I had learnt and tested myself.

The great strenght of the system is to have eliminated the human being, the weak part of all trading strategies. It's always the human who goes wrong after some time, not complying with the rules for whatever psychological justifications. I've observed myself trading. I knew that I should have done that, but sometimes I find incredible reasons to justifiy something I should not have done and further the loss that came with !!!

That's for Nords: of all the incompetent expert systems I've designed (and to be honest quite some !), I think this one is the one which has most credentials. Its backtesting has gone over thousands of runs (each run over decades) and millions of trades. It has been running on the markets daily for two years and I have had no deviation between real accounts (where trades are placed) and the positions as monitored by the system.

Again it is just a well trained and well disciplined automat. But do we need more for successful trading ? I've not said for genius trading.

What I'm not at ease with is genius trading. You always find someone who does completely unreasonable things (in terms of trading practice) and either fails (like Nick Leeson) or succeeds... But the difference between the two is none. They were both unreasonable, but one was lucky enough to be right. Would Nick have traded the Nikkei the other way round (as stupidly) he would have become a genius and not a fool. In fact he would still have traded badly !

As for the LISP part, I'm glad we've got still some LISP (men and women ?) on the board. Unfortunately (though for some numerical processing with high precision and there are quite a lot with trading LISP is not great) the system is not written in LISP or PROLOG (PROgrammation LOGique). Bill has won over me and it is in Basic and C# in .NET. But in the end, it does not matter, it is still a very sophisticated programme.

Its weakness is there. Sophistication and maintenance. Updates the databases at night connecting to reuters. Starts computing at 4 am. Works two hours on a P4 and sends emails to me so that when markets open (in europe) everything is OK. And a lot of software. So OS migration, versoning of underlying tools all that put a burden on me and I'm sometimes worry to find the energy to keep it working unless I'll find more people interested in using it.

But I'm not going to troll here !!!
__________________
gnoti seauton
poyet is offline   Reply With Quote
Re: 10M USD Reference Portfolio
Old 03-21-2006, 01:34 PM   #52
Moderator Emeritus
Nords's Avatar
 
Join Date: Dec 2002
Location: Oahu
Posts: 26,617
Re: 10M USD Reference Portfolio

Poyet, short of putting your entire portfolio into the computer's hands and posting scanned copies of your tax returns, you're doing this in the best way it can be done.

And you're retaining creative control. It's better than being a quant at a brokerage house.

__________________
*
*

The book written on E-R.org, "The Military Guide to Financial Independence and Retirement", on sale now! For more info see "About Me" in my profile.
I don't spend much time here anymore, so please send me a PM. Thanks.
Nords is offline   Reply With Quote
Re: 10M USD Reference Portfolio
Old 08-06-2006, 11:54 PM   #53
Moderator Emeritus
Nords's Avatar
 
Join Date: Dec 2002
Location: Oahu
Posts: 26,617
Re: 10M USD Reference Portfolio

Poyet, I don't know if you're a Business Week subscriber, but there's a short article on their website about kaburobos:

"Now a pair of Japanese companies -- tech startup Trade Science and online securities firm Monex Beans Holding -- has teamed up to provide robot fund managers for individual investors in Japan. If all goes as planned, next year investors will be able to choose from among 10 so-called Kaburobos, each offering a different investment strategy (tested using data from the past 15 years to see how it would have performed against market indexes). Using algorithmic programs, the robots will make the day-to-day decisions about what stocks and indexes to trade -- and when."

1991-2005 backtesting... it'd be good to see more data than that. Have you seen any other research on this?
__________________
*
*

The book written on E-R.org, "The Military Guide to Financial Independence and Retirement", on sale now! For more info see "About Me" in my profile.
I don't spend much time here anymore, so please send me a PM. Thanks.
Nords is offline   Reply With Quote
Re: 10M USD Reference Portfolio
Old 08-07-2006, 12:36 AM   #54
Full time employment: Posting here.
 
Join Date: Mar 2005
Posts: 699
Re: 10M USD Reference Portfolio

Interesting. Looks like they are also holding robotic trading contests to help pick good algorithms. Maybe Poyet could even enter -- they don't require you to be in-country. It looks like they are considering having successful authors be able to earn some fraction of the profits that their robot earns in real life as a fund manager.


http://www.kaburobo.jp/
http://www.kaburobo.net/
(All documentation seems to be in Japanese at this point, but I bet you could send them an e-mail if English if needed.)
__________________
bpp is offline   Reply With Quote
Re: 10M USD Reference Portfolio
Old 08-14-2006, 12:31 PM   #55
Recycles dryer sheets
 
Join Date: May 2005
Posts: 228
Re: 10M USD Reference Portfolio

Thanks Nords and bpp for looking into this interesting subject.
I'm pretty confident that automated trading will generalize in the future.
Funny that the Japanese enter the game.

It's difficult to backtest on more than say 15 years or so for various reasons. Indexes can be traded in backtest conditions for much longer and I've traded the dow since 1929 for instance. This is the piece of cake part of the game...

Usually one maintains an updated database, i.e. a database of stocks which are listed today with their correct tickers so as to implement operations of the running system. In my case, all these stocks are stored in various folders according to their GICS sector and the database undergo a regular update and cleaning every month (updating tickers, unlisting companies having disappeared because of mergers or bankruptcies, etc.) . Stocks removed are parked in the cimetery as "unlisted".

When you run the backtest to check the consistency of the trading strategies implemented you perform that on the latest version of the database. This means that you trade current stocks in the past as long as they are available. But trading stocks in the past which have been unlisted is exacty what one should do if he/she wished to push further back the horizon of the backtest exercise but it is more difficult. Because for a "current database" most stocks tend to disappear beyond the 15 years horizon. Say more than 70% which then makes the run irrelevant.

To perform on longer time scales one would need a dynamic database and a very flexible system. Dynamic as stocks would enter and leave the game, as tickers would change all the time, as the system would update itself as it runs, considering new stocks, acknowledging that some are gone and that many have their tickers changed many times over the time scale we talk about., etc...

Of course this is what happens when we run daily the operating system. But the update and incremental adjustments are one way. We move towards the future by staying current with no capacity to re-create what was the past in its exact organisation.

I do not say that this is not doable. It is an entire project. And the question is that even though I'm sure we'll learn something from doing it the reward / effort question is raised !

Finally a comment. I've recently noticed that some users / clients working with my system and having made significant money with it, kind of completely loose their common sense. And that's a big danger as they tend to interpret the system's action in total decorrelation of the real market. Their understanding of the market is through the system's actions !

My two cents for today.
__________________
gnoti seauton
poyet is offline   Reply With Quote
Re: 10M USD Reference Portfolio
Old 08-14-2006, 12:43 PM   #56
Moderator Emeritus
Nords's Avatar
 
Join Date: Dec 2002
Location: Oahu
Posts: 26,617
Re: 10M USD Reference Portfolio

Quote:
Originally Posted by poyet
Usually one maintains an updated database, i.e. a database of stocks which are listed today with their correct tickers so as to implement operations of the running system. In my case, all these stocks are stored in various folders according to their GICS sector and the database undergo a regular update and cleaning every month (updating tickers, unlisting companies having disappeared because of mergers or bankruptcies, etc.) . Stocks removed are parked in the cimetery as "unlisted".
When you run the backtest to check the consistency of the trading strategies implemented you perform that on the latest version of the database. This means that you trade current stocks in the past as long as they are available. But trading stocks in the past which have been unlisted is exacty what one should do if he/she wished to push further back the horizon of the backtest exercise but it is more difficult. Because for a "current database" most stocks tend to disappear beyond the 15 years horizon. Say more than 70% which then makes the run irrelevant.
I'd think that survivor bias would be exactly the problem that would derail any long-term conclusions or automated trading systems. Look how much the S&P500 has changed over the last 25 years, especially the 1990s, let alone the DOW over the last century.

You say that it'd be a hard slog to put together an accurate database (especially when dividends are correctly incorporated) but I would think that it'd be a form of portfolio insurance from unpleasant surprises...
__________________
*
*

The book written on E-R.org, "The Military Guide to Financial Independence and Retirement", on sale now! For more info see "About Me" in my profile.
I don't spend much time here anymore, so please send me a PM. Thanks.
Nords is offline   Reply With Quote
Re: 10M USD Reference Portfolio
Old 08-15-2006, 02:48 AM   #57
Recycles dryer sheets
 
Join Date: May 2005
Posts: 228
Re: 10M USD Reference Portfolio

Quote:
Originally Posted by Nords
I'd think that survivor bias would be exactly the problem that would derail any long-term conclusions or automated trading systems. Look how much the S&P500 has changed over the last 25 years, especially the 1990s, let alone the DOW over the last century.

You say that it'd be a hard slog to put together an accurate database (especially when dividends are correctly incorporated) but I would think that it'd be a form of portfolio insurance from unpleasant surprises...
If it were "investment strategies" I would perhaps concur, but as far as trading systems are concerned I disagree. My best trades, LONG (e.g. CYTO, HYSQ, DTSX, etc. all well over 1000%) and SHORT were all placed on stocks that have been unlisted in the meantime and which are therefore not "accessible" to the kind of backtest performed. These are missing opportunities for a good trading system. Though it introduces a biais, I would tend to think of it as minimizing the backtested performance for a trading system.


__________________

__________________
gnoti seauton
poyet is offline   Reply With Quote
Reply


Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)
 
Thread Tools Search this Thread
Search this Thread:

Advanced Search
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off


Similar Threads
Thread Thread Starter Forum Replies Last Post
SWR - Amortize your portfolio RgrGd FIRE and Money 5 03-28-2007 11:29 AM
High Dividend Portfolio vs Portfolio Diversified Across Asset Classes ScaredtoQuit FIRE and Money 36 01-12-2007 01:24 PM
Minimum starting portfolio to achieve success Dorus FIRECalc support 1 08-10-2006 05:03 AM
Conservative distribution phase portfolio peteyperson FIRE and Money 13 03-22-2005 07:48 AM
Volatility in your Portfolio ESRBob FIRE and Money 5 03-03-2005 10:09 AM

 

 
All times are GMT -6. The time now is 09:55 PM.
 
Powered by vBulletin® Version 3.8.8 Beta 1
Copyright ©2000 - 2017, vBulletin Solutions, Inc.