Print
The business of natural selection
by EurekAlert
Thanks to Richard Prins for the link.
http://www.eurekalert.org/pub_releases/2008-03/ip-tbo031408.php
The business of natural selection
Could a mathematical model predict the demise of a business?
Many gamblers claim to have a "system", whether they're shooting craps, backing horses, or punting on the stock market. Now, researchers in Taiwan have devised an approach to spotting when a company is likely to fail based on the principles of natural selection. They report details of their system in a forthcoming issue of the International Journal of Electronic Finance.
Ping-Chen Lin of the National Kaohsiung University of Applied Sciences in Kaohsiung and Jiah-Shing Chen of the National Central University, Jhongli, in Taiwan, explain how the financial status of any company can be of interest not only to its owners and employees but to a range of creditors, stockholders, banks, and individual investors. However, there are so many changing and interconnected factors that can lead to success or failure that it is usually considered an impossible task to predict whether a company will fail.
The researchers have now borrowed some of the principles of evolutionary biology to come up with a computer algorithm to make such predictions possible. They feed different variables, such as earnings per share, liabilities and net income, into their genetic-based hybrid algorithm, which assigns a weighting to each value. The output of the algorithm is a new set of variables that are then selected for how well they fit the next set of financial results from the company. Those that fail are discarded, or reduced in weight, and those that match the actual results more closely are fed back into the algorithm for the next round.
By using actual data from successful and failed companies and feeding this into the algorithm the researchers build up the fittest set of variables and weightings. This allows the algorithm to evolve so that it can then predict the financial future of any given company based on current income and expenditure, and tax obligations.
The team has blind tested the predictive power of their system on several companies successfully. "Our experimental results show that this hybrid approach obtains better prediction performance than when using a single approach effectively," the researchers say.
http://www.eurekalert.org/pub_releases/2008-03/ip-tbo031408.php
The business of natural selection
Could a mathematical model predict the demise of a business?
Many gamblers claim to have a "system", whether they're shooting craps, backing horses, or punting on the stock market. Now, researchers in Taiwan have devised an approach to spotting when a company is likely to fail based on the principles of natural selection. They report details of their system in a forthcoming issue of the International Journal of Electronic Finance.
Ping-Chen Lin of the National Kaohsiung University of Applied Sciences in Kaohsiung and Jiah-Shing Chen of the National Central University, Jhongli, in Taiwan, explain how the financial status of any company can be of interest not only to its owners and employees but to a range of creditors, stockholders, banks, and individual investors. However, there are so many changing and interconnected factors that can lead to success or failure that it is usually considered an impossible task to predict whether a company will fail.
The researchers have now borrowed some of the principles of evolutionary biology to come up with a computer algorithm to make such predictions possible. They feed different variables, such as earnings per share, liabilities and net income, into their genetic-based hybrid algorithm, which assigns a weighting to each value. The output of the algorithm is a new set of variables that are then selected for how well they fit the next set of financial results from the company. Those that fail are discarded, or reduced in weight, and those that match the actual results more closely are fed back into the algorithm for the next round.
By using actual data from successful and failed companies and feeding this into the algorithm the researchers build up the fittest set of variables and weightings. This allows the algorithm to evolve so that it can then predict the financial future of any given company based on current income and expenditure, and tax obligations.
The team has blind tested the predictive power of their system on several companies successfully. "Our experimental results show that this hybrid approach obtains better prediction performance than when using a single approach effectively," the researchers say.
Add to del.icio.us
Posted: Saturday, March 15, 2008 | PermalinkComments 1 - 25 of 25 |
Reload Comments | Back to Top | Page Numbers2. Comment #144155 by Mitchell Gilks on March 15, 2008 at 8:10 am
This seems like it could easily turn into a self-fulfilling prophecy.
I'm skepticial, but then, I know diddly skwat about business.
Other Comments by Mitchell Gilks
3. Comment #144159 by aporeticus on March 15, 2008 at 8:23 am
I'm skeptical, too. External factors affect businesses tremendously, but I don't know of those are inputs. Other Comments by aporeticus
4. Comment #144162 by Bonzai on March 15, 2008 at 8:27 am
It is yet more hype.At one point it was using neural networks to pick stocks, which died a quiet death.
A few years ago they were talking about using chaos theory to do business,.I heard business professors on TV and obviously these clowns knew nothing about chaos theory beyond a few sound bites. Maybe they meant creating a lot of confusions so that they can make money by pulling a fast one, I have no idea.
I actually hate business type with an even bigger passion than I do religious fundamentalists. At least the fundies have some genuine human emotions, I can't say the same about the talking suits. The other day I was reading a book in a coffee shop, this woman behind me began to yack away on her cell phone. She was going on and on about "synergy", "the butterfly effect" and so on, for almost half an hour she was babbling gibberish in business lingo peppered with bastardized scientific jargons. I couldn't even be sure the conversation made sense to her, certainly not to me. I felt like telling her to STFU and pouring coffee all over her notebook. But then I decided to leave before I lost it and got arrested for assault.
Other Comments by Bonzai
5. Comment #144164 by nother person on March 15, 2008 at 8:30 am
GIGOOther Comments by nother person
6. Comment #144165 by Steve Zara on March 15, 2008 at 8:30 am
It is yet more hype.
As always. Everyone wants to be the first to discover how to use Azimov's fictional "psychohistory"!
Other Comments by Steve Zara
7. Comment #144168 by Logicel on March 15, 2008 at 8:36 am
Can this system account for such variables as the company's in question competitor's well being/decline, Lady luck (ex., an unexpected oil, or other commodity find?), etc.?I am a discretionary trader, and I spit in these people's faces.
Other Comments by Logicel
8. Comment #144179 by Dr Benway on March 15, 2008 at 9:13 am
"Fitness" isn't merely a function of an organism, but of the dynamic relationship between organism and environment. To comment upon fitness, these models must assume a fairly static future environment.
Heh. How likely is that?
When an article touts a new model sounding sciency and like a big breakthrough but omits any mention of the model's limitations, it's usually crap.
What amazes me: spammers sending out hot stock tips actually provoke enough people to invest to drive up stock prices a couple of points, so they can sell at a profit.
Other Comments by Dr Benway
9. Comment #144185 by 5keptical on March 15, 2008 at 9:31 am
A friend of mine was using neural-networks (or variations thereof) over 10 years ago and was quite successful with it (became a vice-president in a financial firm due to his system's performance compared to others at the time).The use hasn't died a quiet death - the specifics of their modeling software are trade secrets - the more information one gives out, the more others will know how you will react to the market, so the techniques are not publicized.
All the players use prediction software (though now it's probably moved on to Bayesian networks) and the models have to be constantly refined to take into account that others are using similar techniques - i.e. it must constantly evolve in a changing environment.
Other Comments by 5keptical
10. Comment #144188 by SPS on March 15, 2008 at 9:35 am
Survival of the richest?Other Comments by SPS
11. Comment #144225 by andrewtott on March 15, 2008 at 12:08 pm
Business systems are chaotic (especially public markets!!)- the past is not necessarily a guide to the future.Other Comments by andrewtott
12. Comment #144291 by mixmastergaz on March 15, 2008 at 2:54 pm
Mathematics is a language I can barely understand, but as I read this I could hear the booming voice of 'Deep Thought' declaring that the answer to the ultimate question of life, the universe and everything is...
is...
Other Comments by mixmastergaz
13. Comment #144325 by the_ultimate_samurai on March 15, 2008 at 3:49 pm
considering we cant use natural selection to guess the evolutionary change of a certain species (due in large part to not knowing what the environment will hold in the future) i dont see how we could predict the evolutionary change in the market using the same measure.too many variables need to be followed..and of course there is the impact that this technology would have on the market anyways. im reminded a bit of heisenbergs uncertainty principal (another scientific principal we shouldnt apply to business) the very act of trying to measure this will invariably cause a change in what we are measuring. so if all these machines spit out this stock, people buy that stock, money goes into that stock, they make some success and it confirms itself, if it says this will do bad, everyone pulls out, money goes away from this company, they lose money and return poor profits...it confirms itself again.
Other Comments by the_ultimate_samurai
14. Comment #144327 by Terminally Nerdy on March 15, 2008 at 3:52 pm
yay pyschohistory, steve! Lets have telepaths and robots, too, taiwan!Other Comments by Terminally Nerdy
15. Comment #144339 by MrPickwick on March 15, 2008 at 4:38 pm
Interesting, very interesting... btw, there's this swamp land I've got for sale in Florida... Other Comments by MrPickwick
16. Comment #144355 by Grantaire of JC on March 15, 2008 at 5:35 pm
I am a skeptic, show me proof! Show me a successful test run with 50 Fortune 100 companies and you may have my attention. Show me a wide range of success with companies from utilities to drug companies to penny stocks and then I am interested. ( Did they predict the housing market crisis in the U.S.? )Oh, MrPickwick, hold on to that swampland. In New Jersey that's prime property these days.
Other Comments by Grantaire of JC
17. Comment #144364 by lievemebe on March 15, 2008 at 6:01 pm
Like others I am skeptical. Does the algorithm include random mutation of business elements? Also, If the algorithm is successful, every business will use it and it's back to square one.Other Comments by lievemebe
18. Comment #144365 by fontor on March 15, 2008 at 6:10 pm
Come on, everyone. Tons of people use evolutionary algorithms (including me, in language research). They're incredibly useful, and I don't see why they can't be used to predict survival in this context.I did think it a bit odd that they'd try using static info as data, though. Interesting approach, but I might try blasting the 'population' with tons of random but likely events. Probably run the experiment a bunch of times, just to see if we get the same weighting outcomes.
The proof is how well the model predicts survival of actual companies in the real world. If it can, then we have yet another example of the incredible utility of evolutionary theory. And that would be cool.
Other Comments by fontor
19. Comment #144387 by robotaholic on March 15, 2008 at 10:35 pm
I don't know how effective this algorithm is but it seems reasonable that if given enough information about a macroscopic system and it's environment the future could be predicted with some measure of accuracy. The more information the better: the company, it's competitors, and the whole economy...all the information contained within our past and future light cones lol.. Other Comments by robotaholic
20. Comment #144389 by Bonzai on March 15, 2008 at 10:54 pm
I don't know how effective this algorithm is but it seems reasonable that if given enough information about a macroscopic system and it's environment the future could be predicted with some measure of accuracy
Then how come we can't even make long term weather forecasts? Weather is quite simple comparing to the market.
Nonlinearity and chaos prevent any accurate prediction for individual trajectories,--say the fate of your company,--for a complex dynamical system.
Other Comments by Bonzai
21. Comment #144392 by mmurray on March 15, 2008 at 11:29 pm
There are simpler mathematical models than this. You can subtract the company income from the company expenses and test the resulting number to see if it is positive or negative. It would be interesting to see how different the test companies were from the training set but I can't see the article on the journal site.
Michael
Other Comments by mmurray
22. Comment #144395 by Teratornis on March 16, 2008 at 12:04 am
In reply to comment #144162 by Bonzai:
The other day I was reading a book in a coffee shop, this woman behind me began to yack away on her cell phone. She was going on and on about "synergy", "the butterfly effect" and so on, for almost half an hour she was babbling gibberish in business lingo peppered with bastardized scientific jargons. I couldn't even be sure the conversation made sense to her, certainly not to me. I felt like telling her to STFU and pouring coffee all over her notebook. But then I decided to leave before I lost it and got arrested for assault.
Why get mad when you can ridicule people instead? You could have shouted "Bingo!" See:
http://en.wikipedia.org/wiki/Buzzword_bingo
Another interesting game is "Egregious thread drift." I will demonstrate. Over on Wikipedia a bunch of Muslims are trying to remove all the pictures of Muhammad:
http://en.wikipedia.org/wiki/Wikipedia:Wikipedia_Signpost/2008-02-11/Muhammad_image
They have a petition which attracted 100,000 signatures. There's also a counter-petition for the godless heathens to sign:
http://www.thepetitionsite.com/1/fight-islamic-pressure-to-censor-wikipedia
Other Comments by Teratornis
23. Comment #144396 by Teratornis on March 16, 2008 at 12:08 am
In reply to comment #144389 by Bonzai:
Then how come we can't even make long term weather forecasts? Weather is quite simple comparing to the market.
I'm not even a financier but I'm pretty sure that if someone could predict short-term behavior in the financial markets as accurately as meteorologists predict tomorrow's weather, that someone could get obscenely rich in a short time. There are lots of ways to place bets in financial markets on the short-term moves in commodity prices, currency exchange rates, etc.
I just looked up the Wikipedia article about derivatives, and I learned that financial markets let people bet on the weather, neatly tying together the disparate threads of meteorology and finance in your message.
http://en.wikipedia.org/wiki/Derivative_(finance)
Other Comments by Teratornis
24. Comment #144407 by D'Arcy on March 16, 2008 at 2:38 am
Before wealth can be bought and sold on markets it must first be created by living human labour and nature given materials. The most sophisticated technology still requires 100% human input. Machinery can save labour in the productive process, but it can't create new wealth. Machines eventually wear out and have to be maintained and replaced from time to time. The wealth created, belongs of course to the factory owners and not the producers who are paid wages or salaries.
The fact that capitalists bet on the various markets of the world on the likely economic performance of companies does not influence the total amount of real wealth in the economy. The betting merely enables the change of ownership from A to B. In practice not everyone can bet on the winner.
Other Comments by D'Arcy
25. Comment #144409 by notsobad on March 16, 2008 at 2:45 am
Michael:"There are simpler mathematical models than this. You can subtract the company income from the company expenses and test the resulting number to see if it is positive or negative."
I take it this was a joke.
Other Comments by notsobad
Over 1.5 million copies sold
Books by Richard Dawkins
The Oxford Book of Modern Science Writing
by Richard Dawkins
The Selfish Gene
by Richard Dawkins
The Blind Watchmaker
by Richard Dawkins
The Ancestor's Tale
by Richard Dawkins
Unweaving the Rainbow
by Richard Dawkins
1. Comment #144147 by Animavore on March 15, 2008 at 7:52 am
Other Comments by Animavore