Wednesday 19 March 2008

When economists make predictions, just walk away ...

crystal_ball_lg 'Mathematical economists' predict all sorts of things, and all sorts of people believe them -- but there's a reason contrarians still make money even if Bear Sterns can't: the 'mathematical economists' with their crystal balls really don't have the first clue what they're talking about.
These are the people, it should be remembered, who told the government they'd be in deficit when they're not (oops); who said for years that government surpluses would be small (when they turned out to be huge); that Kyoto would be a net benefit to New Zealand taxpayers (it's going to cost us billions) ... we're heading for recession, says Cullen, we're not, says Clark; we are says the BNZ, we're not says BERL ...   let's face it, none of them really has the first clue what's going on.
And that's just the gee-whiz crystal-ball gazers who do the sums for government!  Then there's the ones who pull down the really big salaries, the chaps who are all over the TV and radio making predictions -- which mostly consist in agreeing with what every other pundit is saying -- but it turns out with these types that they've got no more idea about what's going on than your taxi driver does, as a new study fresh from the printer makes clear.  What you see in the graph below is a chart of NZ's Trade Weighted Index for the last fourteen years -- a crucial figure on which billions of dollars are spent -- measured against the predictions of all those big-name chaps with a calculator, a big salary, and not a clue what's going on.  Take a look:
         TWI-Crystal-ball-gazing
You'll be wondering what those little fan-shaped things are that seem to bear no relation to the actual TWI?  Those suckers are the range of predictions made by most of the big names you hear cluttering up your airwaves with their soothsaying (the red line is the median forecast, the grey line is the range of their predictions, the solid black line baring little to no relation to either of these is what actually happened).  Just to stress this point again: none of these turkeys is able to predict the future any more than you could by throwing a dart at the wall.  Or to put it another way, when you hear these mathematical economists making predictions about the future, don't believe a word of it.  You might just as well read your tea leaves.
Here's the real lesson: Economics doesn't give you a crystal ball allowing you to make quantitative predictions about the future.  It doesn't give anyone that kind of crystal ball -- and anyone who says it does is either lying to you, or to themselves.  As Ludwig von Mises pointed out fifty years ago, economics is not a science capable of quantitative predictions:
People by and large know today that a boom brought forth by a policy of credit expansion and "easy money" cannot last forever and must sooner or later lead to a slump. They do not want to be taken by surprise and ruined... As they believe that economics is the art of predicting tomorrow's business conditions, they consult the economists.
"How will business be in the coming months?" asks the newspaperman when interviewing the economist. No convention of businessmen is held without the solicited presence of a professor of economics, or the head of a bank's research department, who in guarded language produces a cautiously qualified prediction about the nation's, or the world's business. Whenever and wherever a businessman catches sight of an economist, he tries to sound him out about the future state of the market.
But we've already seen that the country's highest-earning economists have no idea, do they.
Economics can only tell us that a boom engendered by credit expansion will not last. It cannot tell us after what amount of credit expansion the slump will start or when this event will occur. All that economists and other people say about these quantitative and calendar problems partakes of neither economics nor any other science.
Economics predicts the outcome of definite modes of conduct, in our case, of a policy of credit expansion. But this prediction is qualitative only. Economic prediction can never disclose anything about the quantitative relations concerned. There is not, and there cannot be such a thing as quantitative economics ... [and any] forecasts about the course of economic affairs cannot be considered scientific.
  George Reisman drives home the point in his book Capitalism:
Despite popular beliefs, economics is not a science of quantitative predictions.  It does not provide reliable information on such matters as what the price of a common stock or commodity will be in the future, or what the 'gross national product' will be in the next year or quarter ...
...or what the Trade Weighted Index will ever be, except in the past. The future is unknown.  There's no way to put a number on it. Real economists know that, and don't pretend otherwise.
Proper economics -- not the mathematical junk that squeezes real market processes into a Procrustean Bed of mumbo jumbo -- that is, real descriptive economics of the kind practiced by Austrian economists will, as Reisman goes on to say, "provide an important intellectual framework for making personal and business economic decisions."  A proper knowledge of economics will help defend business and economic activity against regulators and politicians who continually seek to destroy both.  It will not teach businessmen how to make money, he says -- a skill "they possess to an incalculably greater degree than economists" -- but it will explain why it is to the self-interest of everyone that businessmen should be free to make money.
That should surely be enough for anyone, you would think, without witchdoctoring up the world with lots of overpaid and under-successful clairvoyants.

31 comments:

Anonymous said...

PC said...
And that's just the gee-whiz crystal-ball gazers who do the sums for government! The ones who pull down the really big salaries who you hear all over the TV and radio making predictions: they've got no more idea than your about what's going on than your taxi driver does, as a new study fresh from the printer makes clear.

PC, you have to differentiate between mathematical economist and the crystal-ball gazers such as those celebrity economists that are frequently making comments in our evening news. These crystal-ball gazers include Cameron Bagrie (ANZ), TONY ALEXANDER(BNZ), Nick Tuffley (ASB) and a few others. These guys don't model anything at all. They just estimate future movements based on reports compiled for them by their respective junior economists (who have just got out of varsity). There was no mathematical modeling involved in compiling those reports.

I talked to David Boyce of ASB securities recently offering to do some market modeling for them, and he told me, that they don't do that sort of thing at all, they just make judgment based on intuitions.

So, don't lumped together the real mathematical economists and the psychic economists (ie, crystal-ball gazers) because they're completely 2 different domains. Mathematical economists predictions get it right MOST of the time compared to crystal-ball gazers who are often got it wrong most of the time.

Anonymous said...

Actually PC the Economist magazine had an interesting prediction method for when an recession will end.

They looked up articles in major newspapers in the US and the month that had the highest number of times the word recession appeared was actually the end of the recession and the start of growth in the economy.

I think with most gee-whiz crystal-ball gazers they are about 6 months behind what is actually happening.

However, one granted method of things going bad is when taxi drivers start giving you stock tips.

Then it's time to bail pront.

Anonymous said...

PC said...
As Ludwig von Mises pointed out fifty years ago, economics is not a science capable of quantitative predictions.

PC, you so treat Ludwig von Mises' words as an authority in everything economics. First, I suggest that you should step out of your comfort zone and try to read more about how the discipline economics evolve from being descriptive into becoming a quantitative field. You're obsessed with economic prediction, but economics is not all about prediction. There is the application of the field of optimization in economics. How one minimizes risk while optimizing returns (regardless of where the market moves)? This is very important for investors & traders to use economic optimization, as if you ended up loosing because of market move, the amount is minimal as compared to one who didn't use economic optimization. This field was pioneered by Harry Markowitz (Economic Nobel Prize winner), James Tobin, William Sharpe and others.

Economics is now deeply quantitative in its nature and I think Ludwig von Mises would have realised that in his final years.

Peter Cresswell said...

The fact that economics is now deeply quantitative says nothing at about whether it should be.

The distinction you make between the failed crystal-ball gazers and the mathematical economists is as phony as the idea that mathematics can predict future human behaviour, or any other of the sundry nonsensical bases on which the mathematical modeling of human economic behaviour is based.

Perhaps if you yourself were to step out of your comfort zone you might realise that that the reason for economics becoming a mathematical rather than a descriptive science were and remain deeply flawed -- just as the are the predictions made by the mathematical would-be clairvoyants.

You might begin your baby steps away from the comfort zone with that linked article by Herr Mises; you could continue with an article by Gene Callahan on 'Logical Economics vs Mathematical Economics'; and perhaps conclude your afternoon with paegs of 8 and 9 of Reisman's book 'Capitalism,' which you can read online here.

I make no predictions however about the likely outcome of your reading. ;^)

Anonymous said...

The article ("Logical Economics vs. Mathematical Economics") that you linked to at Von Mises is laughable. It is obvious that the author has no clue to mathematical modeling. Economic forecasting is about likelihood that all, and for anyone to say that it is about accuracy, is either doesn't know what the whole idea of forecasting is about or otherwise he is a person that just being dismissive about everything or anything that he doesn't understand, regardless.

Here is one of his comment:

Gene Callahan said...
Mathematical equations can be useful for modeling the result of people following through on previously-made plans. Once a batter in a baseball game decides to swing at a pitch, we can use an equation that, based on the initial force the batter chose to apply to the bat, predicts its progress. This equation will be of little use, however, in predicting whether the batter will change his mind and check his swing.

But this batter's change of mind the way he swings is more likely to be captured by the model based on previous data, then one who has seen the past data and not able to spot those patterns. Human minds are not able to spot important predictive hidden informations from a huge tabular datasets (be it economic, scientific, social etc,...) with thousands or even more, a mathematical model can capture this information easily. Again, models can be dynamic & static. Dynamical modeling is more robust compared to static, because recent inputs (recent economic news or recent metric that has just arrived) can be fed into the model where it updates its model parameters. So, models are evolving, exactly as how humans incorporate new informations that has recently been made available.

One, thing about Gene Callahan's comment makes me think that he has no clue at all to what he is talking about. Some dynamic economic models don't have an equation, such as something like:

Z = x^2 - 3*x*y + 5

Most dynamic models maps the input, say X into an unknown economic systems with certain parameter characteristics, say D that produce an output Y a functional relation similar to the followings:

X -> D -> Y

X and Y are observables, ie, economic datas are available already. Systems D is unknown and it is likely to be different all the time depending of known input data X and output data Y. Once D is estimated (D is not an equation), then it is possible to predict the value of Y from an unseen value of X, ie, a value of X that was not included in the model building.

So, if you think that modeling is useless according to Mr. Callahan, then try reading the following article (see link below) and see how mathematical frequently beat the market. This success is less to do with luck, but to do with the models. There are more of these articles on the internet and I will post them here so that you can see that models outperform humans and that means, that Von Mises & Reishman are wrong.

How a Computer Knows What Many Managers Don't

I come to think that Von Mises & Reisman were more like Einsteins when Quantum Mechanics emerged in the mid-1920s. The great scientist was disillusioned with QM, that he dedicated the rest of his life to refuting it (instead of making new discoveries with the new theory - QM), but this was the unproductive years of his life, since he didn't make any new theories or discoveries over this period at all till he died in 1955. Either Einstein thought that QM was wrong philosophically and most likely to be useless in making predictions, that know new discoveries or inventions is going to come out of it. The great scientist just stood still in time, while new inventions after new inventions as a result of QM emerged. May be Von Mises & Reisman are the Einstein equivalent in Economics. Perhaps they resent not winning a Nobel Prize where most of them have been awarded to mathematical economists.

Peter Cresswell said...

FF, in commenting on the motivations of Mises and Reisman and the like, you make the same mistake your mathematical models do: drawing too many conclusions from too little data, or data you can never ever know.

The various business cycles we so frequently experience -- booom, bust, bubble and squeak -- are more a function of central bankers (for example) assuming they can plan' people's behaviour, and of forecasters assuming they can predict it.

That we all too frequently experience bust (and that our crystal ball gazers got it so wrong so frequently on the TWI) suggests there's something seriously lacking in the fundamentals.

Anonymous said...

.. an enginer, a biologist and an economist were sitting around a campfire with a can of beer and no way of opening it.

"Let's make a slingshot, use centrifugal force to hurl the can against that rock and it will pop open", said the engineer

"Let's confine it in a container surrounded by fast-growing algae, and the pressure will pop it open" suggested the biologist.

"No, there's a much easier way - let's just assume a can opener" said the economist.

Anonymous said...

the drunken watchman said...
.. an enginer, a biologist and an economist were sitting around a campfire with a can of beer and no way of opening it.

I bet you that it was a can of Waikato beer, drunken watchman?

Have you set the fishing net recently or over the summer period so far? Lots of black snappers I guess?

Anonymous said...

"Mathematical economists predictions get it right MOST of the time".

That is bullshit! The record demonstates that "mathematical modelling" is irrelevant when it comes to understanding economics. Most of the time such models are arbitrary and the predictions made on the basis of such models erroneous.

You should carefully reread what the original essay by PC actually reported.

What you are describing is not economics anyway, it's crystal ball gazing. The models you jokers build are all about your hunches and feelings just as much as the bank employees you criticise. At least they have real EXPERIENCE to fall back on.

How do you KNOW what behaviour or activity to include and what to exclude from your model? The model can't tell you that answer while you are in the process of creating it.

How do you know which man's action is important enough to include and how do you select a weighting for it, compared to another man's actions? The model can't tell you that either. And certainly not while you are in the process of creating it.

What to measure? How to measure it? Model can't help you there. You just do not know for sure.

How about understanding what the metrics you intend to use are and how they were accumulated? What about understanding whether they are relevant, accurate or merely artificats? What are the premise and assumptions? Model is blind.

What about unknowns and unknown unknowns? How do you account for data you have not got and don't even suspect exists and certainly do not understand? You are not omniscient and nor are your mathematical models.

This modelling faith is witchdoctoring; no more than that. Silly stuff!

What you NEED to do is read Von Mises and Reisman carefully. Try hard to UNDERSTAND what they are telling you. Perhaps you'll get it.

BTW Mises did not change his mind on this subject. The fact that you wish he'd have done so is not evidence to validate your statement. Given this rather pathetic inaccuracy there must be serious concerns about what you assume, what your premise are, what you include in your models and how you design them.

Next thing you'll be telling us you can "mathematically" model philosophy to determine the optimal ideas.

Naaaaaaaah! It's bullshit mate! If it is economics you really would like to understand (and not sooth saying or fortune telling) then you really do need to study Reisman & Mises.

LGM

Anonymous said...

Falfulu Fisi,

There is no academic evidence to suggest that models exist that can systematically predict the value of future asset prices such as oil, stocks, foreign currencies etc such that economic profits can be made. If mathematical modelling is as valuable as you claim, maybe you could provide some evidence of academically published work which demonstrate this so called "skill". Generally, the evidence suggests that the current value of an asset to be an unbiased estimator of future prices, an efficient price incorporating the "markets" best estimates of future value.

Optimisation - which you refer to in your first post - is in many ways based on the proposition that predicting future asset prices is not an economically justifiable proposition. Using past correlations, volatilities and returns it suggests how to construct portfolios of assets to obtain "efficient" portfolios. If mathematical models could be shown to predict future asset prices, then optimisation - I suggest - would not be required.

Julian Darby

Anonymous said...

That is bullshit! The record demonstates that "mathematical modelling" is irrelevant when it comes to understanding economics.

Which economics? Descriptive or Quantitative?

Most of the time such models are arbitrary and the predictions made on the basis of such models erroneous.

Obviously, you haven't got a clue about numerical modeling? Ok, let’s take you on an area, which you may be familiar with such as in Physics & Engineering. Schrödinger’s proposed his well-known wave equations with arbitrary assumptions, there was no evidence at all that this wave equation has any physical meanings at all, because the equation contains a complex number in it, I repeat, no evidence all just arbitrary assumptions. Complex numbers has no physical interpretation at all. It turned out that this equations made useful predictions of how the physical world (in microscopic) behaves, therefore lead to technological breakthroughs. It turned that Schrödinger’s equation failed to model certain types of waves known as solitons, because of its linearity. Tsunami ocean wave is a non-linear wave similar to solitons, ie, they're govern by the same physics of non-linear wave propagation. With the failure of linear Schrödinger’s in modeling non-linear waves as solitons, it was modified to be able to deal with non-linear situation where it lead to non-linear Schrödinger’s. Non-linear Schrödinger’s has been applied successfully to the design of high-speed/high-volume fibre optic soliton transmission systems.

Now, do you get my point or not? Technology evolves and it is no different from numerical models, be it economics, physics, systems biology or whatever. The models are only going to get improved (ie, more generalize) over time. If the physical models don't evolve or progress over time, then civilization would have had TV , mobile phones available when Maxwell invented radio waves. But in reality, models improve over time and this is fact of everyday life, and you must get over this.

You should carefully reread what the original essay by PC actually reported.

I re-read PC's post, and that was why, I emphasized that he not lumped together mathematical economists and accountants. Can you see the difference? Mathematical economists use numerical models and accountants add up numbers from a spreadsheet and make guess work. Are you with me here? The graph that PC posted on this thread is nothing of the sort of forecasting technique that is used by mathematical economists. The graph looks like one from an accountant or a child play from someone who professes to be a mathematical economist. I know what a forecasting graph is supposed to look like, because this is something that I develop software in.

What you are describing is not economics anyway, it's crystal ball gazing. The models you jokers build are all about your hunches and feelings just as much as the bank employees you criticise.

And what do you define economics anyway? Are you talking accountings or are you talking descriptive economics? So, you say that Markowitz's work was in Anthropology. Algorithmic trading is based on Arbitrage theory and it is an economic theory not a theory from Anthropology. It looks to explore opportunities that arises or exists in seconds or a blink of an eye. These opportunities don't hang around for long, ie, why those models are there to scan and ready to react in milliseconds when those opportunities do arise .Jesus, you've just told off Owen McShane on the other thread not to use wordsmithing, now you're doing wordsmithing about economics.

When you build a model, you start with what you think is how the systems work. Once the model is formulated then it is validated against past data to see if there is any indication that it fits or shall we say have some agreement according to some tolerance. That is how the real world works, be it physics, fluid dynamics, economics or anything else that could be mathematically modeled. The model evolves, since there will be unforeseeable or unexpected parameters that the modeler never thought off or perhaps left out in his initial formulation and again this is how the real world works. No one is going to invent a model that it stood unchanged in time forever. Even Relativity & Quantum Mechanics have had modifications over their entire history since they were first formulated.

At least they have real EXPERIENCE to fall back on.

Descriptive economic experience is something easily achieved by anyone. All you need to do is grab a book (with no complex equations) and read thru, perhaps in a weekend. Any toilet cleaner or rubbish collector can read and understand Descriptive economic, because there is no calculus involved. I am not interested in descriptive economics, since it doesn't make my mind tick. I delve into quantitative economics because it makes me tick. Sometimes I stayed up all night trying to figure out how to code a model from someone else's publication (ie from finance/economic literatures).

It is typical that you preach about Mises & Reishman are right and everyone else (other economists are) is wrong.

How do you KNOW what behaviour or activity to include and what to exclude from your model?

Read my previous paragraphs. By now you should understand how numerical modeling is supposed to be done. When someone formulates one, it is often that he/she would missed certain things, but models are modified as time progressed, because certain knowledge were not adopted when the model was first formulated, because the theorist him/herself didn't know about. Quantum Mechanics was not relativistic, the reason was that the founding fathers never thought that relativity is applicable in quantum mechanics, until Paul Dirac proposed of relativising of quantum mechanics. Look what happened, a whole new field & inventions were developed when Dirac published his work of relativistic quantum mechanics. Again, the main point is that models improved over time, since no one has the psychic ability to read the unknown and formulate a perfect model that is in agreement with observation 100%.

How do you know which man's action is important enough to include and how do you select a weighting for it, compared to another man's actions? The model can't tell you that either. And certainly not while you are in the process of creating it.

Here, is a stupid statement. There is no such thing in economics as a model for an individual. The model applies to an ensemble (participants in an economic dynamical system) or a collection of individuals. Economic theories that have been established apply to the masses and not to track what Helen Clark ,John Key or any other individuals is going to do at time T = t, or T = t+1, or T = t+2, and so on.

What to measure? How to measure it? Model can't help you there. You just do not know for sure.

Models are to give you indications or say estimations. Your mistake, is that you think that model gives you certainty and everything.

How about understanding what the metrics you intend to use are and how they were accumulated? What about understanding whether they are relevant, accurate or merely artificats? What are the premise and assumptions? Model is blind.

These have been published in the economics/finance literatures, and there is no need for me to repeat them here. Shall I suggest some refereed journals for you to go over and dig for info on that?

What about unknowns and unknown unknowns? How do you account for data you have not got and don't even suspect exists and certainly do not understand?

Again, read my example about Schrödinger’s equation at the beginning of my post and get the idea? Obviously Schrödinger didn't even foresee that his equation would be useless in predicting or modeling non-linear wave? Yes, but it was useless. Schrödinger didn’t think about that non-linear scenario and that is human nature. You can't know everything beforehand. You develop a model, and then at a later stage someone else spotted a weakness in your model, where the model is modified (by that person or someone else) to deal with the new unforeseen situation. This is how life is all about. Remember we're not psychics to foreknow everything in advanced.

You are not omniscient and nor are your mathematical models.

Did I claim to be omniscient? You're putting words into my mouth? And your comment is getting sillier and sillier.

This modelling faith is witchdoctoring; no more than that. Silly stuff!

Modeling is what economic knowledge grows and developed. Without modeling, then economics is dead, ie, there is nothing there to learn since all the descriptions have been there all along in the last 50 years or so.

What you NEED to do is read Von Mises and Reisman carefully. Try hard to UNDERSTAND what they are telling you. Perhaps you'll get it.

As I have stated previously, is it hard to understand Mises and Reisman descriptive economics philosophy? No it isn't hard, a toilet cleaner would find it very easy to understand Mises and Reisman's philosophies. WHY? Because it is not bloody hard to read about them. There is not difficult calculus in there. So, if I read them, I won't try hard, because they're not hard concepts to grasp.

BTW Mises did not change his mind on this subject. The fact that you wish he'd have done so is not evidence to validate your statement.

No, I didn't wish that he did. I thought that he would have realized how stupid his criticism of mathematical economics was. Just the same as Einstein, when he could have invented more by adopting quantum mechanics and at the same time, try to solve its philosophical absurdities. The mathematics of quantum mechanics is undisputable, however its interpretations is where the disagreement lies. Einstein could have used the theory to derive new models in physics, while keeping an eye on its philosophical inconsistencies.

Given this rather pathetic inaccuracy there must be serious concerns about what you assume, what your premise are, what you include in your models and how you design them.

Next thing you'll be telling us you can "mathematically" model philosophy to determine the optimal ideas.

Naaaaaaaah! It's bullshit mate! If it is economics you really would like to understand (and not sooth saying or fortune telling) then you really do need to study Reisman & Mises.


Now, you jump from one idiotic comment to another. Mathematical model for philosophy? Have you seen such model? If you have, then I am certainly keen to be pointed out to it. BTW, I am not interested in descriptive economics at all. My interest is in economics numerical modeling, so I may read Mises & Reishman, but perhaps when I want to relax with doz of steinlager.

Anonymous said...

FF

You really do not understand economics at all. Your incantations amount to a belief in mysticism; little more than that.

I note you spend a lot of time discussing everything except

...... economics.

Funny that.

The other point to note is that you have not got a solid understanding of economics at all. You admit that you have not read Mises and Reisman. In the absence of achieving that you are inferior to any toilet cleaner who already has.

LGM

Brian S said...

FF,

The only thing Markowitz is good for is writing academic papers. Same with Black-Scholes.

Hint: price movements do not follow log-normal distributions.

Anonymous said...

Julian said...
There is no academic evidence to suggest that models exist that can systematically predict the value of future asset prices such as oil, stocks, foreign currencies etc such that economic profits can be made.

Yes, Julian, because you probably still think like other descriptive economists in the way how economic things developed or how economic agents (individual) interacts where they are suppose to be static. Economic system is dynamic, which means that there is no single model that can predict (or estimate) the future accurately. However, because it is a dynamical system, there are models which had been or currently being developed to deal with the minute changes that occur almost continuously in this situation. So, if you think that there is one single model that gets it right all the time , then you're mistaken (I guess that this is what PC & LGM are bitching about, ie, if the model gets it wrong 40% and right about 60% of the time, then it is not good enough therefore it must be dismissed). I would rather use a model that gets it right 60% or above, most of the time, then not to use one.

Here are some useful pointers to your question about using optimization for predicting/forecasting prices. You have to take note that some economic related publications aren't publish in traditional economics/finance literatures but they're scattered in many disciplines , such as Computer Science, Mathematics, Engineering, etc... So, don't be surprised if you see that these publications are not your traditional economic/finance journals.

There are 2 main types of optimizations mathematics. One is the traditional deterministic, in which still dominates the Markowitz's portfolio optimization and the other is stochastics, which is non-deterministics. Its advantage over its deterministic counterpart, is that it is highly unlikely to get trapped in a local minima (local trough) and not the global minima. This means in mathematical terms that if you're lost in a jungle where there are heaps of peaks & troughs in the area (of course all these peaks & troughs don't have equal heights & depths, some will be intermediate between the highest peak and the lowest trough), then you might end up in a trough you think that is the lowest, but actually, there might be a trough that is much deeper than the one that you're getting trapped in, since you don't have a spotter plane flying above to inform you that where you are is in fact the lowest or not. Stochastic optimizations will likely to find the lowest trough in that terrain of the many variables that you modeled compared to deterministic optimizations, which does use linear algebra, that is susceptible to a phenomena in numerical analysis called ill-conditioned

Avoiding getting trapped in local minima helps in the accuracy of the prediction if it is forecasting model. It also helps to correctly quantify the minimum achievable risk (global minimum) in an investment portfolio rather than a pseudo-minimum risk (local minimum). An investor wants to avoid pseudo-minimum risk at all.

One of those recent popular stochastic optimization method is the Particle Swarm Optimization (PSO). The Support Vector Machine (SVM) in ref #2, is not stochastic, but its optimization routine is deterministic. Stochastic optimization methods are promising as it has the capability to capture economic emergent behavior (game theory anyone?) in an economic dynamical system which is truly a self-organised Complex System itself. As PC's frequently quoted message about culture. Culture is not a piece that you put into a museum where it stays the same forever. Culture is an evolving thing, so as our mathematical models about economic systems. It improves over time as new methods displaced old methods that are no longer generalizable to the wider domain of economic theory.

#1) Stocks’ Trading System Based on the Particle Swarm Optimization Algorithm

#2) Support Vector Machine Regression for Volatile Stock Market Prediction

#3) Computational intelligence in economics and finance

#4) Particle Swarm Optimization for Constrained Portfolio Selection Problems

Ref #1, #2 and #3, you can find them at the Auck University main library , in the computing section since they're computing publications, because the main library subscribes to most of Springer's & ACM's computing journals. Ref #4, you can find at the Engineering School library, since they subscribe to most of IEEE journals.

Finally, if your modelers or analysts at Russel are interested in the latest techniques in general optimizations (some do cover economic analysis, just search on economic related papers over there), then there are a number of freely available papers from Optimization Online. Here is one that I am currently learning of how to implement it (software-wise). Again, the online journal is not an economic journal perse, but since the subject of optimization is widely adopted in a variety of disciplines, there is no surprise that economic related papers are sometimes made available from there.

- A Log-Robust Optimization Approach to Portfolio Management

Anonymous said...

LGM said...
You admit that you have not read Mises and Reisman. In the absence of achieving that you are inferior to any toilet cleaner who already has.

Funny enough, a toilet cleaner who had just alerted me to the Not PC post from last weekend, titled: Saturday morning ramble #159 on Saturday, March 15, 2008. And guess what, the last article in the Ramble list quoted the following:

Climate Panel in the Hot Seat

In that article, it quoted 2 mathematical economists, Kesten C. Green and J. Scott Armstrong who published a paper that criticized the lack of knowledge by IPCC scientists about forecasting methods in their report (ie, IPCC report). Green/Scott are experts in forecasting methods & mathematics.

Here is the full paper by Green/Scott:

Global Warming: Experts’ Opinions versus Scientific Forecasts

Did you or PC spot something funny here? The toilet cleaner who alerted me was saying, surely that PC and the Mises lovers like LGM, must realized their inconsistency & nitpickings here?

Knowledge & expertise in mathematical forecasting by mathematical economists such as Green/Scott are wholeheartedly endorsed as a counter to the proponents of AGW, but at the same time, this very thread here, PC (supported 100% by LGM) said that mathematical economic forecasting is bollocks, simply because Mises & Reishman, et al, said so.

How come that a paper in mathematical economic forecasting has been hailed to have more authority and as a counter-argument to AGW alarmism, but on this thread it is being dismissed as useless for no reason other than, because Mises said so? PC has blogged about this Green/Scott paper last year (2007).

Is there nitpicking here or is there simply something called : just follow what Mises had said, because he said so.

You have to be consistent PC, because you're the only last standing man locally here in NZ to confront AGW alarmism. I bet that your doubters who frequently making comment in this blog like Eddie Visit Frequently would jump at your inconsistency and criticize you for nitpicking.

And one point goes to the toilet cleaner (never read Mises in whole life) who tipped Falafulu when he (toilet cleaner) spotted this nitpicking about mathematical economic forecasting and zero point goes to LGM for insisting that you only spot such inconsistency if one reads Mises.

Anonymous said...

Yes yes FF. All that matters to me at the end of the day is can we make money from it?

As you know all FX dealers use stochastics, Elliott Wave, and so forth.

If your system is any good do let us know. People don't care what Mises might have thought about it, as long as it works.

Actually that might explain why navel gazers - to quote Mr Key - are crap capitalists ;-)

Peter Cresswell said...

FF, you can talk all night if you want to about the ability of mathematical economics to make predictions -- and looking at your posts I see that you have -- but I notice you haven't properly addressed the 'elephant in the room' here.

To whit, that graph up there showing that over the last fourteen years the economists from a combination of the country's largest banks and financial houses have been unable even to predict the TWI with any accuracy.

All the talk in the world isn't going to change that.

A is A. They got it wrong.

What you might want to think about is WHY they got it wrong, why they keep getting it wrong, and why nonetheless they still keep getting calls to offer their expertise when, based on studies like this one, they have none.

Peter Cresswell said...

Anonymous, you said: "If your system is any good do let us know. People don't care what Mises might have thought about it, as long as it works."

If it were possible to get ahead of the market by creating a 'system' that predicted future prices, then the system itself would end up setting those prices. That fact alone is enough to demonstrate the illusion of such a system.

Anonymous, you also said that "navel gazers ... are crap capitalists." In essence, that's the very point that's being made. In fact, if you read the above post you would have seen Reisman make the very point there.

Reisman's point (if I may repeat and expand it) is that the navel gazers have no tools to offer that can help the businessman to make money -- if they did they'd be making money themselves. The value of economics lies elsewhere. To give you the whole quote from which only a portion was used in the original post:

"Business activities can endure and flourish only in a society which understand economics and which is therefore capable of appreciating their value. The value of economics to businessmen should be thought of not as teaching them how to make money (which is a talent they posses to an incalculably greater degree than economists [no matter how many models they dream up]), but as explaining why it is to the self-interest of everyone that businessmen should be free to make money. This is something businessmen [and John Key] do not know which is vital to them (and to everyone else), and which economics is uniquely qualified to explain."

In other words, there is a necessary division of labour going on here: businessmen need economists to defend them making money and to keep the government and the whiners off their backs; economists need businessmen to keep producing so they can keep making the rest of us wealthy -- but no one should be fooled into thinking that economists somehow have a magic bullet that can predict the future.

They don't.

Owen McShane said...

There is a fundamental difference, or conflict of interest between economists and businessmen.
One task of the economist is to advise governments how to create a legal framework which promotes competition so that all profits are finally driven out or close to zero. This stimulates innovation.
Naturally the businessmen are always trying to find means of maintaining profits and his quite happy to rent seek and seek subsidies and maintain monopolies if they can. This healthy tension helps keep the total system honest but also shows up the peril of the populist notion that " a few good business men could run the whole country." Indeed they could but it would be to their advantage.

Anonymous said...

Owen

Your contribution is statist bullshit that is damn near as ignorant as what that FF creature vomits forth. He has an excuse, but you should know better.

For example, take a look at this dribble, "One task of the economist is to advise governments how to create a legal framework which promotes competition so that all profits are finally driven out or close to zero."

What utter crap! Aside from not understanding what profit is and why it is necessary (similarly not understanding the nature of competition and innovation), you continue to labour under the erroneous assumption that governments are necessary to control the activity of business. They are not.

The valid purpose of government is solely to defend freedom of the individual. You may, perhaps, have heard of the three basic individual rights (which are actually three aspects of the same principle). They are right to life, right to liberty and right to pursue happiness. There is no justification for government to involve itself in business, let alone the economy, whatsoever. It is not the proper role of government to attempt to control the economy.

Just as church and state should be separate, so should government and economy.

The error you are making is exactly the one you made when you supported that travesty, the RMA. You got it wrong. Time to admit your mistake, apologise, retract and revise your view.

LGM

Anonymous said...

Anonymous said...
If your system is any good do let us know. People don't care what Mises might have thought about it, as long as it works.

Anon, who should I try to make contact with in your organisation, to let them know about my system, when I am ready to go to market? I have developed a number of models (I didn't invent them - I basically implement them from refereed research papers) that deal with forex trading, including this model described here. Both technical analysis & fundamental analysis functionalities are included.

I am gonna join the NZ shareholders association soon just to promote my stuff. I am currently proto-typing it at this stage and I am eyeing a possible move to the Icehouse (Auck Uni Business Incubator for StartUps) or the AUT tech-park incubator in Penrose. As currently I am still a one man band, I will be looking to gain funding from any venture capitalist who would think it is worth investing in my software so that to further the development in the scale that is needed to expand perhaps beyond NZ borders.

The software is going to be web-based so, there is no need for anyone overseas to do any installation, once someone does register he/she can just login from anywhere anytime from a browser. Initially, it will connect to the NZX & ASX live (ie, realtime), but other offshore market exchanges will be included over time.

NZX has offered developers such as myself a one-month connection to their server (NZX) for free in order to test the software, but after that, there will be a connection fees. I am aware of similar systems available mainly overseas (only one locally), but analytical modeling works on one thing only. The models that performed better is all that matters. I have had talked with 2 mathematicians at AUT and a few 2 physicists from Auck Uni Physics Dept, one professor from the Engineering Robotics division, about establishing a long term R&D partnership in developing new models or implementing models that are available from the literatures.

So, I do have long term plan and not just a one shot algorithm, which some future competitors may overtake you with new models, because you have been simply just left behind in the competition with your one shot algorithm. Models that are state-of-the-art today would be obsolete tomorrow and that is why linking up with University researchers is important in my view. And of course, this linking to University R&D staffs depend on how much funding that I can get access to.

Anonymous said...

fisi: FWIW, there is no Nobel prize in economics. It's the Sveriges Riksbank/Bank of Sweden prize, generally awarded for the best socialist apologetics. (And as for your "mathematical models" getting it right: four letters - LTCM!)

Economic forecasting is about likelihood
Please see http://www.mises.org/humanaction/chap6sec1.asp et seq.

I come to think that Von Mises & Reisman were more like Einsteins when Quantum Mechanics emerged in the mid-1920s.
I think you have it backwards. In effect, Mises introduced QM and all the other "economists", so called, are still refusing to acknowledge that their "classical" models are so much nonsense.
Economics is not physics, and your constant attempts at equating the two just serve to show how little you understand.

Anonymous said...

FF I'll find out after the Easter break and give you an email address to contact re your system.

Theother anon isn't me.

Anonymous said...

Brian S said...
Hint: price movements do not follow log-normal distributions.

Yeah Brian, price movements do follow a power law distribution. But it doesn't mean that Black-Scholes shouldn't have been developed. Remember, that models are evolving (as Reishman quoted, if I am correct here that culture is not something to be put in a museum and it is expected to stay the same forever), and the discovery of the power law distribution of price movement tells us that models are not static. Here is a good paper about it:

A unified econophysics explanation for the power-law exponents of stock market activity

An evolving model is similar to what you have argued here at Not PC previously about starting with a problem, then that leads to new formulation of how to attack or solve that problem.

Log-normal distribution was questioned when economists encountered the problem with the log-normal model that did'nt generalize, which they never foresee at all when the model was first formulated. But this is not a reason to dismiss the model, since no one can foresee (being psychic) or seeking help from the dead to help or inform us how a specific model is suppose to be formulated? We make assumptions, test the assumptions against real world data, deduce conclusions, then announce the conclusions to your peers. Your peers will try to scrutinize your conclusions and if they find fault or anything that you didn't foresee, then they will correct (or modify) your model. This new model is announced to your peers. Again your peers will scrutinize your model, test it against some new type of data that wasn't used previously, and if they discover something that you missed, they then add-on (or perhaps derive a new model that has the same outcome as your old model but it is more generalise than the old one). I don't need to describe it further (simply intended for the uninformed on this thread) since you know what I mean because you have done it yourself (scientific peer review publication). BTW, I have seen the abstract of your papers on Google that you co-authored.

Anonymous said...

To give a 'real life' example of this sort of thing...

Yesterday, (Thursday 20th April), I was reading the Sydney Morning Herald online as usual.

There was an article from a journalist know-it-all, quoting a myriad of economists, fund managers, analysts and 'experts', saying that Wednesday's sharp rise in the ASX market (up 3.5% for the day) was not a 'one off' bounce, but the start of a sustained rise in share prices.

After giving the 'experts' and their opinions the consideration they deserve...I immediately contacted my Broker and said to short 20 SPI200 index futures contracts when the market opened at 11:50am NZ time.

Needless to say I was filled at 5200 points and the market sank to 5142 points before closing out my short position for a rather successful day.

The key point in all of this is if these 'experts' were any good they would all be rich and successful...(as Sir Robert Jones keeps pointing out)

Anonymous said...

Opps..yesterday was, of course, the 20th of March..ha ha..*blushes*

Owen McShane said...

Re LGM and forecasts.

Auditing Scientific Forecasts.
Over the last few months the Centre has been exploring means of monitoring and auditing the process whereby scientific information is turned into policy. Kesten Green is an expert in this field. He spoke at the Heartland Conference and is based in Monash University, Melbourne, but is currently working in New Zealand.
I recommend his excellent joint paper (with Scott Armstrong) "Global Warming: Forecasts by scientists versus scientific forecasting" which is at
http://forecastingprinciples.com/Public_Policy/WarmAudit31.pdf
which in turn is within the group’s general web page "Public Policy Forecasting" at:
http://forecastingprinciples.com/Public_Policy/index.html
During my several years experience in managing venture capital funds I became acutely aware of both the strengths and weaknesses of the financial models and forecasts which are presented in support of applications for funding or for preparing business plans directed at joint venturing and licensing.
Such forecasts are also at the heart of annual reports and information memoranda supporting normal prospectuses for fund raising, and mergers and acquisitions.
Fund managers soon learn to spot the models that have started with the desired end results (say 30% return on capital) and plugged in the inputs needed to achieve that result, rather than building the model from the bottom up.
We soon learn how sensitive such outputs are to small changes in the inputs if the model runs out for five to ten years – let alone the fifty and one hundred year projections so favoured by the IPCC and the like.

Over the years the business community has accepted the need for all such models and forecasts to be subject to independent audit by acknowledged experts in the field.

However, scientists are now routinely making forecasts based on computer models without having their forecasts subject to independent audit of their forecasting techniques. A recent example is the report prepared by the Ministry of Transport which concluded that precisely 399 New Zealanders, aged 30 years or over, would die prematurely because of vehicle emissions. As soon as the report was released it became accepted as "fact" and is routinely quoted as "fact" even though the error spreads are massive.
A suitable "forecasting audit" process would put a stop to such nonsense.

The Centre believes we need to have the requirement for such forecasting audits firmly embedded in the public consciousness. We need to demonstrate the use of such audits with an example which is simple to understand, is placed within a simple legal framework, and which can be distributed as an "international" template.

The aim would be to use a particular case to develop a template could be used to encourage policy-makers in other jurisdictions and other fields to seek similar forecasting audits so that eventually such audits become as well established as "best practice" as audits of financial forecasts are today.

The whole field of environmental law desperately needs such audits for claims about future hazards from ozone depletion, contaminated soils, nuclear waste, dioxins, mercury, vehicle exhausts, and other “toxins”.

The Centre is finalising a proposal, with Kesten.
NO doubt LGM will dismiss it as statist bullshit.

Owen McShane said...

lgm and competition.
LGM descends to abuse and ad hominem attacks from his cowardly cloak of anonymity.
I sign my name so he knows who I am.
He claims I do not know anything about profit and innovation which is strange seeing that I lectured in the University MBA courses and the Auckland Insititute of Studies MBA courses in the management of innovation and change for over ten years.
I suggest that LGM might learn something from reading the economist Schumpeter.
He also fails to notice that this thread was triggered by newspaper story in a daily paper in New Zealand and the commentary is about the role of economists in an existing state such as NZ.
The thread is not about the design of a Erewhon type Libertarian Ideal "state" with no government or institutions of government which exist here and overseas.
I am describing and commenting on the world we live in and some of us have to work in.
I wonder if LGM realises how many people who are potentially Libertarian are driven away by this sort of outburst. Most of us want to make the present world a better place - not design a new Ayn Rand community on Mars, although I admit that would be fun and I would love to have a crack at it.
And finally I had virtually no input into the establishment of the RMA which was drafted in Treasury during the MId eighties.
I did attend seminars where we were told there would no more zoning under the RMA and it would focus only on externalities such as adverse effects on the environment. Anyone who is anti central planning would support such a move and the RMA actually does not mention the word zone and we were all told it was the end of land use planning.
However, the courts and the lawyers and sadly the general public would not have a bar of it and I have been fighting this corruption of the Act in commentary, Hearings and the Courts ever since.
I suggest you go lightly on the abuse. One day you may need my help in fighting the officials. You would not be the first who has had to apologise before asking for my help in the courts.

Anonymous said...

Oh come on Owen, don't be disingenous.

Ad hominem would be writing something such as, "Noddy, you low punk, you are a bastard", or, "Noddy, you are a dirty, low-down cheating crook", or, "Noddy cheats on his wife with the neighbour's daughter" and evading debating the ideas Noddy presents during debate. It is when an attack is launced against the person (Noddy) instead of the issue under examination. The objective of ad hoiminem is to discredit the person while evading directly addressing the ideas he posits.

Now I have not given you an easy time. I admit to that. But what I directed my critique against are the false ideas you promoted. I called them statist bullshit (which they are) and, having demonstrated what they mean and that they are wrong, I recommended you retract them and amend your them. First and foremost, I attacked your ideas.

As to your response, who gives a tinker's plop about preaching to paying students in a classroom for 10 years or whatever? It's irrelevant. That activity does not validate your position. Appeal to authority isn't a substitute to a reasoned argument and real knowledge.

Fact is, the presence of the RMA has been beneficial to certain professional and business interests. You didn't oppose it at incepetion and this was made clear during several interviews on free-to-air radio. You were challenged to oppose the damn thing on principle and you refused to. At the time you reported an involvement in consultations to improve implementation and that you agreed with the premise.

The "corruption" of the RMA resides not solely at the superficial level of those officals and bureaucrats whose existence is to weild it. There are other willing participants, such as "consultants", "experts" and sundry "representatives" who merrily invoice the victims (property owners) extracting significant fees, imposts and charges for the provision of various dubious "services". These expensive "services" (really a form of rent seeking) would not be required in the absence of said RMA, but you already know that.

The root cause of the problems with the RMA exists at the level of the premise and principle upon which it relies. It is a premise which you defend; namely that it is the proper role of government to control all property. The government is the final arbiter on property matters, not the property owner. Similarly in economics, for you, the government is the controller of the economy, not the individuals whose transactions form it.

Consider your claim that it is the purpose of government to eliminate profit, for the purposes of competition. Here is an example of promotion of ultimate government control over how property can be utilised and exploited. You feel it is the government's role to destroy the businessman or entrepeneur's profit (in other words, his income and also his motivation) in order to foster some arbitrary value, in this instance competition (why that, why not some other arbitrary value like equality)? Other promoters of big govt authority, such as sundry socialists, prefer to control property for purposes such as racial purity, environment, future generations, mankind (the species), the nation, the state, the thousand year reich, the empire, the working class, etc. etc. etc. None of the promotors of those values choose to recognise the three individual rights I previously mentioned, yet protection of these is the sole valid purpose of a government. Instead their objective is to negate the individual rights in favour of the collectivist values (some of which listed previously). In the process certain parties gain considerable prestigue, social status, wealth, priviledge and, the brass ring, power over others. Government control necessarily requires destruction of the individual's property right (whether he be a businessman, entrepeneur, investor or consumer).

You have been supporting the notion that it is proper that the government governs (as in rules) the people and their property. Admit it! In such a system the government is set up to award priviledges and permissions, to institute monopolies and to award rents, not to defend or promote an individual right to property. Obviously, I oppose that view and consider it immoral. That's why it was important to put you on the spot and make specific what the ideas you posit represent.

Even though it is unlikely you will alter your approach, it is important that it is made clear what it is you are supporting and what it is you are acting upon. What is being clarified here is the premise and its result when reduced to practice. Whatever you claim your ideals to be, they do not include the consistent promotion of freedom, nor of a government whose sole function is the defense of individual rights, such as the property right. Your view amounts to little more than government rules over all and that govt should exercise its authority according to your ideals (as opposed to those of, say, some city council or district planner, Karl, Noddy, whomever) or delegate authority to you or to persons who agree with you. Too bad for the rest of the population.

---

Owen, it was your volutarily made choice to promote yourself as a public figure. It is understandable that it assists your personal interests to so do. Fair enough. Nevertheless there was nothing preventing you from posting as OMS or XYY or whatever. Too late to squeal so loudly now!

Anyway, how about just debating the point at hand?


---

I wonder if you bothered to consider that I'm not a libertarian? I've mentioned it on these pages several times before. Still, it is noted how some, on ending up on the losing end of a debate, lacking reasoned contribution, resort to the weak claim of how some fictious "others" are scared off and away by the vigorous demolishing of a weak argument! If someone is scared off by a little spiritied debate, then they are of not much substance. What use such soft-headed mushy darlings.... precious didums!

Owen, your approach is faulty. Look around you. The results of the application of central planning and associated big govt command systems is failure; property rights are dissipated, resources are consumed for little or no gain, savings eliminated, wealth is destroyed. The evidence is available to those who are prepared to witness it. Those results are exactly what drove me away from your class of ideology.

Now, that's some real plain speaking for you!


LGM


PS Schumpeter? Yup. Read him. Not impressive. Try Reisman if you're really interested in understanding economics.

Owen McShane said...

lgm
A few facts.
The RMA was drafted in the mid eighties and introduced as law in 1991. I was not involved.
In 1995/96 I was asked by the Gov of the Reserve Bank to report on why house prices were rising when other inflation was under control.
I reported back blaming the RMA and Growth Management etc etc. I was pilloried but what I predicted has all come true.
Then in 1997 Simon Upton asked me to comment on land use controls in the RMA and why they were still rampant when the RMA was intended to prevent the direction and control of the use of land.
I was not to comment on section 5 etc to keep the focus. I accepted the task. This was not the launching of the Act. The Minister was trying to figure out what had gone wrong. I identified an international trend towards Growth Management or Smart Growth and recommended changes to halt it.
The govt was voted out. Never happened.

I did not say the Governments job is to drive out competition. One function of government is to protect property rights (in all forms) and to provide a legal framework which generates trust and enforces contract. Adam Smith pointed out that merchants combine against the public if given the chance and hence the legal framework should encourage vigorous competition and prohibit monopoly or conspiracy against the public.
The end result of competing individuals is to drive out excess profit and thus stimulate innovation which is what generates wealth. Human innovation - creativity - is what generates wealth. But when favoured business had monopoly or licence rights then they can rent seek to their heart's content.
Economists now understand this and are best used as our agents advisiing government how to rule against monopolies, subsidies, licences and other privilege. When people like Tindall advise the public they always lobby for subsidy, special interests, and intervention. Climate Change is the latest excuse for such interventions and business is rushing to the subsidy party.
That is why I spend most of my time trying to expose it for the Scam of the century - which it is.

Anonymous said...

PC, here is an interesting article for you & LGM to read from Herald columnist Diana Clement. Yep what she's talking about is Harry Markowitz mathematical optimization theory. She didn't mention Markowitz or optimization, but efficient allocations is what Markowitz is all about.

Diana Clement: Don't lose your shirt - diversify

BTW, I've heard that your investment club lost about $8000 in your portfolio recently ? If you use Markowitz, perhaps your loss could be much lower than that. I have offered many times in the past to your investment group to do Markowitz allocation for free for your portfolio assets, but you guys weren't/aren't keen at all. Well, that's your loss Clubwiz.