Physicists vs Mathematicians?

[Dorothy]

The following is the url for an Associated Press article that appeared in our local paper yesterday:

http://www.journalnow.com/wsj/business/MGBOJK8WFAD.html

Comments?

[/url]

[Dr T Non-Fan]

1. Auburn? Honestly.
2. I don't think his predictions are all that accurate. Might be better than current models, but still isn't good enough.

Now imagine measuring the planets' collective gravitational pull. The curve is pretty wild, but an average can be plotted - with a Levy curve. One stock is impossible to fit into such a model, but the curve can be adjusted to fit a "basket" of stocks - the S&P 500 or a mutual fund - with fair accuracy, Oks said.
Sounds familiar to Course 6 material. Sounds as if he's reinventing wheels.
In Alabama, however, he might actually be inventing the wheel.

[Wag, the Dog]

The only part that I did not understand is: exactly how did Eugene Oks get the horse to squat over Jacqueline Kochak?

[Will Durant]

when companies lose value - because stock prices plummet - they stop growing, limit excess inventory and lay off unneeded workers
I thought companies stopped growing, etc., when their REVENUES (or prospects for future revenue) went down.
:duh:

[TwistedMentat]

when companies lose value - because stock prices plummet - they stop growing, limit excess inventory and lay off unneeded workers
I thought companies stopped growing, etc., when their REVENUES (or prospects for future revenue) went down.
:duh:
In certain areas of advanced physics, cause and effect get a little fuzzy.

[Paddyboy1]

I have to say either this guy is some kind of crackpot or else the reporter doesnt know the first thing about math or statistics.

1. His change in methodology sounds like he says he has come up with a better way of determining market volatility or perhaps a better way of diversifying stocks. But his prediction (October 9th was the market bottom and that growth will be slightly under 10% a year) is a market-timing prediction - why would this matter if you're using a lognormal curve vs a gauss curve or levy curve?

2. His description of his testing of his assumptions on historical data sounds extremely unverified:

"No other model has ever yielded such accuracy," he said. "There's still some risk, however, because there's always some chance reality will be different from predictions."

The most accurate model in history! Yeah, thats why you're talking to a reporter and professors in Alabama and not your butler in New York.

3. And what exactlly did you test?

"I can go back 10 years, run the model and see what was predicted for a year-and-a-half later," Oks said. "Then I compare the prediction with reality. I've found the model to be accurate within 20 percent, plus or minus half a year."

That means that if Oks' model were to show the next stock market peaking 75 percent higher than today, the peak could actually be from 60 percent to 90 percent higher.

What on earth does this mean? I've read it 3 times. Wouldnt the range be 55% to 95%, and what does plus or minus half a year. If he's saying he's giving himself a 40% band to work with and he's only predicting 18 months into the future, and it just has to hit inside that band at one point in the 6 months preceding or 6 months following the 18 month point, than that is silly. Also he is going back 10 years, was his model parameterized considering this 10 year period or some other period?

4. His current predictions sound like no-brainers trying to attract funding. He sees Oct. 9 as the market bottom! What date was he saying the bottom was on Oct. 8 is my question. Anyone can find the low point on a chart, and we've had such a huge runup since Oct. 9 that it will probably take some time to get back down there again. As for 9% growth, for the next 3 years, that sounds awful close to the long-run S&P growth. But it is all going to be up front! Quick give me your moeny to invest before its too late!

[Dorothy]

Thanks. I was curious as to how others would view the article. I can't imagine why this physicist has such an antagonistic attitude toward mathematicians.

My own reaction was one of surprise, since I have been modeling projected fund values as part of variable universal life guaranteed death benefit reserve calculations, using the lognormal distribution. I suppose that is not quite the same as the distribution that Dr. Oks advocates, but, then again, not as limiting in terms of sampling extreme values as the normal distribution.

My supervisor, on viewing the workings of my model for the first time, was very pleased and suggested that I might present the model at an SOA meeting. Although I was quite gratified by his praise, in acknowledging it I stated very sincerely that I would be very surprised if others in the actuarial community had not created models that were as good or better. I think that Dr. Oks must be very isolated indeed if he really thinks that his model is the best that has ever been created.

About the only thing I know about Auburn is that it has a rather unimpressive web site that I hope is not representative of the technological capabilities of the whole of that institution.

[Paddyboy1]

Dorothy,

If I remember CAS part 8 correctly, a normal distribution of periodic stock market changes (like if each day's percentage return for a stock was drawn from a normal distribution) results in a lognormal distribution of ultimate stock values. I'm guessing you fit a lognormal to ultimate fund values, which is consistent with modeling market fluctuations with a normal. In other words, your method is the orthodoxy that this physicist doesnt like.

[Dorothy]

Paddyboy,

The monthly accumulation factorin my model is an exponential function in which sample values from N(0,1) are incorporated into the exponent along with the mean and standard deviation of S&P 500 returns over the last 50 years or so. My understanding is that the resulting distribution is preferable to the normal distribution as its tails are longer and fatter.

I don't have the advantage of having studied the investment material. I have just constructed a model using information that my boss, an FSA, obtained from a seminar he attended a couple of years ago.

At any rate, you have to admit that the guy seems a little loopy.

[Paddyboy1]

Dorothy,

The Black-Scholes option pricing model uses a similar method, except instead of monthly returns it does instantaneous ones. I think that smart people do have problems with this method, but this guy probably isnt one of them. My whole point is that when he says he disagrees with the bell curve he's referring to a bell curve of periodic % changes, which results in a lognormal distribution of end-of-period changes just like you did in your model.

[exactuary]

FWIW, periodic normal returns approach lognormality at endpoints as the number of periods increase and/or the length of the periods is shortened. Because the normal distribution does not multiply well (it adds well), it is usually better applied to cross-sections of returns within a single period. Because the lognormal does multiply well (but does not add), it is tractible for longditudinal work. Thus, when working on multi-period problems, it is best to assume lognormal for each period thus assuring (rather than asymptotically approaching) a lognormal endpoint.

[Paddyboy1]

I think as the number of period increase it will approach lognormal, but I dont think short periods will get it there any faster.

[exactuary]

For a fixed end point, it will increase the number of periods.

[Paddyboy1]

Well, yeah. Just to nitpick: when you said

"number of periods increase and/or the length of the periods is shortened"

the "or" part really doesnt work. Because if the length of the periods are shortened and the number of periods *dont* increase, yours is a false statement. In this case we use the logical operator "and", not "or". You should have said

"number of periods increase"

[exactuary]

Well, as an exact actuary, I am glad I only screwed up the language.


And, of course, I always knew what I meant.

[Paddyboy1]

And, of course, I always knew what I meant.

I know you knew.

[exactuary]

And, of course, I knew that you knew that I knew, which is why I said "of course," of course.

[Bama Gambler]

In Alabama, however, he might actually be inventing the wheel.

Watch it! Not everyone in Alabama is stupid, just most of them. Rooollll Tide!

[Dr T Non-Fan]

I cannot be held responsible for what I wrote 19 days ago, BG.

[Bama Gambler]

fair enough

[Dr T Non-Fan]

Since you're in a good mood, please change "days" to "minutes."

[764dak]

I prefer mathematicians over physicists any day of the week and twice on Sunday.

[Brad Gile]

I prefer mathematicians over physicists any day of the week and twice on Sunday.

:iatp: :)

[echo]

I prefer mathematicians over physicists any day of the week and twice on Sunday.

Why?
Why do you prefer mathematicians?
Why do you dislike physicists?
Why, most importantly, would you post to a thread that's been dormant for over 9 years?

[OldGuy]

Physicists are using mathematics to figure out how something works. Mathematicians are playing wih their food.

Actuaries use mathematics to figure out how to make things work. Like insurance.

I perfer physicists.

[echo]

Physicists are using mathematics to figure out how something works. Mathematicians are playing wih their food.

Actuaries use mathematics to figure out how to make things work. Like insurance.

I perfer physicists.

+1

[Avi]

I believe my opinions are encapsulated in my sig.

[echo]

I believe my opinions are encapsulated in my sig.

An engineer, a physicist, and a mathematician, and a statistician are all staying at a hotel. In the middle of the night the engineer wakes up to find that his trashcan is on fire. He runs to the sink, fills his ice bucket with water and douses the flames. Then, just to be sure, he runs back to the sink, refills the bucket and dumps more water into the trashcan. With the fire out, he goes back to sleep.
A little while later, the trashcan in the physist's room spontaneously breaks into flame, waking the physicist. He whips out his slide rule, does some calculations, then runs to the sink, fills his bucket with exactly .75 liters of water, and douses the flames. Having put out the fire, he goes back to sleep.

A few minutes later, the mathematician wakes up to see that his trashcan is on fire. He whips out a piece of paper, scrawls out some equations, then goes back to sleep, comfortable that a solution exists.

Meanwhile, the statistician is running from room to room lighting trashcans on fire -- he needed more samples.

[Avi]

An engineer, a physicist, and a mathematician, and a statistician are all staying at a hotel. In the middle of the night the engineer wakes up to find that his trashcan is on fire. He runs to the sink, fills his ice bucket with water and douses the flames. Then, just to be sure, he runs back to the sink, refills the bucket and dumps more water into the trashcan. With the fire out, he goes back to sleep.
A little while later, the trashcan in the physist's room spontaneously breaks into flame, waking the physicist. He whips out his slide rule, does some calculations, then runs to the sink, fills his bucket with exactly .75 liters of water, and douses the flames. Having put out the fire, he goes back to sleep.

A few minutes later, the mathematician wakes up to see that his trashcan is on fire. He whips out a piece of paper, scrawls out some equations, then goes back to sleep, comfortable that a solution exists.

Meanwhile, the statistician is running from room to room lighting trashcans on fire -- he needed more samples.
:rofl:

[764dak]

Why?

Mathematicians are better at mathematical finance.


Why do you prefer mathematicians?

They are also not cocky scientists.


Why do you dislike physicists?

They are cocky scientists.


Why, most importantly, would you post to a thread that's been dormant for over 9 years?

I wanted to post in this thread.

[764dak]

Mathematicians are playing wih their food.



What do you think about applied mathematicians?

[764dak]

An engineer, a physicist, and a mathematician, and a statistician are all staying at a hotel. In the middle of the night the engineer wakes up to find that his trashcan is on fire. He runs to the sink, fills his ice bucket with water and douses the flames. Then, just to be sure, he runs back to the sink, refills the bucket and dumps more water into the trashcan. With the fire out, he goes back to sleep.
A little while later, the trashcan in the physist's room spontaneously breaks into flame, waking the physicist. He whips out his slide rule, does some calculations, then runs to the sink, fills his bucket with exactly .75 liters of water, and douses the flames. Having put out the fire, he goes back to sleep.

A few minutes later, the mathematician wakes up to see that his trashcan is on fire. He whips out a piece of paper, scrawls out some equations, then goes back to sleep, comfortable that a solution exists.

Meanwhile, the statistician is running from room to room lighting trashcans on fire -- he needed more samples.

This is quite enjoyable to my sense of humour.

[Arthur Kade]

What do you think about applied mathematicians?
They are playing with their cooked steak.

Pure mathematicians are playing with the cow while it's alive.

Pure mathematics is on the whole distinctly more useful than applied. - G. H. Hardy

I have absolutely no idea wtf I'm talking about; I just wanted to post in this thread too.

[mathmajor]

I always used to think I was a mathematician, but I'm not. As much fun as it is studying rings and fields, I'd rather work on something more tangible. Actuarial is a perfect fit for me. People come to me all the time wanting to know if prob/stats can be used to do something.

"Excel has all these probability functions, certainly one of these will do something!"