Jul

5

 We have a summer intern with us from a university where he has been taught that prices are random and markets not predictable, EMT, anyone have any data, studies etc I can show this poorly educated fellow to enlighten him?

Rocky Humbert writes: 

Larry,

It sounds like you picked a summer intern from a university that is using obsolete textbooks.

Virtually no academics (including Fama) still believe in the gospel of strong-from EMH. I don't think it's possible to "disprove" semi-strong and weak-form EMH because the theories are constructed in such a way as to leave wiggle room.

If you are suggesting that all forms of EMH are incorrect, then I beg to differ.

Lastly, data mining to find low probability events (as some speclisters have suggested) does not necessarily prove nor disprove a hypothesis anymore than pointing to Winston Churchill as proof that cognac and cigars lead to a long and vigorous life. Most of the time, the market is darn efficient. And that's one reason that markets are the best way to allocate resources.

Russ Sears writes: 

Perhaps the best set of data I can think of to disprove ALL forms of the EMH is the interest rates over the last 50-60 years. In the 60's the Phillips curve took over the feds interest rate models since then the bias has been more control of the interest rates is always right. Likewise from 85 to now feds have stopped both inflation and any liquidity crisis (real or imagined). Granted it is a bit of cherry picking to calculate the chance of randomly reaching 85's interest rate levels from 1960 and then multiple that by the chance of coming from 85's levels to 2014/15 levels

I lost a job because in the interview I told the guy in charge of the modeling for a one of the biggest insurance companies that I thought he was wasting the companies money having 2 Phd's calculate the interest rate scenarios using the random walk. The company hadn't even tested any of their correlation of their interest rates competitiveness to their change in lapse rates. But they wanted to have a risk neutral yield curve monthly binary tree model built 30 year out quarterly nodes with several orders of accuracy. If you used such a model for the past 2 X 30 year periods each actual outcome would at best been so remotely possible that only a naive statistician would not see the coin flips were rigged.

I was told that the interviewer thought I was too simply and couldn't handle the sophistication of the math they wanted. Academia seems to thrive on sophistication for job creation sake, not money making sake. Not coming from the Ivies or having a Phd I assume that the only reason I got the interview in the first place was that I had made my past two companies millions betting on long term gamma, for almost nothing. So what do I know.

Even the idea behind the Feds "control" screams non-random walk. If you stifle the short term natural swings it is bound to have long term consequences. 

Gordon Haave writes: 

"I was told that the interviewer thought I was too simply and couldn't handle the sophistication of the math they wanted. Academia seems to thrive on sophistication for job creation sake, not money making sake."

That very accurately describes all of economics and everything surrounding the Fed, although it is not for job creation sake but rather for obfuscation sake. There is nothing more satisfactory than telling an economist that the fed is printing money only for them to rant and rave that the fed doesn't actually print money, and then saying "I know, but the effect is the same".

Then the response is always "it's more complicated than that". But they will never really tell you why in a meaningful way.

Russ Sears writes: 

Perhaps I should read the paper before I comment but my bigger point was to actually be a "science", actuaries and other modelers need to form a hypothesis/model and THEN look at the actual results to at least adjust that model if not scape it altogether. The math is made to predict the data. Not the predictions must be based on the beauty of the math theory Otherwise it is a philosophy not an art.

Academia loves philosophy because it implies the philosopher should be in charge. They dispose science because it implies academia must be humble to the wisdom of the crowd. If you're predicting rate of change long term then it is not enough to validate your models using first order changes such as lapse rates. You must validate second order effects such as shock lapse rates and long term drifts. It shows it gets messy when the philosophers are in charge.


Comments

Name

Email

Website

Speak your mind

Archives

Resources & Links

Search