It's somewhat amusing to see Apple throw away market leadership now, just like it did in the 80s

[KrazyHorse]

Hey, without the EMH I wouldn't be able to predict the spot price of oil a year from now!

Using futures market data to do that is a very, very bad idea...

[KrazyHorse]

Read the nuance -- the EMH falls under the "model" category, not predict.

And how does the EMH allow me to model the future?

[Hauldren Collider]

Read the nuance -- the EMH falls under the "model" category, not predict.

Wait, what's the difference? Don't you use models to make predictions?

[Asher]

And how does the EMH allow me to model the future?

Where did I mention using the EMH to model the future?

The EMH is a model to explain the present.

[Asher]

Wait, what's the difference? Don't you use models to make predictions?

Models are constructs to aid in understanding.

[Hauldren Collider]

Where did I mention using the EMH to model the future?

The EMH is a model to explain the present.

Uh, then it wouldn't be falsifiable. That doesn't make any sense.

[Kuciwalker]

Warren Buffett?

If Warren Buffet is the only weakness in the EMH then it is a really, really good theory.

[Asher]

Uh, then it wouldn't be falsifiable. That doesn't make any sense.

Of course it could be falsifiable. The market is not fully rational, so no model could be completely accurate.

[Hauldren Collider]

Models are constructs to aid in understanding.

And then that understanding is applied to make predictions, that's science. I'm confused, I think you are equating two different kinds of predictions.

[Asher]

And then that understanding is applied to make predictions, that's science. I'm confused, I think you are equating two different kinds of predictions.

Economics is not a science. Models in social sciences are not the same as models in hard science.

[Kuciwalker]

The EMH as a theory makes the following prediction: whether investors' returns are serially correlated (as alluded to by both me and KH earlier). This is a testable prediction, it has been tested, and it has been found to be sufficiently true.

[KrazyHorse]

The market clearly doesn't know what the future price of a stock would be. Or even have a good idea.

It depends what you mean by "good". If you mean "can predict with subpercent accuracy" then no. If it means "better than Asher, me, Kuci or almost anybody on the planet", then yes

Warren Buffett?

Yes, congratulations, you have mentioned one of the very few people in the world who MAY be able to significantly outperform the market. And due to his outperformance (on the order of a few percent a year over 50 years), he is one of the richest men in the world. Where are your tens of billions?

Yet the stock price is so high because people believe it'll continue.

Yes, millions of average people and thousands of professional investors have all laid real money on the line, and the current price is the aggregate of all of their opinions. I'll take their judgment over yours any day of the week.

I don't think any one person can accurately predict every stock, quite obviously. But it is fairly easy to see some stocks that are absurdly priced as people get carried away.

And I'm sure if you guessed 50 times you would be right some of the time. But I can see no reason to believe that you would KNOW AHEAD OF TIME WHEN YOU WERE WRONG.

[Hauldren Collider]

Economics is not a science. Models in social sciences are not the same as models in hard science.

Why not? Is biology a soft science?

[KrazyHorse]

The EMH as a theory makes the following prediction: whether investors' returns are serially correlated (as alluded to by both me and KH earlier). This is a testable prediction, it has been tested, and it has been found to be sufficiently true.

There has been found to be a small amount of serial autocorrelation at the extreme top end of the distribution of money managers (top ~3% IIRC)

[Kuciwalker]

Yeah, I read (and linked to) the Scott Sumner post talking about it, but didn't want to add in a big and ultimately pointless caveat; thus "sufficiently".