Comments on: One quadrillion dollars on inspectd.com

By: Has Mike Kania Disproven Efficient Market Theory Using Inspectd.com? on InvestorGeeks

Wed, 09 Apr 2008 03:18:04 +0000

[...] March 25, 2008: Ugly gets to 1 Quadrillion dollars. [...]

By: Tyro

Tyro — Thu, 27 Mar 2008 13:20:05 +0000

Ugly - I’ve checked a few sites for a clear statement of the Efficient Market Hypothesis. Do you have one that spells it out clearly enough to make predictions like this?

I’m not convinced yet of your accomplishment without knowing how inspectd selects their stocks and timeframe (e.g.: a bias for a bull-market or a bias for stocks which haven’t disappeared or bankrupted), nor what criteria you may be using to evaluate the charts (e.g.: if you recognize them), nor whether we’re looking at some survivor bias. I certainly agree with you that the EMH is bunk, but I’m not convinced that this is the best way of doing it.

Every time I looked to find a clear statement of the EMH, it’s so wispy that it could be interpreted to mean just about anything. In some senses, any price fluctuation at all invalidates the EMH since they would take it away from the “efficient” price; in other senses, any price fluctuation can be explained by reactions to news.

By: Ugly

Ugly — Thu, 27 Mar 2008 12:32:53 +0000

Tyro: even the weak form of the efficient market hypothesis says “Excess returns cannot be earned by using investment strategies based on historical share prices.” But that is exactly what I have shown can be done. TA shows this can be done, so I think it directly challenges the EMH.

By: Tyro

Tyro — Wed, 26 Mar 2008 16:14:16 +0000

Part of the problem with the Efficient Market Hypothesis is that it’s so imprecise. News comes in all the time, so you can always find some news event to point to and say “look, this move was caused by this news”. There’s also the question of timeframe: at what timeframe is the market supposed to find the “right” price?

When you start talking about TA, I don’t think you’re challenging the Efficient Market Hypothesis since I think TA can co-exist with an Efficient Market. Instead, I think you’re challenging the Random Walk Model, and that can be disproven without bothering with TA or any of the possible biases that are present with inspectd. You’d just need to analyze the price moves and compare it to a normal distribution. Benoit Mandelbrot has already done this and proven that price moves are not distributed normally and are not random in this sense.

By: Brad

Brad — Wed, 26 Mar 2008 14:49:22 +0000

A paper on algorithms inside parallel file systems. Not certain it will get accepted at SC.

By: Ugly

Ugly — Wed, 26 Mar 2008 13:27:54 +0000

Brad, when you get past your deadline let’s work together to prove this. I think it would be worthwhile and satisfying to statistically prove that TA works and that the Efficient Market Hypothesis is incorrect. I think I’ve put forth a considerable amount of evidence to justify my argument - but I’d love to make a statistical proof of it - one that can’t be denied.
What are you doing for SC08?

By: Brad

Brad — Wed, 26 Mar 2008 03:44:37 +0000

I thouht some more about your experiment to prove TA works. Although it isn’t strictly neccesary, to show a more statistically valid sample, you would need to bet the same amount on each trade. If you don’t compounding will skew the computation. It won’t alter the results, however it is likely to increase the variance, as the result of the first trade may have a greater impact than the value of the 2nd trade, causing the value of subsequent trades to vary wildly. Again, it isn’t strictly necessary to do it that way, but it would lead to an easier apples to apples comparison. And of course if TA is going to win, that would show a stronger result.

I’ve thought of several different ways to run this experiment that could be interesting. But I have a conference deadline (Supercomputing, where Kurzweil gave the keynote in 06), and can’t mess with it right now.