4 ms·
They almost certainly can't be too efficient. There's a proof that if markets are efficient then P = NP. This isn't really about whether markets are or can be
by cryptonector 9y ago
They almost certainly can't be too efficient. There's a proof that if markets are efficient then P = NP.
This isn't really about whether markets are or can be efficient. This is about corporate governance. Boards of directors should not permit CEOs to play these games, and compensation of CEOs should be structured to not create such destructive incentives.
- goldenkey 9y agoPerfect efficiency does not exist. Some efficiency does. There are three variants of the hypothesis: "weak", "semi-strong", and "strong" form. [1] Insider information isn't some conspiracy -- people have gone to jail over it year after year. With information disparity, efficiency is always going to be a fraction of what it could be. Just think about two parties -- one thinks the stock should be high because of insider information. And lets say they are also mega-rich big bucks. The other party is people without the insider info, they think the stock should be low. So Mr. Mega-Rich buys a lot of shares, the stock goes up. Then it falls back down because no one else wants to buy it at that high price. These oscillations will continue as Mr. Mega-Rich takes advantage of his insider information - and the more trading volume on the buying end, that he provides, the more choppy the market for that equity will be. If Mr. Mega-Rich is continuously buying, that choppyness won't go away - lest other parties take his actions to be information itself, and adjust their appraisals. Wall Street I and II aren't the most accurate pictures, but they certainly aren't fiction. If you aren't the shark, you're getting sharked. But because of 21st century prosperity, you still might be happy with your returns, even while paying the tax of giving money to those with better information. [1] https://en.wikipedia.org/wiki/Efficient-market_hypothesis https://en.wikipedia.org/wiki/Efficient-market_hypothesis
- chibg10 9y ago> If you aren't the shark, you're getting sharked People say this a lot, but I think it's misleading to most who hear it. This is certainly true in a trading context (i.e. where the buy/sell time horizon is short enough where the underlying asset's economic value doesn't change, though the market price might), but not in an investment context (where it's expected that the inflow of capital will create economic value and the stock price will reflect that). The takeaway is that you should leave daytrading to the pros, and instead bet on humanity improving the means of production over long time horizons. Unless you're actually a pro and have good reason to believe you can beat the market.
- tlb 9y agoBeware of "perfect" as a requirement. For instance, if you can do perfectly accurate calculations with real numbers in constant time, that also leads to P=NP (because you can encode entire problems in very long strings of digits and solve them with a few operations.) You can have efficient-enough markets, like ±0.001% efficient, without requiring P=NP.
- dv_dt 9y agoI think it's not efficiency in the sense that you're talking about, it's that the picking and pricing of stocks as a signal of their effectiveness back to the company that is broken.
- kd0amg 9y agoThere's a proof that if markets are efficient then P = NP. The only "proof" of this I've seen asserts that there are 2^n boolean-valued functions on n-bit vectors.
- jessaustin 9y agoDoes the apparently dubious proof assert that there are exactly that many functions? Because if the assertion is rather that there at least that many, that assertion is true but really understating the case. E.g. for n=3 there are 2^(2^n)=256 functions, and after that it really blows up.
- kd0amg 9y agoIt gives that as an upper bound on the size of the search space of boolean functions with n-bit input, representing technical strategies that consider the last n market movements. It goes on to claim that if there is a historically winning technical strategy, a nondeterministic Turing machine can write out a full description of that strategy in polynomial time. The remaining problem is how to uniquely describe in polynomial space an element of a set with doubly exponentially many elements, but the authors don't bring that up because they're working under the assumption that there are 2^n functions rather than 2^(2^n). The assertion that the efficient market hypothesis is equivalent to P=NP also comes with no proof that future performance of a trading strategy will match its past performance or that anyone will actually deploy a perfect strategy if one exists.
- jessaustin 9y agoHahaha ok if that was supposed to be an upper bound then something is very wrong!