3 ms·

I rarely do this any more, because I consciously assume that an expert knows more about their field than I do, unless I am certain of specific facts. (And, perv

I rarely do this any more, because I consciously assume that an expert knows more about their field than I do, unless I am certain of specific facts. (And, perversely, I want this article to be true, because I loathe the patent system in its current incarnation and I think it is a serious problem for our society.) That aside, there were a few things from this that didn't make sense to me: > The theory is simple: a company's stock price represents the stock market's best estimation of the company's value. This is contrary to all of the very little bit that I understand of the stock market. Investors attempt to find companies which are undervalued, and invest in them; by definition, that would mean that the market is not accurately estimating the company's value. A company's "real value", however you care to measure that, might be some kind of attractor in the market graph, but that's about the best I'd give it. Not that I know anything about this stuff. > If the company's stock drops by, say, two percent in the days after a lawsuit is filed, then the market thinks the lawsuit will cost the company two percent of its market capitalization. I am more certain of four problems here. One, that companies can take a stock hit for a variety of reasons, all, some, or none of which may be related to a patent issue. Two, although I can't be arsed to find an example at the moment, I'd wager a fair amount that there have been no shortage of data points where a company's stock market value rises, or stays level, in correlation with a patent issue. Three, "the market" doesn't "think". The market is simply an abstract construct composed of a huge number of individuals following individual goals and algorithms. Anthropomorphisation of complex systems often leads to faulty conclusions -- like using those systems to evaluate the value of individual elements. And fourth, since we can't accurately tie the company's value to its stock price in the first place, then concluding that a two-percent loss of stock price represents a two-percent loss of the company's fortunes is erroneous. > But with a large sample of companies, these random factors should mostly cancel each other out... And I am very certain that this is an error. Why "should" these random factors cancel out? Have there been any studies finding that they do, or are there any lemmas in statistics that supports this? Aren't the accumulation of errors in statistics usually represented as minima and maxima values, i.e., "we think the value could be here, but it could be as low as there or as high as there..."? I like the conclusions but I'm uncertain of the claims.

URSpider94 15y ago
>> The theory is simple: a company's stock price represents the stock market's best estimation of the company's value. >This is contrary to all of the very little bit that I understand of the stock market. Investors attempt to find companies which are undervalued, and invest in them; by definition, that would mean that the market is not accurately estimating the company's value. A company's "real value", however you care to measure that, might be some kind of attractor in the market graph, but that's about the best I'd give it. Not that I know anything about this stuff. It's a tautology that the value of something is defined by what people are willing to pay for it. When the markets are open, shares of a large company are changing hands hundreds to thousands of times per minute. It sounds like you are suggesting that a company has some kind of "real value" that's divorced from its current selling price -- if so, how would one determine what that is? The efficient market hypothesis (which most pros believe is accurate, at least in its weakest form) tells us that it's not possible to pick stocks that are "undervalued." The EMH dictates that all information relevant to the value of a company is incorporated in the price, so it's not possible for any one investor to rationally discern that a particular stock is undervalued.
dredmorbius 15y ago
The "investors find undervalued stocks" proposition is also testable: find investors for whom this is a stated investment strategy, and track the performance of their picks vs. the market at large. A small number of fund managers do have long-term above-market returns (Warren Buffett comes to mind). However many have only a short run (even Buffett made much of his bank in earlier times). There's a lot of back-and-forth on the efficient market hypothesis, but (other than market-wide run-ups based on bubbles / inflation / loose credit) by and large it holds with small opportunities for arbitrage. More likely you'll see manipulation (cornering markets, playing "penny stocks"). Regarding stock performance following an event, this can be statistically controlled in a well-designed study. I'm not saying that the referenced paper describes one, or that it doesn't, just that this is feasible. Control can come both from a well-selected sample size, and including other control variables in your statistical model. "The law of large numbers" as applied to random samples will ensure the normalizing tendencies. A cogent argument could be made that stock markets and industry segments aren't particularly amenable to random sampling, though as I said above, other statistical controls also exist.

URSpider94 15y ago

>> The theory is simple: a company's stock price represents the stock market's best estimation of the company's value. >This is contrary to all of the very little bit that I understand of the stock market. Investors attempt to find companies which are undervalued, and invest in them; by definition, that would mean that the market is not accurately estimating the company's value. A company's "real value", however you care to measure that, might be some kind of attractor in the market graph, but that's about the best I'd give it. Not that I know anything about this stuff. It's a tautology that the value of something is defined by what people are willing to pay for it. When the markets are open, shares of a large company are changing hands hundreds to thousands of times per minute. It sounds like you are suggesting that a company has some kind of "real value" that's divorced from its current selling price -- if so, how would one determine what that is? The efficient market hypothesis (which most pros believe is accurate, at least in its weakest form) tells us that it's not possible to pick stocks that are "undervalued." The EMH dictates that all information relevant to the value of a company is incorporated in the price, so it's not possible for any one investor to rationally discern that a particular stock is undervalued.

dredmorbius 15y ago

The "investors find undervalued stocks" proposition is also testable: find investors for whom this is a stated investment strategy, and track the performance of their picks vs. the market at large. A small number of fund managers do have long-term above-market returns (Warren Buffett comes to mind). However many have only a short run (even Buffett made much of his bank in earlier times). There's a lot of back-and-forth on the efficient market hypothesis, but (other than market-wide run-ups based on bubbles / inflation / loose credit) by and large it holds with small opportunities for arbitrage. More likely you'll see manipulation (cornering markets, playing "penny stocks"). Regarding stock performance following an event, this can be statistically controlled in a well-designed study. I'm not saying that the referenced paper describes one, or that it doesn't, just that this is feasible. Control can come both from a well-selected sample size, and including other control variables in your statistical model. "The law of large numbers" as applied to random samples will ensure the normalizing tendencies. A cogent argument could be made that stock markets and industry segments aren't particularly amenable to random sampling, though as I said above, other statistical controls also exist.