The "stop-loss" controversy: Fisher, Nelson, Montier and me

One of Ken Fisher's entries in "debunkery" which I do agree with is the one summarized pithily on the book jacket as "stop-losses should be called stop-gains" -- though in this case I'm not sure whether I agree with his main reason for saying so (i.e., that "stock prices aren't serially correlated" -- Stein and DeMuth have shown that 10-years running average of price measures for the real [inflation corrected] S&P500 _are_ predictive enough of the market's returns over the next 20 years that a long-term investor using those prices is much better off than one doing "buy and hold" instead, for example; yet this example of extreme long-termism is as contrary to the popular myth of the perfectly efficient, aka "not serially correlated" market prices, as any other bit of chartism, from the venerable Dow Theory, to the "momentum" craze, &c... I'm as intuitively averse to the chartism/technically mumbo-jumbo as clearly Fisher is, but definitely not because of dogmatic adherence to "efficient markets", which I find just as unconvincing... so, I'm studying up and refreshing on all sorts of such theories, of both ilks, to try and fight my own confirmation bias on the matter).

Rather, I think S.A. Nelson has it right in his "ABC of Stock Speculation" masterpiece. There are two main ways of speculating in stocks (in 1902, when Nelson's book was written, all trading and investing in the stock market was also called "speculation"): the only way that's ever made really great fortunes, based on deep study (and continuous re-checking) of firms' actual values -- and short-term, pure trading/gambling approaches where the underlying firm is hardly considered, and everything hinges on "what the market will bear"... the market price (which of course a lot of theorists, from Dow himself all the way to dogmatic "efficient market" theorists, will and do at times claim embodies all there possibly is to know about a firm). In the second case, stop losses are not a bad idea (though Nelson justifies them, essentially, by a model of the grand-scale stock manipulation that the short-term trader is trying to coattail-ride on).

If I get into a stock because I'm gambling that it will rise a lot soon, and it goes down substantially instead, then the stop-loss may indeed reduce (though never of course eliminate) my losses on the losing gamble. But if I get into a stock because I'm convinced after thorough study that it's really worth 50 and it's now priced at 35, then, if the stock's market price further decreases to 25 (and, re-checking carefully all my reasoning as to why it's really worth 50, I'm confirmed in that opinion), then the stock is now even more of a bargain, and quite possibly some funds that were previously otherwise employed should be freed to purchase more of the stock in question. While perhaps not "efficient", the market will eventually sync up with a firm's real value -- the patient value investor counts on that! In this approach to investing, stop loss orders make absolutely no sense!

The old saw "ride your winners, cut your losers" only makes sense in a pure trading, not investing, perspective, as "winners" are intrinsically defined here as "stock which increased in price after I purchased them" and vice versa for "losers". That's what makes it particularly sad to see the very worst piece of advice in Montier's "Little Book of Behavioral Investing" -- that "stop losses may be a useful form of pre-commitment that help alleviate the disposition effect in markets that witness momentum". Only for a pure trader, somebody who needs to accept the current market price as the total determinant of the underlying firm's overall worth -- which is really very contradictory with most else that Montier is saying in that otherwise decent book...!