InvestorPlace Blogs is powered by Marketocracy. Marketocracy has authorized Investor Place Blogs as an official registrar for voting through Marketocracy's Investment Research Rating service. Registered members of InvestorPlace Blogs are linked with a Marketocracy account to establish voting power based on their performance of trading and posting on stocks.
I have to ask myself that question after reading in the WSJ that Thursday's peculiar market action, with good stocks down and bad stocks up, was caused by leveraged "Quant" Funds unwinding their positions. Quant Funds, as described by the article, run their portfolios by means of computer programs, assuming that stocks will conform to various statistical patterns and taking both long and short positions accordingly, meanwhile applying leverage in the form of borrowed money.
My system starts with the assumption that a stock's range, in terms of either its Price/Sales or Price/5 Year average EPS, is a lognormal distribution. I review ten years worth of historical financial information, and make the assumption that the high and low Price/5 Year Average EPS (or whatever metric I am using) are 4 standard deviations apart. I assume that the midpoint price of the stock is halfway between the extremes. My Excel workbook assigns a percentile to the stock's current price on that basis, and I make my stock selections from among stocks that are trading below their 30th percentile as I compute it, generally buying somewhere around the 5-10th percentile. I start selling at the 50th percentile, generally I am out by the 70th.
This relies on the assumption that the stock price will revert to its mean, the same sort of assumption that the Quant funds have been relying on for many years. I have reinvented the wheel. For the past three years, this system has worked for me when I was long the stocks involved. The reverse concept, shorting stocks that are high in their range as I see it, has been hazardous: I made a small amount of money doing it, but approximately 1 in 5 short positions went against me badly. The right hand side of a lognormal distribution can extend out a long way. Fortunately the SLO doesn't permit shorting, so I will stay out of trouble.
By no means do I make my decisions on a mechanical basis. To begin with, the minimum and maximum Price/5 Year average EPS ratios are sometimes so extreme that I discard some values as outliers. Also, many companies have "one time" charges or bad years that need to be looked at and left out or included in. Small changes in these assumptions can lead to big changes in the expected price ranges. Some companies have a bad business model or an obsolete product or technology and being cheap doesn't make them a good buy.
Frequently value candidates developed by this system have problems with slowing growth or shrinking margins. I operate under the rebuttable assumption that management will resolve the issues. This is where I read Management's discussion in the 10-K or 10-Q, to see if I can develop an affirmative answer, that they have identified some solutions to the problems and are actively pursuing corrective action. If they have a strong balance sheet and cash flow, they have the resources to perform their task; and, unless they misdiagnose the problem, improvements will occur over time.
Also, this system performs poorly when an industry is in transition, going from a series of fat years to lean years, or vice versa. For example, it got me out of various steel and energy stocks when those industries started to experience improved margins, which were more durable than I thought they would be. Homebuilders, which are now suffering loss of both revenue and margin, would be another case where I question if my methods are workable. This is a critical point where judgment is required.
The positive aspect of this approach is that I don't make a lot of serious mistakes working the way I do. A stock's price movement has three dimensions - direction, magnitude and timing. Direction I usually get right. This is a plus, because you need to go up 100% to compensate for a 50% loss. On magnitude, I don't catch a lot of big moves, but I catch the easiest part and reduce the number of round trips. To improve magnitude, I am working on making fewer trades, focusing larger amounts of my resources on my best picks. Like many value investors, I tend to sell early. Timing is a question of patience. I will wait for as long as two years before I give up on a situation.
To answer my own question, quantitative methods are a tool, a starting point, and not a substitute for qualitative decision making.
|