Of the ninety-two stocks in the Low P/E Bin as of yesterday, seven of them have a positive 52-week price change. Two of those seven are limited partnerships in the business of distributing propane - "Hank Hill" companies. A third is a petroleum pipeline company. The other four are a mixed bag: a Canadian insurance company, a British pharmaceutical firm, a Columbian alcohol seller and an American phone company - one of the few "fallen angels" in the communications-services industry. Here's the list of all seven:
- AmeriGas Partners, LP
- Suburban Propane Partners, LP
- Sunoco Logistics Partners
- Fairfax Financial Holdings Ltd
- AstraZeneca plc
- Compania Cervecerias Unidas SA
- Qwest Communications International
The one that gained the most over the 52-week span? Cervecerias, the Columbian booze company.
Admittedly, with this metric, the Low P/E Bin doesn't stack up very well with respect to the overall market. Using Google Stock Screener, I found 1532 companies with market cap over 500M that aren't ETFs. Of these, 265 were in the positive 52-week-change category: 17.29%. The ratio for the Low P/E Bin is a much smaller 7.692%.
In other words, a randomly-selected non-ETF stock with market cap over 500M has a 0.1729 chance of having a gain over the past 52 weeks. One from the Low P/E Bin has only a 0.07609 chance. To have a 50% chance of getting at least one positive 52-week'er from the Low P/E Bin, you'd need to draw nine times from those 91 stocks with replacement after each draw. [Actually, the chance is 0.5095, but it's the closest to 50%.] Nine draws from the larger set, with replacement, have a 0.8190 chance of getting at least one 52-week positive: almost 82%.
More to the point, a random draw of ninety-two stocks from the larger set has only a miniscule chance of winding up with exactly seven 52-week positives - far lower than 1%*.
Conclusion? The Low P/E Bin is a set with relatively fewer stocks making 52-week gains than its corresponding universal set. I don't know how much influence this finding has on the low P/E undervaluation anomaly, but I can see it having some...
"These things? All of 'em are down except for no-growers** There's no way I'd put my money into any of 'em!"
*: Through using an oversimplified and somewhat inaccurate model, I got 0.3977%. I used this formula: p7*(1-p)(92-7) * 92!/(7!*[92-7]!). The formula calculates the odds of exactly 7 fifty-two-week positives, and (91-7, =) 84 fifty-two-week negatives, in any combination. p, which equals 0.1729, is the chance of picking a 52-week gainer from the larger universe of 500M+ market cap non-ETF stocks in Google Finance's database. Since this formula does use the p-value for an infinitely-sized universe, my math is off. Trouble is, I'd have to go through 8,760,554,088 different combinations, with the precise odds calculated for each, to arrive at an exact answer. A finite set does that to ya.
The computation chores were made possible by the factorial calculator on this webpage. The others I tried sent me to Overflow Junction for some three-and four-digit factorials I used in some preparatory calculations.
**: Apparent non-growers, that is. One of the seven listed above, AmeriGas Partners, had a 10-year logarithmic-regression-derived earnings per share growth rate (annuals' EPS from continuing operations, or EPSCO, growth rate) of 20.08%. Suburban Propane's 10-year EPSCO growth rate was 16.34%. Since both are limited partnerships, and LPs are basically designed to whittle down shareholders' equity over time, calculating the 10-year return on equity for both wouldn't give a meaningful answer. Suburban tends to increase its distribution in line with EPSCO growth, while AmeriGas tends to tuck some away. For what it's worth, Amerigas' 10-year ROE, using a 10-year sum of earnings divided by a 10-year sum of shareholder's equities, is 33.46%. Because of the whittle effect, though, it's larger than an equivalent corporation's would have been.
Fairfax's EPSCO growth rate is 24.07%, but its 10-year average return on shareholder's equity is only 9.615%. Both of these results come from numbers in Canadian dollars as measured by Canadian GAAP measures. So, these ratios are Canuck all the way. Ye hast been warned.