Another screen for the skeptics, those who think quant investing has no relationship to sensible value investing and therefore can't "work".So, let's start from first principles.People put money into a company (equity) to earn a return. More return is better, less is worse.In particular, more return for a given amount of equity is particularly good if you're the one putting up the equity.Express the return as a percentage of the equity it's called the return on equity (no!) or ROE.The most important thing about any good long term investment is that it have a sustainably good return on equity. All else is a means to that end.Obviously you want that to be accomplished using a good business model,not merely with risky leverage, so adding a leverage cap is a good idea.When looking at individual companies you'd look at the way the firmactually goes about its business to ensure that the good returns will be able to continue in future, but hey, this is quant world, and we'll simply skip that and let the failures be papered over with a bit of diversification.We need only ensure that the average is good enough, and that we never bet the farm on one stock.A truly great business can allocate new capital within its same areaof business at the same high rates of return, but not all firms withhigh returns on equity can do that. If their management is smart, they pay it out (or perhaps do buybacks) rather than invest in low return expansion.This is known as a cash cow. They can make very nice investments, even if they are "second best".The WD40 company is sometimes cited as an example of such a firm.There are all kinds of ways to try to make money in equities, but fortoday let's concentrate on just those firms that do have very high returnson equity but don't necessarily have new ways to invest that capital.To recap, dumb ones will be reinvesting the money at low returns, smart ones will be paying out the earnings.We'd prefer the smart ones.So, the two key ingredients that might give away a cash cow is high ROE (without undue leverage) and a great dividend.Throw in a bit of a final sort on momentum because, hey, this is the mechanical investing board, right?It's no secret that companies in fashion with the market do betteron average in the short term than those that are out of fashion.The market is not always wrong; sometimes prices are falling for a reason,and a bit of momentum allows us to pick up on the [meagre] wisdom of the crowd.So, we have a genuinely simple screen:- Find the top 50 stocks ranked by the formula [ROE times dividend yield].The multiplication means both factors have to be high at the same time, not one low and one very high which would pass the test if we just added them.- Eliminate those with high leverage, "high" defined here as having total assets over 4 times shareholders' equity.As a side effect, this is a screen that will never pick a bank.- Take the top few stocks ranked by total return in the last year.That's it!Check out the equity curve graph on this fella.Follow the link, click "run", and scroll down to look at the picture. Trust me, it's worth it.http://www.backtest.org/CashCowROE:8912SBenwG0IcdyG0ItasG0In... Screen S&POverall 23.3 9.5 1989 72.5 35.2 1990 -8.6 -4.8 1991 54.6 30.8 1992 -2.0 7.1 1993 38.8 10.2 1994 -3.8 1.5 1995 33.1 38.7 1996 8.4 23.2 1997 39.5 27.9 1998 16.5 34.5 1999 -1.2 18.4 2000 30.7 -10.8 2001 31.8 -9.3 2002 7.7 -22.1 2003 28.6 28.5 2004 33.8 10.2 2005 44.7 7.5 2006 7.5 13.8 2007 22.5 5.6 2008 0.1 -39.4 2009 59.6 32.7 2010 37.1 15.2 2011 30.0 2.5 2012 18.9 14.2It's not very brittle at all. Taking the top 30 or top 100 by the ROE*DY metric works nicely too, and that's really the only magic number.A slightly tighter leverage cap might be good, say 3.7x, but a simple 4x works well enough.I've done a GTR1 test (average of all daily starts) and that's nice too.It seems to indicate that top 100 might be better at the sort step,at least when using a smoothed momentum sort which is what I usually do.e.g., I used ratio of [average price in the last month] to [average price 7-13 months ago] as the sort.This kind of smoothing reduces turnover, improves returns, and generally increases the steadiness of the returns (lower downside deviation).What about variations?Some people like to look at return on assets (ROA) rather than ROE. This takes out the benefit of leverage entirely, so the leverage check isn't necessary. This works too.Just replace the ROE field ("enw" in backtest.org speak) with ROA = net income / total assets ("nin/tas" at backtest.org).http://www.backtest.org/CashCowROA:8912SBninG0IcdyG0ItasG0XO...Again, check out the pretty equity curve.One of the nice things about cash cows is they don't change much.Both screens works with longer holds very nicely.CAGR 23 monthly, but still CAGR 19 with an annual hold (January start).In fact in many cases it does better with two month holds.Because of this good result at longer holds, it makes a nice "dozens" screen.Each two months, sell what you bought 6 months ago and buy the top-ranked stock in the current rankings that you don't already own.This gives 3 staggered separate 1-stock portfolios added together.Works well with the ROE version http://www.backtest.org/CashCowROEdozens:8912h2DZ%28CashCowR.....and also with the ROA version http://www.backtest.org/CashCowROAdozens:8912h2DZ%28CashCowR...Since those sometimes give slightly different picks you can blend those two together.Not a lot of distinct stocks at any given time, usually either 4 or 5, but the result sure looks nice.http://www.backtest.org/CashCowDozensBlend:8912BL%28CashCowR...I for one sure wouldn't have minded a 16% profit in 2008 in an all-equity portfolio.In all of the above screens, look at the steadiness of the equity return graph.One of the nicer things is that it simply doesn't seem to notice thatthe market has gone net nowhere since 2000.e.g., the blend beat the S&P by 28%/year (screen) to 2%/year (S&P) 2000-2012.http://www.backtest.org/CashCowDozensBlend:0012h2BL%28CashCo...But getting back to the beginning, I like the simple unadorned CashCowROE, the first link above.http://www.backtest.org/CashCowROE:8912SBenwG0IcdyG0ItasG0In...What you do with the picks is pretty flexible.No, real world returns in future won't be like this, and there willinevitably be the occational plunge where the portfolio shrinks by half.That happens with all equity strategies, including the broad US market.But I suspect these screens will be materially market beating on average in the next decade.Jim
Screen S&POverall 23.3 9.5 1989 72.5 35.2 1990 -8.6 -4.8 1991 54.6 30.8 1992 -2.0 7.1 1993 38.8 10.2 1994 -3.8 1.5 1995 33.1 38.7 1996 8.4 23.2 1997 39.5 27.9 1998 16.5 34.5 1999 -1.2 18.4 2000 30.7 -10.8 2001 31.8 -9.3 2002 7.7 -22.1 2003 28.6 28.5 2004 33.8 10.2 2005 44.7 7.5 2006 7.5 13.8 2007 22.5 5.6 2008 0.1 -39.4 2009 59.6 32.7 2010 37.1 15.2 2011 30.0 2.5 2012 18.9 14.2
Best Of |
Favorites & Replies |
Start a New Board |
My Fool |
BATS data provided in real-time. NYSE, NASDAQ and NYSEMKT data delayed 15 minutes.
Real-Time prices provided by BATS. Market data provided by Interactive Data.
Company fundamental data provided by Morningstar. Earnings Estimates, Analyst Ra