Using a 5 year ROE/ROA: my understanding of the business cycle is that its usually 5-7 years, sometimes up to 10 years.
Have longer business cycles in general (or in any particular sector) ever presented any problems, e.g. outlier results?

That would be a very good question when looking at a single firm. But in "shotgun" investing, the goal is simpler, merely to reduce the number of dumb picks due to a misleading metric. In the case of ROE, the two main things to watch out for are overleverage, and a single year even that makes the numbers look unusually good. Any multi-year average, even two years, would cut down on the latter problem. This is partly because you aren't looking for a single company nor even a single good ROE, but comparing a firm's ROE to those of other firms--if there was a recession, it will have hit them all.

Given the choice I would probably use 3 years, maybe 4 years as my goal, rather than longer than 5 years, as businesses do change and you want *passably" up to date metrics.

"Gap: 8.5%/year, nothing to be sneezed at for something so simple."
I remember you have mentioned this gap several times over the years.
Do you know offhand, to what extent that gap relates to each of a) valuation/multiple expansion and b) intrinsic value improvement?

The gap that I mention might not be a good reliable number in future, but it's indicative of a real effect.
I think it comes from the following observation: all good firms have high returns on equity. After all, that's what you want if you start a company: a good return on the money you put in. The reverse isn't true; you can't say that all high ROE firms are good businesses. But if you do require a high ROE, you're eliminating a lot of firms which are both weaker and *obviously* weaker, which shifts the odds in your favour. For shotgun investing, that's about all you need.

More subtly is WHY a persistently good ROE works. Equity here is being used as a good-enough proxy for the cost of a competitor to enter into the same business. If you have a high ROE over a long period of time, and you are in a business subject to competition, others will notice. Sooner or later someone will raise some money and go into business and undercut you, to which you will have to respond. No matter how you respond, your profitability will fall. So, if you have a high ROE that is *not* falling over time, this is very good prima facie evidence that your business has substantial barriers to competitive entry: a moat. The very best firms will have (a) not too much leverage, (b) a very high ROE, (c) ROE that is consistently high over many years with no declining trend, and (d) shareholders' equity rising on trend at a good rate. The last is wonderful: not only are they getting a great return on capital, they are able to get that same high rate of return on newly deployed capital. This was true of Walmart for many years, for example. 20% ROE on $20 billion of equity, then 20% ROE on $40 billion of equity, then 20% ROE on $60 billion of equity...pretty soon you figure out it's a good business.

FYI, my quick and dirty leverage test: a safe firm will have long term debt less than 5 years of net earnings. I usually use earnings rather than assets--ever tried to make a loan payment with a chunk of a factory? For the very best firms with the most reliable earnings streams, I might push that debt level as high as 10 years of earnings. I used to count Hershey in that category, as an example.

One other thing about ROE: a firm with pretty consistently positive earnings but negative book value in effect has an infinite ROE, not a negative one. Such firms should sort to the top of your screen, not the bottom--you can arbitrarily assign them all an ROE of 100%. Some firms are so resilient that they need no assets at all to earn a living. Moody's, Coke, tobacco firms (ick), McDonald's, Oracle.

This sort of statistical edge pairs well with a judicious bit of screening to avoid just a small number of the worst disasters: the ever popular crap filter. Nothing too strict, just skip the most egregious. It's amazing how much better a portfolio does if you dodge just a few big losers.

Jim