Subject: Re: Diversifying away from Berkshire
If I have a list of the top, say, 30 stocks as ranked by the given criteria, how much can I hurt my return by leaving out five based on some personal bias?

you'd be surprised : )

If you eliminate several from the slate at random, it is extremely unlikely to make a meaningful difference.
But if you pick them judiciously, it can hurt an amazing amount. Maybe.
Murphy's law for quant investing: the ones you eliminate are the ones that will do best.
Apparent junk has a habit of popping when you least expect it.

Certainly a quant slate can be useful simply to generate ideas for further research, if your intent is to research a firm deeply and make a 'rifle' type investment.
A biggish position you know and trust well, on the punch ticket theory.
There is nothing wrong with using ANY method to generate lists of firms to read about. Stealing ideas is probably the best method.
But this is the situation where you really study and understand the firm before investing.

At the other extreme, investing in a quant slate can make good sense, as Mr Graham noted on many occasions.
Think of an entire slate of stocks as you would any single 'punch ticket' stock.
Spend the same amount of time studying the quant technique that you would spend studying a single 'punch ticket' stock.
And think of it correspondingly as a comparable portfolio allocation.

But...the middle ground can be tricky.
A little stock analysis, like a little knowledge, can be a dangerous thing : )
The only thing I do in this regard is eliminating a few categories of stocks entirely.
For example, Chinese domiciled ones.

FWIW, I believe these would be the current picks, in descending order of reported rate of growth of sales per share.
LNG CQP KLAC LPX IT MA ABBV AAPL BX IDXX
CDW SHW SMG CAH MCO WAT DVA SPGI TPX AMGN

Note, the method I tested uses data fields that are updated only annually. That's one of the reasons the turnover is so low.
Thus it's likely that some of these stocks do not have good ROE or good sales growth at the moment.
This is on purpose...the key thing is that it reproduces the same 'warts and all' data source that I tested (with hindsight!) over the last 37 years.

It's interesting to see several Berkshire holdings in there.
Apple of course, but also Mastercard, Moody's, and Davita.

Regarding the turnover, if you reconstitute the portfolio every six months, you'd trade about 6.5 of your 20 stocks each time on average.
I suspect many would stay on the list for years at a time.


There is one minor flaw with this screen: I think it would be useful to include those firm with positive earnings but negative reported ROE because of negative shareholders' equity.
These typically get reported as firms with a negative ROE, but from an investment point of view it's more like an infinite ROE.
They need no net assets at all to produce their earnings: the perfect company, in that one way.
Some really fine firms fall into this category, and they perform well on average, but they get eliminated from the screen as described above.
Ideally I would do it like this:
First, for firms with regular positive earnings and negative equity I'd assign a notional high ROE of something like 100%.
Then sort all firms on high ROE, a mix of real ones and notional ones.
Some recent 'positive earnings, negative equity' picks might be:
SNBR SBGI DPZ SIRI NSP ORLY MSCI COMM MCK HLF LOW VMW SBUX TDG AZO

Jim