For investment purposes, a company's output may be defined as cash it deploys in order to grow the business or reward shareholders.
R&D, capex, advertising, dividends, buybacks and acquisitions meet this definition.
Using per share amounts and assigning a multiple to each output, the results can be summed to produce a valuation.
Back-testing validates the resulting metric as a stock selection criterion.
Let's start with a back-test. The period chosen, from 1/1/2008 to the present, verifies that this method of selection would have provided serious outperformance before, during and after the financial crisis.
11% alpha is a very appealing statistic.
The screen is as follows:
- Member of S&P 500
- Dividend greater than 2%
- Sum of Outputs Valuation > Price X 0.85
- Sorted by 5 year average ROI and the top 20 selected
- rebalanced at 3 month intervals
Sum of Outputs Valuation
The formula is somewhat complex and relies on custom variables created in Portfolio123. Here's a spreadsheet:
IsNA is a way of dealing with blank fields. SharesFDQ means shares fully diluted most recent quarter. The rest of the field names are self-explanatory.
Excess Current Assets is set up to reward a current ratio greater than 1.5. Long Term Debt is a subtraction, I used the 0.5 multiple because debt can often be beneficial so I didn't penalize it heavily. Capex as used here is net of depreciation: a company that doesn't have capex greater than depreciation will be penalized. Buybacks is net of issuance. Buybacks and Acquisitions are not always positives and may not repeat every year. As such, I assigned a low multiple to these items. Company financials usually relegate advertising to the verbal part of the filing. For that reason, the information isn't available to the database and no provision for this expense was made.
The multiples were pulled out of the air. It would be possible in theory to do multiple linear regressions to determine average values. I made some efforts along those lines, without producing any meaningful information. I didn't tweak the multiples much, because the initial guesses were close enough to give a very satisfactory result.
The 0.85 factor which was applied as a hurdle for selection based on price is relative. It was selected because it passed a sufficient number of prospects to enable at least 20 stocks in portfolio for the time period involved. There is a period from about 2002 to 2006 where it would be necessary to use a lower factor, around 0.65, to get enough stocks to pass the screen.
5, 10 and 15 year back-tests all produced favorable results, using the screen as shown above. For the 2002-2006 period, better results could be achieved by adjusting the relative price cut-off from 0.85 down to 0.65. Otherwise, the screen produced too few candidates to maintain diversification and avoid buying situations where the stock was cheap for a reason.
The screen produces a lot of Information Technology, perhaps because of the key role of R&D. By increasing the maximum number of stocks from 20 to 40, it's possible to bring in some utilities and increase the representation of industrials and materials.
I'm long four of the stocks listed: Intel (NASDAQ:INTC), Baxter (NYSE:BAX), Coach (NYSE:COH) and Altera (NASDAQ:ALTR). I've written bullish articles on Intel, Baxter and Altera, and my opinions are unchanged. I have an unrealized loss on Coach, and have sold covered calls at the 38 strike, which will lock in the loss if called away.
The screen is not all-purpose: the candidates produced are worthy of further consideration, due diligence in the usual way. It performs better in down markets than up, and may be regarded as defensive in nature. Performance would have been outstanding in the wake of the Tech Bubble, as well as during the Financial Crisis.
My current investment thinking is focused on cash deployment: I want to look at what a company is doing to grow the business and reward shareholders. The analytical task involves forming an opinion on the efficacy of capex and R&D, whether a company is overpaying on buybacks, whether debt is being used constructively or for potentially troublesome financial engineering, whether cash is being hoarded or held at safe levels, etc.