Lower is better in all of them. First equation gives a sensical return but the second one makes no sense with the top 50% of buckets having negative returns
Curiously #3 and #4 give the same return.
Why is the system spitting out such variable results on what should essentially be identical?
Yes, make a ranking system with those variables. Rank each of them at 100% and run it on the CAD universe for past 5 years, 13 week holding but I"m sure it holds for all
The only way that I could replicate that result was by changing the performance test to short.
One significant difference here is that we return NA for EV if EV would be negative. The idea is that this is the value that someone would pay to acquire the company. A negative EV means, usually, a ginormous cash pile. No one will pay for you to haul that off.
Anyway, because of that behavior of precalculated EV versus the calculation from its components, I would expect there to be some substantial ranking differences simply because the negative EV companies are removed in the P123 EV.
With this formulation, higher values are better. You definitely want to punish companies with subzero EBITDA when doing rankings instead of making them N/A.
Here are the problems with your formulae. 1. P123 has rightly rejected the common EBITDA measure in favor of operating income before depreciation, and the designers have improved that formula. It’s very close or equivalent for most companies, but there are some differences here and there that matter. 2. The conventional EV, as Paul points out, becomes NA for negative EV. This can lead to some really strange results when using EV-based ratios. Let’s say a company has a negative EV. Then its price goes way up. All of a sudden, it’s EV changes from -$5 million to $1 million and it suddenly shows up on your screen as a company with an INCREDIBLY low EV/EBITDA ratio. In fact, its theoretical EV/EBITDA ratio just got a whole lot worse when its price increased. Despite what Paul says, companies with negative EVs tend to outperform companies with high EVs in the same industry. So these are companies you don’t want to exclude from your formulae. Using the Max(5, $EV) formula makes companies with negative and very low EVs equivalent. 3. Ignoring minority interest and preferred equity doesn’t make sense. Those numbers need to be there. They make a big difference for some companies.
Lastly, the EV/EBITDA ratio, like the P/E ratio and price-to-book ratio, are so overused that they give investors like us very little advantage. They’ve been essentially priced into the market. I suggest using three-year rather than TTM figures, but also including some less conventional ratios like unlevered cash flow to EV, tangible book value to price, gross profit to EV, r&d expense to price, working cap to price, dividend yield adjusted for payout ratio, economic value added to price, forward sales to price, shareholder yield to price, and so on. A good mix of valuation ratios will get you much better results than just relying on one or two.
Thanks so much. Just testing out your formula quickly gives sort of a bimodal and roughly expected result.
Why doesn’t formula include market cap?
Is there a difference in the order (i.e. can I expect same results from writing it traditional order and keeping the traditional ranking direction
And if I wanted to get “EV/SALES” do I just use the denominator over SalesTTM?
P123’s MktCap is different from Price*SharesFDQ, which is the standard for valuation ratios. MktCap is geared more toward estimating a company’s size in general. There’s a thread or two about it somewhere.
There’s a big difference in the order in terms of negative numbers. If you want lower-ranked stocks to get higher results you have to put NA for all the companies with negative EBITDA. My version gives them lower ranks.