There is no doubt that confidence in any method (valuation or otherwise) is increased with long term successful use.
I tend to think this can be broken into two aspects.
To be sure, any method worth discussing here has some elements which take some practice and honing of skill. Those like FCF that involve a certain amount of computation or spreadsheet use will take some work to those who are not familiar with spreadsheets or the type of computation. Those ... all ... require honing of skills in reading financial statements to know what to include and what to exclude from various figures of interest. This aspect is the transition from "gee, that sounds complicated" to "oh, I've got it now" to comfortable use.
But, the other element in confidence is not just one's own use. After all, any valuations that any of us made today would not be proven good or bad for some years to come. And, quite frankly, since today's judgements are likely to be supplemented by many other judgements between now and then with an evolving perspective during this time, it is unlikely that five years from now we will have any more than an anecdotal sense of whether today's judgements were the right ones.
However, this doesn't mean that we can't take a candidate method, roll back the clock 5 years, apply it to a test company as it was then, and then roll the clock forward, repeating the test once a year or so, and see how the sequence of valuations corresponded to the actual history. Given the forum, we would want to do this for several gorillas of different types and then we would want to do it for other one-time candidate gorillas which obviously haven't made it. We would see how the analysis made the companies look in pre-bubble, more "normal" times and how the company was changed in appearance during the bubble. We would ask ourselves in this hindsight analysis whether we would have received the kind of buy or sell signals at appropriate times and we could model what responding to those signals would have meant to our investment in comparison with holding all the way through.
Having done such a test, I would still have to worry whether the method only worked because of the peculiarities of the last 5 years or whatever I chose to test and whether the next 5 would be at all similar, but any method which did not succeed in giving me good signals in this test wouldn't seem like a very good candidate for a method going forward, would it?
While I am in favor of making valuation tools available, and would urge their consideration by those expressing dissatisfaction with their results, I know that any such tools can be "mis-handled" to give misleading answers. One learns proper handling by practice, not watching. Those with incentive to do so, will practice.
I sure have no argument with you there. But then why practice on something which can't pass a test like this? Why adopt a method going forward which can't handle predicting the past? And, unless that test is made, how do I have any idea whether the numbers or signals it is giving me have any value? |
