One of the nicer things ever said about the kinds of analysis we do here at Political Calculations recently surfaced over at the ChicagoBoyz blog, as Grurray took on the challenge of explaining the unique brand of stock market forecasting we do.
We can look at ways of fundamentally gauging the valuation of the stock market such as price to earnings ratio or the so-called Warren Buffet Indicator of total market cap to GDP ratio. I like to look at the Q ratio which is a simple comparison of the total price of the stock market to the replacement costs of all companies listed. This is the favored metric of billionaire black swan investor Mark Spitznagel, who by the way wrote a most excellent book, The Dao of Capital, about Boydian investment strategies.
By this measure, the market looks to be at a pricey level compared to other points in time. 1907, 1929, 1937, and 1968 were all years when the stock market peaked and saw a significant decline. The problem is it’s also at the same level as 1997, which had a small pause before marking the half way point in multi-year rally. We generally have seen regression to the mean in the past, but that doesn’t necessarily suggest it has to ever happen again. We could be waiting a long time for a sanity check to take hold, especially if the definition of sanity has changed.
A shorter term answer possibly comes from the world’s best econometrics blog Politcal Calculations. They believe, convincingly in my opinion, that expectations for future dividends drive stock prices in the near future, absent any surprising shocks to upset the apple cart. Those of us who used to watch Larry Kudlow on CNBC (since his show was cancelled there hasn’t been any reason to watch that silly network anymore) remember he used to say ‘earnings are the mother’s milk of stocks’. Well if that’s true than dividends are your father’s pemmican.
What they do is take values of dividend futures traded on the Chicago Board of Options Exchange and apply a multiple (and some other math) to convert them to expected stock prices. Their calculations show a possible slide in prices for the next few weeks to few months. It has worked reasonably well in the past with a few caveats.
There are different instruments traded for different times in the future. Prices can and do take leaps from one trajectory to the other. It usually happens when someone from the FED talks about raising rates, and then the financial press speculates what specific month or quarter it can happen. In this way, stock prices behave similar to quantum particles bouncing from one energy level to another. It’s not a good way to pin down exactly where stocks are going but just gives a range.
The other caveat is this measurement only works when the market is in a state of relative order, and not buoyed or rattled by some overly cheery or dreary news. While at a smaller level the market seems to obey quantum mechanics, at the macro level it acts like a natural system, following mathematical probabilities such as those observed in predators hunting or even groups of people foraging. The market moves from more easily observable and predictable periods until the forageables (earnings and dividends) run out, in which case it moves into chaos and unpredictability until new expectations are established.
What will trigger rapid moves in either direction and out of the current financial horse latitudes is anybody’s guess. There’s a big vote in Greece this weekend, but how many times has that situation reached a cliffhanger? Perhaps too many to matter anymore. As unsatisfactory as it sounds, what usually occurs is something we weren’t expecting, not an event that seems to replay itself over and over again. The best we can really predict is that we won’t be drifting forever, and the time will come when the stock market will move far away from this level. The key is to stay ready for it when it finally does.
Here's the link to our stock market forecasting math and here's our math-free description of how stock prices work.
Now, to address Grurray's "other caveat"! When you get into our model, you'll find that our math actually does continue to work when stock prices go on what's called a Lévy flight, which is the pattern that Grurray is referring to when he links to articles describing how both predators and people adapt their hunting and foraging methods when food is scarce. We know it continues to work in that situation because we invented behind our analytical methods in the latter months of 2008, when stock prices were collapsing (we publicly introduced the discovery we made that enabled us to develop our forecasting methods on 10 December 2008).
Behind the scenes, in the early months of 2009, we were finding that we could anticipate where stock prices were heading as they were crashing after they had entered into Lévy flight, because what was changing was the expectations for dividends in the future, changes in which preceded subsequent changes in stock prices, which is what allowed us to confirm the direction of causality in the relationship between the two.
Not to mention being able to quickly confirm just a week after the fact that the bottom had been set for that particular Lévy flight for U.S. stock prices. We later proved (again, behind the scenes) in late 2012 that our model is capable of handling the opposite situation. That latter episode also confirmed that the quantum nature of those expectations always applies - even during Lévy flight.
Where Grurray's other caveat is correct is when stock prices are driven by factors other than their basic fundamentals. The classic example here actually predates our model's development by a number of years, where we don't believe that our forecasting model could have tracked stock prices very well without considerable adjustment during the period of the Dot Com Bubble. That event was caused by a disparity in the tax rates that applied to capital gains and to dividends, which greatly changed the results of the investment rate of return math for dividend versus non-dividend paying stocks for investors so much that it continued to distort the entire U.S. stock market until the disparity between the investment tax rates was finally fixed in late May 2003.
Unfortunately, we are not able to put our model to the test of the Dot Com Bubble because we lack the historic dividend futures data during that period of time, which prevents us from being able to accurately re-create what the future looked like in those days.
Other than those significant non-fundamental stock price driving factors, the rest for us is pretty much noise - things that actually are the result of investors reacting to the relatively random onset of new information as it becomes known, the speculation associated with which accounts for much of the apparent day-to-day random walk of stock prices (you know, like the reaction in the U.S. market to what's going on in Greece). We spent several years after our original discovery of how stock prices work collecting the necessary data and documenting the ongoing context behind it to quantify that aspect of how stock prices behave, which allows our model to determine not just the general future trajectory of stock prices, but also the likely ranges we can reasonably expect stock prices will fall during periods with "typical" levels of noise in the market.
We've come a long way. We've gone from only being able to anticipate what the average prices of stocks would be during the next month to accurately anticipating the actual trajectory of stock prices within a small margin of error on each day of the next month (as we've previously demonstrated under optimal conditions). Or if you prefer longer range forecasts, we can reasonably anticipate the general trajectory of stock prices several months in advance.
But not much more than that. Although we can project the alternative trajectories that stock prices are likely to take on a daily basis as much as a year into the future, it is far more likely that the expectations we use in our forecasting model will change well before that distant forecast future arrives.
After all, a man's got to know his model's limitations!