State Space Models

All state space models are written and estimated in the R programming language. The models are available here with instructions and R procedures for manipulating the models here here.

Tuesday, February 22, 2011

Charting vs. Random Stock Walking














Random Stock Walking is a form of technical analysis. On Mad Money tonight, CNBC analyst Jim Cramer did a good job of analyzing the stock OCLR (Oclaro, Inc.) using charting concepts, another approach to technical analysis. In this post, I'll compare the two approaches.

Oclaro, Inc. is a provider of optical network components. Cramer thinks that the stock has a bright future given the future of telecommunications and the lack of competitors. Cramer also thinks that both the fundamental and the technical analysis support the positive forecast.

In the video above, Cramer presents a technical analysis that uses some of the following concepts: trend lines, head and shoulders, and relative strength index. As a statistician, I'm not sure what to make of chartist concepts. Here's a start on that project.

From the standpoint of random stock walking, OCLR is a stock that is hard to distinguish from a random walk. The business-as-usual model (OCLR[t] = 0.49 + 0.85 OCLR[t-1]) with 98% confidence interval [0.70, 0.85, 0.94]) shows that a random walk is somewhat improbable. The graph above displays the dynamic attractor for the model. Notice that 2006 to 2010 draws most of Cramer's attention with a lot of weight being given to the "break out" after 2010.
Given a dynamic attractor that is below the current stock price, it's not surprising that my future forecast is not very encouraging. There aren't many analyst opinions on this stock (here) and the one analyst opinion available suggests a high price target of 23, which is about the upper 98% bootstrap prediction interval for the stock.

The Random Stock Walker approach finds little evidence for a "sky-is-the-limit" breakout that Cramer's technical analysis is suggesting. It will be interesting to follow this stock over the next few years and see where it goes. Cramer does admit that it is a speculative stock.

Thursday, February 17, 2011

Worry, Worry: Steve Jobs Six Weeks To Live?













The National Enquirer (that's right) is reporting that Steve Jobs has six weeks to live (here). Stock analysts are keeping a level head in the face of this "news" (video above). One analyst on Yahoo Finance (here) downgraded AAPL from "Strong Buy" to "Buy" and another analysts upgraded the stock from "Neutral" to "Buy".

The stock was up to 364.90 yesterday but dropped down to the 358 area for most of the day today. The Random Stock Walker models (here) suggest that a stock price between 300 and 350 will be very close to the dynamic attractor for the next few months.

Wednesday, February 16, 2011

Should You Hold On To ARM Holdings?

CNBC stock analysts Jim Cramer recently pick ARM Holdings (ARMH) as one of his growth stocks (discussed in a prior post here). Unfortunately, the Random Stock Walker didn't validate ARMH as a growth stock even though it did validate the other stocks Cramer picked (AAPL, CMG and NFLX). What's going on?

First, as a company ARM doesn't actually make anything! It is a supplier of semiconductor intellectual property. It designs chips used in smart phones but it then licenses the designs to semiconductor manufacturers. ARM provides economies of scale in R&D in what has become a commodity market (semiconductor manufacturing).

Because ARM technology is so central to smart phone design and because cell phone technology is switching from "dumb" to "smart" phones, ARM should be well positioned to ride the "mobile internet tsunami".

If you look at the Random Stock Walker models, ARM is clearly a business as usual (BAU) model driven by shocks and a "short-memory" for the stock price (see THEORY below). It is not a growth stock (long-memory for shocks and stock price) and it is not particularly well linked to the U.S. economy. Therefore, even though the performance for ARMH (in the graphic above) shows improbable growth (beyond the 98% prediction interval) for early 2011, the forecast for the rest of 2011 is for a return to the long-run attractor at a price between 20 and 25 roughly.

Analysts don't like this stock very much either (here): it has a high P/E ratio, eight-out-of-ten analysts have recently downgraded the stock and given it a price target between 10 and 12!
Cramer acknowledges that people got burned by this stock during the dot-com bubble as can be seen from the long-term Random Stock Walker forecast above. Yet, he thinks it is a compelling story. Only time will tell. From the modeling perspective, this stock does not behave the same as other momentum or growth stocks and should not be put in the same category statistically.

THEORY. The BAU model is P[t] = a + b P[t-1] + V where b is less than unity. A pure growth stock would differ in that b would be greater than one. In the random walk, of course, b is (in probability) precisely unity. Since b=1 is actually quite unlikely, what the random walk designation should mean is that the stock price is dominated by random variability and cannot be effectively forecast.

METHOD. Looking at the behavior of ARMH during the dot-com bubble around 2000, one approach would be to eliminate this "shock" from the model and re-estimate. I've re-estimated the model starting in 2003 and the results are fundamentally the same as reported above.