Data Dive

Soda vs. Pop vs. Coke on Twitter

Mon, 16 Jul 2012 14:47:00 -0500

via Nathan Yau at flowingdata we find Edwin Chen using twitter to gain insights into current cultural trends, we see that “Coke” is the dominant phrase in ROW (Rest Of World), but has limited penetration in the US. This makes sense, since Coke dominates the international markets with 50%+ market share versus the 20%+ that Pepsico has.

Venture Capital Returns by Vintage

Fri, 13 Jul 2012 11:07:00 -0500

Drew Conway’s blog is worth following if you’re interested in the intersection of politics and data science. But I came across this old post he did of VC returns by vintage for a publicly released Calpers data set and thought I would link to it here.

Exploring the Tour de France with R and ggplot2

Thu, 12 Jul 2012 19:57:00 -0500

Exploring the Tour de France with R and ggplot2:

Some stunning visualizations of the Tour de France.

The Banking Industry Returns Thru the Financial Crisis

Thu, 12 Jul 2012 19:35:00 -0500

I recently took some high level data from the companies covered in the Value Line Investment Survey (both the Standard edition and the Small and Mid-Cap edition) and decided to plot it out to see how bank valuations have flexed over the last several years. All data is courtesy of ValueLine (from 2003-2011) and all plots were made with ggplot and R. This isn’t the most rigorous of analysis, but it was pretty useful for some projects I’m working on. Since this only analyzes banks that are currently in the survey, it suffers from survivorship bias. Given the state of the banking sector today and its recent history, this is an issue.

The first graph charts Price to Book versus Returns on Equity for all the banks in the survey. I’ve stratified out the banks by large cap, midcap, and smallcap.

You can see that there are a large number of companies that trade above or below the trend line, but the trend is pretty strong regardless of market cap size.

Next we look at Price to Book versus Returns on Equity by year.

Again, the trend is very apparent, but the slope of the effect is diminished as valuations get compressed post-crisis. You can see that as we went through the financial crisis, the market began paying less of a premium for superior performance. As investors became less trusting of bank balance sheets, they started discounting the price. The idea of paying 4x book value for a bank is laughable today, but in 2003 or 2004 seemed entirely rational.

The next chart is a histogram of Price to Book by year.

This gives you a good sense of the intense clustering of value in the banking industry. In short, regardless of capabilities, most banks trade the same today on a multiple basis. This could mean there are opportunities for savvy investors to play a long-short strategy effectively here.

As further evidence of this, check out a boxplot of Price to Book values by year, stratified out by small, mid, and large cap. Not only have the standard deviations compressed, but so have the number of outliers.

Finally, let’s see how much ROEs have flexed through the crisis.

Couple of interesting points to make here. First, large caps seem to have a better ROE performance than midcaps and smallcaps as evidenced by the tighter variation and higher median. ROEs ranged in the mid-teens for all banks in the 2003-2005 range and then large caps really outperformed in 2006. 2007 was the year when write-downs and capital impairments began to surface and you see the deviations of all market cap sizes explode and by 2008 you begin to see the large cap banks coming to grip with reality before everyone else and take huge writedowns. The 2009-2011 period looks like the beginning of a return to normal performance. If you believe in the loan quality of the banks, they would be a steal here. Obviously, the Street doesn’t agree with that assessment.

Analysis of the court strength of the Miami Heat vs. the OC Thunder

Tue, 12 Jun 2012 15:57:16 -0500

Analysis of the court strength of the Miami Heat vs. the OC Thunder:

Via the New York Times comes this stunning visual analysis of the courtside strengths of the respective teams. I think what’s most amazing is how close the teams are statistically (47% accuracy for Heat vs. 47.1% accuracy for the Thunder) on a team wide basis. However, when you start diving into the stats you see that the Heat plays the midrange more than the Thunder.

Also, Durant and Westbrook have taken almost 50% of the shots for the Thunder. The Heat seems more well-rounded as a team, offensively, though LeBron has taken almost 20% of this season’s shots. Given his 53% accuracy, this makes sense.