Wednesday, November 4, 2009

Sneak-Peek: Historical Series Metrics

Historical Data is difficult. It’s big. It’s often messy. It’s cumbersome to deal with more than a handful of series at a time. Price-histories are pretty well-handled in most financial information systems, but historical series data for Price/Earnings, EBITDA, 30, 60, 90-day implied volatility are usually a bit harder to get to.

Unless, of course, you have Bloomberg.

Bloomberg has one of the most impressive historical data systems on earth. Not many people realize this, but chances are good that if there is a metric (mnemonic) that yields a numeric data point for a security, Bloomberg keeps historical data on that metric. PE_RATIO, for example, can be tracked as easily as price-history. As can complex metrics like Volatility skews (1M_CALL_SKEWNESS).

As is so often the case, however, looking at one historical series can yield only so much information. What we really want to do is compare the properties of one series to another. For example, we may see that the P/E ratio of AMZN is at a historical high, but what of its peers in the sector (GOOG, SYMC, EBAY, MFE, YHAOO, EXPE)? Are their P/E’s also at historical highs? Does AMZN stand out?

We’ve been quietly developing some new tools to help answer questions like this for the next-release of AlphaVision™ for Bloomberg. We call this new functionality “Historical Series Metrics”. Today, I’m going to give you a sneak-peek.

The process for creating a Historical Series metric is similar to creating a filter metric or a calculation metric. Inside the Custom Metric Editor (Tasks -> Open Custom Metric Editor) you will find a new button labeled “New Historical Series Metric”.

Clicking this button will bring up the Historical Series Metric Dialog.

The historical Series Metric Dialog allows you to select the Bloomberg Metric (Mnemonic), the Date Range, and the series calculation type (Min, Max, Mean, Standard Deviation, Decile Latest, Percentile Latest). Above, you can see that I’ve selected PE_RATIO, I’m going to look at the historical time period from 8/4/2007 to 8/4/2009, and I am after the Percentile Latest, which will tell me what percentile of the last 2 years of data the latest value falls within. Let’s have a look.

Below is a screenshot of the Internet sector. AMZN clearly stands out. Why? AMZN’s P/E Ratio today (8/4/2009) is about 69.2. Our Historical Series Metric tells us that AMZN’s P/E Ratio today is in the 73rd percentile of all of it’s closing P/E ratios over the last two years.

What is even more interesting about this picture, however, is that AMZN is the only firm in the internet sector, whose P/E so high on a historical basis. Let’s look at this another way – what if we switch and look at the GICS Industry Group to which AMZN belongs: Retailing.

Whoa! That’s a surprise. Virtually the entire Retailing group is trading at its highest P/E’s in the last two years. Did you know that? I certainly didn’t. Right now, earnings in retailing companies are as expensive as they have ever been over the last two years.

PE_RATIO is quite interesting fundamental research data-point. But what about something a trader might look at day-to-day: Implied Volatility – Specifically CALL_IMP_VOL_30D.

In this image of all firms contained in the S&P 500, I’ve created a new Historical Series metric by cloning the PE_RATIO one and simply changing the metric to CALL_IMP_VOL_30D. I’ve set the color-bar to highlight the firms with the highest (green) and lowest (red) historical 30-Day Call Implied Volatility leaving those firms trading near their medians white. In general, one can see that very few firms are trading at historically high volatilities. Given the historic volatilities in the market a year ago, I think this makes good sense. If you are buying Vol today there’s a lot less out there to pick from. From this image, DYN, PCS, WPO, & CFN look like ones to keep an eye on.