Basics: Measuring historical seasonality in indices to explain price curves on futures

As a long-time viewer of Case Shiller forward curves I have to admit that the first quoted prices on the RPX contracts startled me.  I had always known that different real estate indices had different degrees of seasonality, but the RPX futures curves were a sharp contrast to the Case Shiller graphs.  So, after making public my ignorance, and after a few emails from readers asking me to explain the RPX curves, I decided to dig a little deeper, into publicly available information, to examine the two markets (see graphs below for prices from last week), and to try and create an illustration that might make the seasonality differences between the contracts clearer to students and first-time traders.

In trying to determine what forward curves might look like  I reviewed both the historical RPX and Case Shiller data since 2000.  (See for RPX data and–p-us—- for Case Shiller data).

I then calculated how the index release of a certain month (not the month is was released) compared to the average of index releases of the 6 months before and after.  The intent was to show, regardless of up and down markets, whether the month of an index’s release resulted in out-/ (under-) performance to the surrounding 12 months, over the last 11 years.

I then focused on the months that were relevant (the reference months) for the RPX contracts.  (Recall that the March RPX futures contract will settle on the RPX index value 63 days earlier, so late January, while the Sept contract will settle on an RPX index from late July).  (Since RPX data was available through Dec 9th, the window for +/- 6 months around July 31, 2011 was incomplete, and therefore not used.)

It should be clear from the graph to the right that the July 31st RPX index level was consistently higher than the aveage of RPX indices released over the +/- 6 month time frame (w/ all 10 data points being higher than average), and equally clear the Jan 31 index release being lower than the +/- 6 month average of RPX indices (also in 11 out of 11 years). (Note the strong under-performance of the Jan 31st index over the last three years.  More on that below.)

The calculations resulting in the July out-performance (and January under-performance) were then done for all 12 months for both the RPX and Case Shiller indices resulting in curves shown in the graph below.  This graph shows that  while each index has a seasonality to prices, that the RPX index has had a more pronounced swing over the last 10 years.   I’ll defer to Radar Logic and S&P as to how the index components and actual index calculations might impact seasonality but would note that since the RPX index covers a 28-day period, while the Case-Shiller index covers three months, any underlying seasonality in the index components would be amplified with a shorter measurement window.

I’ve also highlighted on the graph the reference periods for both the RPX contract (reference periods in blue dashes) and the Case Shiller contract (reference periods in red dotted lines).  The RPX March and September contracts each reference two of the more outlier portions of the RPX curve (January, at an average discount of 2.83%, and July, with an average premium of 2.33%, respectively) and as such one should expect to see the seasonality shape exhibited in the RPX futures curve at the top of this page.

In addition, the difference between the January discount and the July premium has average 5.16% – a number that is within range of the premium (~5.6%)  exhibited by the Sept ’12 contract mid-point over the average of the March ’12/March ’13 contract mid-point quotes.

In fact, in choosing to have contracts expire in March/Sept and reference the peak and valley in the seasonality curve, the CBOE/RPX contract gives traders a better way to articulate views on changes to future seasonal curves. One has only to look at the change in seasonal impact of the Jan RPX index over the last five years (the blue dots above) to see how volatile that month’s index has been -relative to prices +/- 6 months.  (These changes are not explained by 28-day versus 3-month averages.)  Debating changes to the size of future seasonal price moves, may be a fertile ground for RPX calendar trades, or outright views.

By contrast, the Case Shiller curve is flatter as the index covers both a wider time frame (90 days) and is a “moving average” index.  As such, the four reference periods for the Nov, Feb, May and August Case Shiller contracts don’t deviate as far from “average”.  Thus one should expect some, but not very much, seasonality in the Case Shiller CME curves.

I want to again qualify my comments by saying that these curves are for illustration only.    I’ve used simple math where the actual index calculations are more complicated.  I’ve talked about the average of the March ’12 and March ’13 RPX quotes when the price path between the two is unlikely to be a straight line.  Actual ,historical seasonal curves for either the RPX or Case Shiller indices might differ (and may differ going forward).    I encourage readers to visit Radar Logic and S&P (host for Case Shller index) sites to stay up to speed on developments that might materially impact future seasonal patterns on this indices.  As with my cautions on trading Case Shiller futures, no one should trade RPX futures without a through understanding of factors influencing the index calculations.

In sum, the numbers used here from this analysis of publicly available data may be useful in giving comfort to the notion that the sawed-toothed pattern in the RPX futures curve, makes sense, both in shape and in amplitude.

As always please feel free to contact me at to discuss this blog, or any aspect of home price derivatives.