Thanks, Arif. This is interesting on multiple levels. If only this were a database that we could all play around with, manipulating different variables to understand different relationships.
For the graph, you use the value of the date sets (whether common dates, semi-key dates, key dates). It would be interesting to see a plot of the change in value of the "key denomination" (usually the 1/2 or the 1/4) of the each type (common dates, semi-key dates, key dates). Thus, instead of investing $100,000 in common date sets or semi-key date sets, what if one invested $100,000 in just the key denominations of each date set. To take one extreme example, let's say someone piled everything in 1997 G1/2s instead of full 1997 date sets. The return would be much greater for the 1997 G1/2 (oh, regarding disclosure, I WISH I had piled a lot of money into 97 G1/2, but I did not. I only have one). The flip side is, I guess it is not always obvious what the key denomination will be. If someone piled all their money in the 2002 G1/2, they would have missed out on the upward explosion of the price of the 2002 G1/10, so the date set may have been the better play for 2002.
The goal of playing around with all of these variables would be to try to inform our investment strategy going forward to get the most bang for our limited bucks. There are several levels to this question (1) do we invest in common dates, semi-key, or key; (2) do we invest in date sets or focus on what we think is the key denomination, or (3) do we focus on what we think are the key varieties. Each investor's decision is based on some of these variables, combined with what happens to be available on the market place when they are ready to buy.
Birdman