SPI Tinkering

I've made a couple of discoveries in the last few days by playing with the SPI and seeing what I could come up with. Here are the new tools that I've come up with.

Adjusted SPI. This is something I've wanted to do for a while, but I didn't want to expend the energy necessary to make it. I added a function that calculates a team's SPI over its last ten games, then factors that into its overall SPI rating. This gives teams on a winning streak a boost to reflect recent performance, while holding back teams that aren't doing so well. I wanted to find a way to take into account how a team is doing currently, since a team that has been bad for most of the year but is playing good recently should be rated higher than the current SPI format would place them. Adding this into the SPI ratings shakes things up considerably. The Chicago Cubs, already dominating the league, earn an extra full point, widening the already considerable distance between them and the Red Sox. The Chicago White Sox and Oakland A's, both slumping, see their ratings drop considerably. While the unadjusted SPI is still the best measure of a team's overall performance and potential, the adjusted SPI more accurately reflects how a team is likely to do over the course of their next few games, which brings us to the next new tool.

Series percentage calculator. By plugging two teams' adjusted SPI ratings into this calculator, you can see each team's forecasted success in an upcoming series. For example, let's consider the series between the Padres and the Mets. By plugging each team's SPI ratings into the calculator, we see that the Mets have a 68.9% chance of winning each game, and a 77% chance of winning the series (either winning two or three games). So far, the Padres and the Mets have each won one game, with the final game being played today. (The Mets lead 2-1 in the top of the 6th.) The calculator predicted a 44.3% chance of the Mets winning the series 2-1 and a 20% chance of the Padres winning 2-1. Based on that, we can see that the Mets are more than twice as likely to win this game than the Padres.

Neither of these tools are foolproof, not being based in actual statistics, but I think they correspond pretty closely with what we can see happening in the league. I'll continue to play with them and post updates as I see fit.