tXtFL Bowl 9.0: Analysis

It's been awhile since Super Bowl XLIX, but we've finally put together our graphical analysis of tXtFL Bowl 9.0. Unlike the last bowl, this decision was on the finest of margins. As you can see by the relationship of the green circle to the dotted line, the final score was just ever so slightly toward the Patriots' side.

From the scatter of blue dots, it's clear that as in previous simulations, any individual game in our 10,000 simulations can be all over the map. The red line shows the regression (forced through the origin), which just barely lists away from the dotted null line. The scores were so close in fact that we re-ran the simulations several times and double-checked individual statistics to ensure that the actual player stats were being loaded instead of default scores.

To our surprise, the final score fell almost exactly on the regression line. A total of 52 of the 10,000 simulations, or 0.5%, had this exact same score. This level of prediction accuracy was unlike tXtFL Bowl 8.0 from last year, where the odd score (43-8) left us with not a single exact score match, even if the overall winner was still predicted accurately.

In the offseason, we'll continue working on tXtFL 4.0 development with additional tweaks to the game model as well as new features such as health and injuries to further refine our simulations. Feel free to grab a beta build, run your own simulations, and let us know how to improve tXtFL!


Popular Posts