The Rational Picks fell below 50% in prediction accuracy over the first 5 weeks of the season. This is unprededented and terrible performance. Historical prediction results suggest The Machine should be in the range of 63 to 71% accuracy for most seaons, with an average of 67%. Five weeks of near 50% performance is something that should happen, on average, no more than once every 13 years. So, yes it’s possible that this is at least partly a statistical fluke.
However, this is clearly no ordinary year so far. A quick glance at the NFL standings tells some of the story. The entire AFC South is at 3-2. The Chiefs are atop the AFC West. The Cowboys are in last place in the NFC East and 2 games back. The Vikings are vying for last place in the NFC North and have fewer net points than the Lions. The Saints (defending Super Bowl champs) are in 3rd place in the NFC South. Clearly, the human experts are struggling as well, with an average accuracy of only 55% for the ESPN experts. This is at least 5 to 10% below their typical performance.
The root cause of the poor performance of the Rational Picks so far this year–whether a statistical perfect storm, a fundamental change in the NFL that represents a break from the past, an undetected error in the advanced algorithms of the Rational Sports Machine, or perhaps some combination–remains to be seen. Meanwhile, day and night the machine crunches away on the numbers looking for that new edge, that different nuance, that special factor that will enable it to catch up and surpass the human experts. The hole is deep, but the sason is still young. For now, all we can say is check back to see how it goes.
