|
I'll add a more philosophical take. Silver and many others are complaining that NET is not as good as power rankings systems like those used in professional sports, or Kenpom and Massey in college basketball. This is true. Those systems will be better at forecasting what teams are better than others at any given moment. But a key difference is power ranking systems are NOT used to determine postseason qualification. So while they may be more accurate, there is nothing to be gained by achieving a better power ranking compared to actual record.
Competitive sports must maintain the idea that winning is the ultimate goal. Power rankings can help us to predict who will win, but the goal should always be winning in itself. Power rankings and Kenpom place no emphasis on actual wins. Winning by 10 points instead of 12 is the same as losing by 1 instead of winning by 1 (assuming pace and opponent are equal). If we're going to maintain the integrity of sports, winning by 1 must be proportionally more important than an extra 1 pt margin. NET is a good compromise of placing emphasis on both winning (like the RPI) and on efficiency (like Kenpom).
When people are disparaging the November NET rankings by saying "well Radford isn't that good because they lost to Duquesne", they are using a preconceived notion of how good Duquesne is. And based off of their record in past seasons, they're probably right. But NET only looks at this year's games. Duquesne is 4-1 with their only loss at Notre Dame (who's only loss is to Radford). They will probably end up being around average, but right now they have a strong record. When evaluating NET, it's important to view it as if you've never followed basketball at all - that you didn't have any idea whether Duke should be better than Radford. That's how NET sees the season, which is very different from power rankings.
|