ESPN articles: Bracketology; Selection Committee to consider new metric

It's about time! There are at least 60 computer models out there and they come in different categories. My algortihm is a "predictor" type algorithm. It is intended to use the data to predict future performance in different situations. Other models are win/lose type models that have a heavy weight on winning. There are other models that are all designed for differing purposes. So each model will have it' strengths and weaknesses depending on what you are wanting to measure. I plan on submitting a recommendation to this committee which will be just one opinion on what type of criteria should be considered. WIns and losses is overall a poor metric since shcedule toughness is a hugely huge factor. Even how you calculate SOS is contentious. You can put emphasis on recent trend as a significant metric. You can emphasize consistency. You can weight performance against better teams (mental toughness) as a significant driver. You can weight home/away variation as a driver. You can weight offense vs defense as a driver. etc, etc, etc. In the end, you just have to decide what attributes or metrics are driving your criteria. This year, my DPPI has K-State ranked 8th in the Big 12 yet they are ranked 34th in the country. Should the 34th best team be in the tourney even if they are 5-13 in the conference but playing in a conference that is loaded top to bottom? There is no right or wrong answer to that question but it does need to be objectified and clarified. My experience is the tourney committee does a decent job for the most part in picking teams but their seeding is not always very good. As a quant, some would say I have a bias to use computers more robustly because they provide the right answer. COmputers only provide the answer you program them to provide and they are a slave to the ability of the data to effectively measure what it is you are trying to measure. A teams performance is not like measuring voltage or current in a circuit. A teams performance is driven by a huge number of variables that each have ranges of values in any given game. But as you start to collect a statistically meaningful size data set the ability of the data to derive meaningful conclusions grows exponentially. So I am all for using this type of approach as long as they are clear what it is they are trying to do. Reading the article it sounds to me like they fully get that concept. After saying all of that, I would love to see a year like this where 8 teams from the Big 12 get into the NCAA because they deserve it.