A disturbing trend I see throughout the discussion of NET rankings, data science, and the use of advanced analytics is something that I see paralleled in my own field of science: People using advanced statistical algorithms to essentially absolve themselves of any responsibility of correct interpretation of the data. These algorithms are supposed to be tools, but people plug in their data and have it spit out the answer and shrug their shoulders and say 'well this is what the data says'. No critical thinking necessary. And if the algorithm is wrong, well that's not our fault, we just ran the data.
But sometimes data needs CONTEXT. Team's can improve throughout the course of a season or they can get worse. At the end of the year, a team's resume does not entirely reflect their current quality. Pitt and UVA are two examples of team's on opposite ends of that equation, and winning games at the end of the year should matter more!
Additionally, different metrics have different meaningful value for interpretation. Strength of Schedule has always been a poor evaluation metric, bc frankly what is the real difference between playing a schedule of rank 250 vs 350? Maybe the potential for 1 additional loss? And team's have no control over what their opponents will do throughout the rest of their games. Pitt scheduled 4 P5 opponents OOC (Florida, Missouri, Oregon State, and WVU), it's not their fault that Missouri and WVU would completely tank their seasons after finishing 19 and 57 in KenPom ranking last season. Lunardi's comment about NCSOS number is asinine bc it 1) focuses on a bad qualitative metric, 2) doesn't take into account the context of that number, 3) overvalues the early part of the season, and 4) ignores literally everything else that says Pitt is a good team.
The whole entire point of there being a selection committee is to make the decision semi-independent of the numbers. You want to use the data as tools to avoid human bias, but you also don't want to be a mindless slave to the data. The committee has the responsibility to put the strongest field of 68 teams together on Selection Sunday and I hope they don't cop out by ignoring what Pitt has accomplished.
But sometimes data needs CONTEXT. Team's can improve throughout the course of a season or they can get worse. At the end of the year, a team's resume does not entirely reflect their current quality. Pitt and UVA are two examples of team's on opposite ends of that equation, and winning games at the end of the year should matter more!
Additionally, different metrics have different meaningful value for interpretation. Strength of Schedule has always been a poor evaluation metric, bc frankly what is the real difference between playing a schedule of rank 250 vs 350? Maybe the potential for 1 additional loss? And team's have no control over what their opponents will do throughout the rest of their games. Pitt scheduled 4 P5 opponents OOC (Florida, Missouri, Oregon State, and WVU), it's not their fault that Missouri and WVU would completely tank their seasons after finishing 19 and 57 in KenPom ranking last season. Lunardi's comment about NCSOS number is asinine bc it 1) focuses on a bad qualitative metric, 2) doesn't take into account the context of that number, 3) overvalues the early part of the season, and 4) ignores literally everything else that says Pitt is a good team.
The whole entire point of there being a selection committee is to make the decision semi-independent of the numbers. You want to use the data as tools to avoid human bias, but you also don't want to be a mindless slave to the data. The committee has the responsibility to put the strongest field of 68 teams together on Selection Sunday and I hope they don't cop out by ignoring what Pitt has accomplished.
Last edited:
