Style of lineup edge production in matchups
Posted: Wed Feb 03, 2016 7:46 am
Here is a big data query idea:
Sort every lineup used in league into three piles. Those with say 70% or more of most commonly used starting lineup's cumulative average usage, those with 50-70% and those with less than 50%. Check offensive and / or net performance for 1s (fts), 2s and 3s against defensive lineups with 4-5 "starters", 2-3 and 0-1. 9 slices of data. Which offenses on average do best / worst against which defenses? How are they good and bad? It is 2016 and I have never seen anything like this in public. Here, there, Sloan, anywhere. (I have been told that idea generation is easy and dime a dozen. I disagree. I can't code a few lines and give you the answer quickly. But with the idea, someone can do that. Without the idea, well...) Anyone with a lineup database and coding chops want to bring this or something close to it to realization? It often takes two...
If you want to win at lineup matchups more often than 1/3rd the time (coaching average), through the different parts of the lineup rotation, it might be nice to know this data. How many teams / coaches do? From this modest positive lineup performance rate, it doesn't look like coaches are that successful at analyzing unique lineup matchups, so would seem somewhat wise to get to know these trends. But coaches will wing it and most will tolerate little interference from analytic peddlers I imagine. Same as it mostly has been.
If you want to move forward in basketball strategy and results, I'd want to know how much harder it is to produce points edge against mostly starters vs. some vs. little. For offensive lineups with most of your starting offensive usage vs. less vs. even less. Perhaps you should attack high degree of defensive starters differently than mostly bench. Perhaps strong bench units that face mostly bench units can't expand that much beyond that matchups against bench defenses. Perhaps some can. Which, how, how far? What does the big data query say? How can teams essentially freestyle lineup matchups without this knowledge (if they do)? Why hasn't basketball analysis, or at least the outsider part, done that much with a lineup version of "hitting" vs. pitching / defense?
If you think starters vs. bench is too opaque and it might be, then look at offensive and defensive biases players (by RPM). But if you do think this, what does that say about starter vs. bench designations by coaches? If you not a strong believer in lineup unit productivity vs. individual productivity, then take the analysis down to the one on one counterpart matchup level too.
Sort every lineup used in league into three piles. Those with say 70% or more of most commonly used starting lineup's cumulative average usage, those with 50-70% and those with less than 50%. Check offensive and / or net performance for 1s (fts), 2s and 3s against defensive lineups with 4-5 "starters", 2-3 and 0-1. 9 slices of data. Which offenses on average do best / worst against which defenses? How are they good and bad? It is 2016 and I have never seen anything like this in public. Here, there, Sloan, anywhere. (I have been told that idea generation is easy and dime a dozen. I disagree. I can't code a few lines and give you the answer quickly. But with the idea, someone can do that. Without the idea, well...) Anyone with a lineup database and coding chops want to bring this or something close to it to realization? It often takes two...
If you want to win at lineup matchups more often than 1/3rd the time (coaching average), through the different parts of the lineup rotation, it might be nice to know this data. How many teams / coaches do? From this modest positive lineup performance rate, it doesn't look like coaches are that successful at analyzing unique lineup matchups, so would seem somewhat wise to get to know these trends. But coaches will wing it and most will tolerate little interference from analytic peddlers I imagine. Same as it mostly has been.
If you want to move forward in basketball strategy and results, I'd want to know how much harder it is to produce points edge against mostly starters vs. some vs. little. For offensive lineups with most of your starting offensive usage vs. less vs. even less. Perhaps you should attack high degree of defensive starters differently than mostly bench. Perhaps strong bench units that face mostly bench units can't expand that much beyond that matchups against bench defenses. Perhaps some can. Which, how, how far? What does the big data query say? How can teams essentially freestyle lineup matchups without this knowledge (if they do)? Why hasn't basketball analysis, or at least the outsider part, done that much with a lineup version of "hitting" vs. pitching / defense?
If you think starters vs. bench is too opaque and it might be, then look at offensive and defensive biases players (by RPM). But if you do think this, what does that say about starter vs. bench designations by coaches? If you not a strong believer in lineup unit productivity vs. individual productivity, then take the analysis down to the one on one counterpart matchup level too.