louisville/uconn
hi, i have been asked by my friend justin to give a uconn/louisville preview. i have also been asked to trash hasheem thabeet, so i will do each in turn.
this is undoubtedly the big east game of the year. marquette will be a good game as well (for both louisville and uconn), but there aren't too many other big east teams that are as deep or complete as uconn and louisville. pitt is one, but their shooting ability has really been exposed in the last few games.
look, let's start with a caveat: it's not clear at all that louisville can match up with price/adrien/dyson, although terrence williams should be able to do something. on the flip side, i think perhaps louisville is a slightly deeper team than uconn, and they might have some success on the interior, since it's not clear that thabeet will be able to help on clark/williams without leaving samuels open. samuels is too good of a finisher on the inside to leave open; unlike other teams that uconn has played, louisville has multiple scoring threats on the interior, so thabeet won't be able to help as much as he might like.
the key to this game is terrence williams. he has to be able to shut down jerome dyson. if dyson goes off for a lot of points, uconn will win. on the other hand, if williams can control dyson's effectiveness, then louisville has a very good chance to win, because earl clark presents somewhat of a matchup problem for uconn. thabeet can't guard clark on the perimeter. that's not his game, and so adrien will have to do it. that presents its own problems, because pitino will spend 15 fouls harassing thabeet (who can be baited) on the interior with jennings, goode, and samuels.
uconn's greatest strength in this game will be the nightmare matchup that price presents. price exposes sosa's lack of defense, making sosa more of a liability on the court. but i think sosa will have to play a lot because taking him off the floor could result in 4 on 5 on offense, which can't happen for any length of time. both knowles and mcgee will be liabilities on the offensive end, and both calhoun and pitino know that. i wouldn't expect to see them used together as much as they normally are. uconn is too good for that.
expect to see a complete slugfest between jerry smith/mcgee/knowles and austrie/kemba walker. jerry smith is probably the best of those guys this year, but walker is legit. but walker/price will cause serious quickness problems for the cardinal guards. many teams have used yucky point guards to get points against louisville; aj price is not a yucky point guard (neither is kemba walker), and this will be a major problem. this problem is magnified by the fact that louisville won't be able to cheat with terrence williams. dyson and austrie are too good to leave open.
i think that louisville's depth makes a difference in this one, but it's really too close to call. i'll take louisville, because of factor #2:
hasheem thabeet's complete inability to show up in big east play.
it's an overstatement, but not by much. it's also a readily testable statement.
here's a regression of pts per game on whether or not thabeet is playing in a big east game. if thabeet scores more points in big east play, this number will be positive...
ppg = -2.768 BIGEAST + 11.05
this is statistically significant, as the standard error is 1.184. perhaps this is not convincing. here is another regression, relating personal fouls to the big east. if playing in the big east doesn't matter, this number should be either statistically insignificant or close to zero, and if thabeet fouls fewer times, this number should be negative. but it's not...
pf = .5914 BIGEAST + 2.25
the standard error here: .25439. we're within a 95% confidence interval.
here's another, less convincing regression:
total rebounds = -1.380 BIGEAST + 8.641
the standard error here: .810, so we're only within a 90% confidence interval.
if you arbitrarily define some metric for "big games," thabeet does much worse. more fouls, fewer points, fewer rebounds, etc. i'm not going to get into that here.
so, thabeet is likely to commit about .5 more fouls (not a whole number, but you only get five and is likely to score three fewer points and grab one fewer rebound. does he play fewer minutes? the answer to this is no: there is hardly any correlation between minutes played and big east games: on average, he plays about the same amount of time. from some arbitrary, low sample size regressions (although with n > 30 we can use the LLN and CLT...), it seems as if thabeet's simply less productive. at the very least, there's no case for him playing better, or rising to the occasion, or anything like this.
what makes this a nice result is the remarkable amount of trash that thabeet talks.
when uconn wins because of good guard play, thabeet can yap about harangody, even though luke went for 27. since uconn hasn't played unc, thabeet can yap about hansborough.
but a parting remark: thabeet doesn't say a word about earl clark.
this is undoubtedly the big east game of the year. marquette will be a good game as well (for both louisville and uconn), but there aren't too many other big east teams that are as deep or complete as uconn and louisville. pitt is one, but their shooting ability has really been exposed in the last few games.
look, let's start with a caveat: it's not clear at all that louisville can match up with price/adrien/dyson, although terrence williams should be able to do something. on the flip side, i think perhaps louisville is a slightly deeper team than uconn, and they might have some success on the interior, since it's not clear that thabeet will be able to help on clark/williams without leaving samuels open. samuels is too good of a finisher on the inside to leave open; unlike other teams that uconn has played, louisville has multiple scoring threats on the interior, so thabeet won't be able to help as much as he might like.
the key to this game is terrence williams. he has to be able to shut down jerome dyson. if dyson goes off for a lot of points, uconn will win. on the other hand, if williams can control dyson's effectiveness, then louisville has a very good chance to win, because earl clark presents somewhat of a matchup problem for uconn. thabeet can't guard clark on the perimeter. that's not his game, and so adrien will have to do it. that presents its own problems, because pitino will spend 15 fouls harassing thabeet (who can be baited) on the interior with jennings, goode, and samuels.
uconn's greatest strength in this game will be the nightmare matchup that price presents. price exposes sosa's lack of defense, making sosa more of a liability on the court. but i think sosa will have to play a lot because taking him off the floor could result in 4 on 5 on offense, which can't happen for any length of time. both knowles and mcgee will be liabilities on the offensive end, and both calhoun and pitino know that. i wouldn't expect to see them used together as much as they normally are. uconn is too good for that.
expect to see a complete slugfest between jerry smith/mcgee/knowles and austrie/kemba walker. jerry smith is probably the best of those guys this year, but walker is legit. but walker/price will cause serious quickness problems for the cardinal guards. many teams have used yucky point guards to get points against louisville; aj price is not a yucky point guard (neither is kemba walker), and this will be a major problem. this problem is magnified by the fact that louisville won't be able to cheat with terrence williams. dyson and austrie are too good to leave open.
i think that louisville's depth makes a difference in this one, but it's really too close to call. i'll take louisville, because of factor #2:
hasheem thabeet's complete inability to show up in big east play.
it's an overstatement, but not by much. it's also a readily testable statement.
here's a regression of pts per game on whether or not thabeet is playing in a big east game. if thabeet scores more points in big east play, this number will be positive...
ppg = -2.768 BIGEAST + 11.05
this is statistically significant, as the standard error is 1.184. perhaps this is not convincing. here is another regression, relating personal fouls to the big east. if playing in the big east doesn't matter, this number should be either statistically insignificant or close to zero, and if thabeet fouls fewer times, this number should be negative. but it's not...
pf = .5914 BIGEAST + 2.25
the standard error here: .25439. we're within a 95% confidence interval.
here's another, less convincing regression:
total rebounds = -1.380 BIGEAST + 8.641
the standard error here: .810, so we're only within a 90% confidence interval.
if you arbitrarily define some metric for "big games," thabeet does much worse. more fouls, fewer points, fewer rebounds, etc. i'm not going to get into that here.
so, thabeet is likely to commit about .5 more fouls (not a whole number, but you only get five and is likely to score three fewer points and grab one fewer rebound. does he play fewer minutes? the answer to this is no: there is hardly any correlation between minutes played and big east games: on average, he plays about the same amount of time. from some arbitrary, low sample size regressions (although with n > 30 we can use the LLN and CLT...), it seems as if thabeet's simply less productive. at the very least, there's no case for him playing better, or rising to the occasion, or anything like this.
what makes this a nice result is the remarkable amount of trash that thabeet talks.
when uconn wins because of good guard play, thabeet can yap about harangody, even though luke went for 27. since uconn hasn't played unc, thabeet can yap about hansborough.
but a parting remark: thabeet doesn't say a word about earl clark.
posted by mike at 11:24 AM
6 comments
