Georgia Tech Game & Other Dribbles ...
- Thread starter DT_PITT
- Start date
Hi Atch:
These are efficiency numbers, and the concept starts off pretty basic.
Offensive efficiency is the number of points scored divided by the number of offensive possessions.
Defensive efficiency is the number of points surrendered divided by the number of defensive possessions.
From there, Ken Pomeroy adjusts this based on the opponent, so scoring more per possession against a good defensive team would be of greater value per se.
In layman's terms, KenPom is measuring scoring margin but is doing it on a possession basis rather than a game basis, since different games will have a different amount of possessions (basketball is a sport that, by nature, will alternate possessions so scoring margin on a possession basis is inherently more relevant).
The most "gah crazy math" thing it does is penalize teams for for numbers against cupcakes and not penalize teams for struggling against good opponents.
In terms of the analytics revolution, it's extremely tame. Things like the SportVU that Duke invested in take things to a level that the public isn't even privy to.
The most "gah crazy math" thing it does is penalize teams for for numbers against cupcakes and not penalize teams for struggling against good opponents.
In terms of the analytics revolution, it's extremely tame. Things like the SportVU that Duke invested in take things to a level that the public isn't even privy to.
But MajorMajors is also correct to note that there are more than one other reputable ratings of defensive efficiency in which Pitt is in the high 30s to high 50s. Team Rankings is one of them, where Pitt is 56. Pomeroy's use of points per possession makes great sense, of course, but once one gets into adjusting the raw numbers, there must be issues in weighing other factors. I don't subscribe to any of these systems to know their methods, but Team Rankings results cannot be dismissed easily. Does anyone know how some of these other systems come up with such a better picture of our defensive efficiency, and how their legitimacy compares to Pomeroy?
For what it is worth, our defensive rebounding% is 103rd in the country this season. This is much better than last year, but worse than Pitt's previous teams. Also, as you indicated, the competition has helped.
Pitt got their asses kicked by Kent State, Purdue, and GT but has been pretty good otherwise. I think Pitt should be alright against all but the best rebounding teams on that end.
Defensive rebounding has been my main concern all year and I'm interested to track it as well.
Most likely it's the difference between the raw numbers and the adjusted numbers. If you play a team that normally scores 1.15 points per possession and you hold them to 1.10 then the raw numbers wouldn't look so good, because 1.10 ppp is below average (defensively). But the adjusted numbers would look pretty good, because you allowed the team to score less than what they normally do.
The fact is that Pitt has played a pretty bad schedule this season to date. So that means that Pitt's raw defensive numbers are better than Pitt's adjusted defensive numbers, because the adjusted numbers account for the fact that the opponents haven't been very good and the raw numbers don't.
Personally, I believe Kenpom biases his adjustments too much based on past performance. Raw numbers Pitt is around 31st in DE at 92.8 but adjusted to 101.7 while Louisville (that played just as poor of an OOC) is 4th in raw DE at 85.9 but adjusted to 91.5 in KenPom. So Pitt is adjusted 8.9 and the Ville 5.6. While Pitt has played against slightly better offenses compared to the Ville . Somehow the Ville gets a better pyth factor - only makes sense if Pitt is paying for the last 3 seasons IMO. The defense is not as bad as some folks state here - inconsistency is it's biggest bugaboo - at times it has been darn right dominating. Recent past it has been consistently poor.
As a side note, I like to use the 2 pt % as a gauge of defense (yes, rbding, steals etc are include but 2 pt% is telling more so than 3% which Kenpom has really shown to be beyond much control of the defense) - so far Pitt is at 42.9% (good for 40th) last year the team finished at 49.1%, then 46.8%, 44.9% (Adams year), 49.3 (the atrocity of Birch), 43.2% (#1 seed and most likely one of the 4 best teams that year independent of the Butler brain fart game). Too early to count on 42.9% to hold but shows promise for this to be a decent defensive team. Why not choose to be an optimist?
I agree, tuckshop. I know that Pomeroy is the gold standard for a lot of sports writers, but it is strange that his adjustment might lead to such a discrepancy. It is even more strange given his recent dismissal of the relative significance of 3pt% defense compared to 2pt% defense. It is a methodological position that runs counter to PointsPerPossession, even with adjustments. Pomeroy's StrengthofSchedule for Pitt is also lower than some other analysts have, and maybe that also factors into his DE number.
Looking at Greenfield's numbers, I checked Greenfield's DE rating for Pitt from last year, which may have limited comparative relevance, but which would put our 2014 performance well above 200. That is at least in line with other ratings, including our own eyes. To my mind, it also lends some credibility to this year's ratings.
There is no doubt that our defense needs to continue to improve. Still, I am not allowing Pomeroy's low DE rating to pop any of my excitement balloons. I am surprised at how many people are still hedging on this team, not just in the media but some posters who have typically focused on positive aspects of our play. The relevant proverb is, "Observers Are Worried; Believers Are Enjoying."
There is a lot to like about this team. Against Notre Dame, as a few others have noted, I was particularly impressed by their composure under pressure.
Looking at Greenfield's numbers, I checked Greenfield's DE rating for Pitt from last year, which may have limited comparative relevance, but which would put our 2014 performance well above 200. That is at least in line with other ratings, including our own eyes. To my mind, it also lends some credibility to this year's ratings.
There is no doubt that our defense needs to continue to improve. Still, I am not allowing Pomeroy's low DE rating to pop any of my excitement balloons. I am surprised at how many people are still hedging on this team, not just in the media but some posters who have typically focused on positive aspects of our play. The relevant proverb is, "Observers Are Worried; Believers Are Enjoying."
There is a lot to like about this team. Against Notre Dame, as a few others have noted, I was particularly impressed by their composure under pressure.
Just two quick comments about the Pomeroy ratings:
1. Our current DE score is exactly equal to the average DE score across all D1 teams, so *by definition* our defense is average right now. It doesn't make sense to cite Pomeroy as evidence of our defense being "bad".
2. If one sorts the teams by their DE rankings, one would notice a bunch of teams who will not sniff the tournament. As such, it's better to compare our score against scores of teams that are likely to be in tournament and compare against that group.
1. Our current DE score is exactly equal to the average DE score across all D1 teams, so *by definition* our defense is average right now. It doesn't make sense to cite Pomeroy as evidence of our defense being "bad".
2. If one sorts the teams by their DE rankings, one would notice a bunch of teams who will not sniff the tournament. As such, it's better to compare our score against scores of teams that are likely to be in tournament and compare against that group.
But, if you rank teams that ARE likely to be in the tournament, aren't we right back in the "bad"category, not the "average" from your first point?
That second comment was to address the notion of "183 is a big number and it's therefore inherently bad". If anything we're middle of the pack in that reduced set as well.
To me, this whole "offense first defense later" performance this season is interesting as an experiment: we know that "defense first" carried us to 2 sweet 16s and a "balanced approach" carried us inches from a final 4. I'm curious as to where our offense will carry us this year, although I feel we'll end like Brazil ended at the latest World Cup when they played Germany. But, as a saying in my native language goes, at least we'll have fallen off a beautiful horse.
Speaking of soccer, which is the *one* sport I grew up with, we'd have endless discussions about whether a penalty kick is missed by the striker or defended by the goalkeeper. This whole thread is a continuation of that argument applied to a different sport.
Announcers said that the players are having a FT shooting competition and egg each other on - would be nice if this applied also to rebounding and shooting % of the player they guarded. Competitions like those help focus. Looks like team's future spins on Young, Artis, and Jones getting defensive religion (who by some measures are the worst defenders that get significant minutes)
Well, after yesterday's game, Team Rankings dropped our DE position by nearly thirty points to 87. Ha ha! I guess they didn't think that Jackson figured to be someone who could hit all those shots, or that Colson and August were supposed to be able to make all the moves they did. If our number for that one game were a season number, we would be literally off the charts, far below the worst of the 351 D1 teams on the list. Notre Dame's one-game number was worse than ours. I guess Team Rankings also doesn't think Artis and Young should be that hard to defend, either. After sorting on the "last game" column, I saw that we and Notre Dame, out of the 351 ranked teams, were 347 and 350 respectively.
Maybe the inference is that Team Rankings aren't that soft with their DE ratings after all. I'll still take their 87 DE rating over Pomeroy's 185 and use Team Ranking's response to the Notre Dame game as evidence that they don't take a day off from work.
But talk about killing an ant with a sledge hammer! I didn't think the defenses were as bad as that. I may have to reside, at least temporarily, with the contingent of posters who don't look at the numbers and rather trust their own eyes. Anyone else?
Maybe the inference is that Team Rankings aren't that soft with their DE ratings after all. I'll still take their 87 DE rating over Pomeroy's 185 and use Team Ranking's response to the Notre Dame game as evidence that they don't take a day off from work.
But talk about killing an ant with a sledge hammer! I didn't think the defenses were as bad as that. I may have to reside, at least temporarily, with the contingent of posters who don't look at the numbers and rather trust their own eyes. Anyone else?
Not to belabor the point, but our defense is often not much better than last year's, at least when the increased offense from the new rules is taken into effect. It will cost us sone games.
I hope we can become the exception to the rule and just outscore people, but the reality is the stats only tell us what happened. Even using them to describe why things happened may be questionable at times.
That's the problem. Our numbers are most certainly not in the middle of the pack of the reduced set of "teams that are likely to be in the tournament." Our numbers compared to that subset are clearly much worse than the relative middle of the pack ranking when considering all teams. That's the real worry, that our defense is bad compared to teams like Louisville and Purdue and Oklahoma and the like. The fact that we are a better defensive team than Maryland Eastern Shore and Eastern Washington is both true and completely irrelevant. Of the teams ranked in the Pomeroy top 80, teams that have a realistic chance of getting to the NCAA tournament and maybe winning a game or two when they get there, we have a worse defensive rating than every single one of them except Notre Dame. If you want to stretch it to the top 100 we would be 96th. We are much, much closer to the bottom of the pack among realistic NCAA teams than we are even to the middle, let alone the top.
To shut down dribble penetration you need quick guards. Pitt doesn't have any. To make up for dribble penetration you need a shot blocker. Pitt doesn't have any.
Harve, saying that our defense this year is not much better than last year seems to be a marginal opinion even among the critical eyes on this Board. We all recognize the issues and the breakdowns and the mistakes, but I think most people would say that our defense is improved from last year. Tuckshop's post about 2pt% is real and is also on point regarding how an interpretive number like DE is calculated.
KGSB's allusion to the endless discussions about soccer penalty kicks is extremely subtle. Although the odds of success are quite different from basketball, I agree that the new rules make it difficult to figure out what defense parameters are meaningful. Players, coaches, and stripes are still working on it, too. Yesterday's game may just have been some sort of anomaly. The "last game" offensive efficiency numbers for us and Notre Dame were 2 and 6, respectively. Those numbers may have fit an expected narrative of "two great offenses versus two poor defenses," but 2 and 6 OE taken together with 347 and 350 DE seems totally logical yet somehow totally meaningless.
The issue is whether our defense is actually as bad as Pomeroy's numbers indicate. There are a lot of systems, but using Team Rankings for a comparative perspective is not unreasonable. Under that system, as bad as they evaluated us for yesterday's performance, we are significantly better than Pomeroy's ranking. Using KGSB's and Joe's exchange above regarding a reduced set of realistic NCAAT prospects, more than half the 86 teams ahead of us in Team Rankings' DE list are out, placing us well within an average range of numbers.
KGSB's allusion to the endless discussions about soccer penalty kicks is extremely subtle. Although the odds of success are quite different from basketball, I agree that the new rules make it difficult to figure out what defense parameters are meaningful. Players, coaches, and stripes are still working on it, too. Yesterday's game may just have been some sort of anomaly. The "last game" offensive efficiency numbers for us and Notre Dame were 2 and 6, respectively. Those numbers may have fit an expected narrative of "two great offenses versus two poor defenses," but 2 and 6 OE taken together with 347 and 350 DE seems totally logical yet somehow totally meaningless.
The issue is whether our defense is actually as bad as Pomeroy's numbers indicate. There are a lot of systems, but using Team Rankings for a comparative perspective is not unreasonable. Under that system, as bad as they evaluated us for yesterday's performance, we are significantly better than Pomeroy's ranking. Using KGSB's and Joe's exchange above regarding a reduced set of realistic NCAAT prospects, more than half the 86 teams ahead of us in Team Rankings' DE list are out, placing us well within an average range of numbers.
Playing good team defense as well as team rebounding is not a switch you turn on/off. I call BS on this theory.
In the past in the tourney, it was usually one guy who beat Pitt. The philosophy of all man-man all the time will never work when you got one guy going off on you. You need a junk defense and the resistance of going to such extremes is why Howland/Dixon struggle. The other thing.. the guy who is going off on you - is probably not playing too hard on the other end. Therefore you attack him mercilessly. Not having a guard who can do this (funny how it is usually guards who would go off on Pitt) is the second achilles heal of the Howland/Dixon regime.
That said... this year... they better find out a way to get some stops at some point. Relying on 85+ night after night ain't gonna work.
Lunz, the raw numbers of defensive efficiency are certainly not interpretive. It's a simple mathmatical calculation of points scored by the opponent divided by possessions.
It only becomes interpretive when "fudge factors" are built into the various models. While a defense has relatively little control over 3 pt percentage, I wouldn't omit 3 pt scoring from DE because in the real world 3 point shots can and do count on the score board. In fact, today many if not most games are decided by 3 point shooting. Ignoring them makes any defensive ranking pretty irrelevant. Likewise, defenses have even less control over FT% made. But as long as the scoreboard counts those points, they've got to go into Defensive Efficency numbers, too.
Besides, defenses do have some control over both methods of scoring in that they can minimize the number of 3 pt attempts by denying 3 pt shots and free throws by not fouling.
I tend to favor Pomeroy's adjusted method because 1) an awful lot of college basketball experts, as well as the few knowledgeable guys I know like it. It's pretty much the language college basketball people talk, and it's got plenty of data for comparison . And, 2) it makes a lot of sense to me to adjust those numbers for quality of opponent.
We're fun to watch and I'm happy we're winning but as I posted in a different thread, we had close games against several teams who will likely finish in the bottom half of the ACC. We won them, so it's likely we're better than our 10th place ACC preseason ranking.
But, it we can't defend teams which are around average, we aren't going to be as successful as we hope.
Which all boils down to we are--at least on Jan 10th--probably a team deserving to be ranked around #20 (+\- 2 or 3 spots) and about 6th in the ACC as Sagarin's computer has had us recently.
Some happy stuff. Looking at Dean Oliver's four factors to winning here is this year's numbers versus last year.
EFG% 2015 55.8% 19th 2014 49.1% 172nd
TO% 2015 15.1% 26th 2014 15.3% 12th
OR% 2015 38.3 % 11th 2014 31.8% 90th
FTR 2015 43.8 40th 2014 34.8% 237th
Except for maintaining an excellent TO%, a pretty significant improvement across the board.
With having the leading FT% being the cherry on top. As well as playing at a 68 possession pace instead of the old plodding 60.
Early in the season, but even if these numbers slip a bit it is still a massive improvement.
EFG% 2015 55.8% 19th 2014 49.1% 172nd
TO% 2015 15.1% 26th 2014 15.3% 12th
OR% 2015 38.3 % 11th 2014 31.8% 90th
FTR 2015 43.8 40th 2014 34.8% 237th
Except for maintaining an excellent TO%, a pretty significant improvement across the board.
With having the leading FT% being the cherry on top. As well as playing at a 68 possession pace instead of the old plodding 60.
Early in the season, but even if these numbers slip a bit it is still a massive improvement.
Last edited:
Harve, when I replied directly to you, I wondered if I was going to get an introductory lecture, but fortunately I had also already read the assigned material on Pomeroy's blog, as have most of the participants in this thread, I assume.
Raw numbers are always raw, but a defensive efficiency rating is not a raw number. What you dismiss as "fudge factors" are explicitly interpretive. The discussion is about significant discrepancies among defensive efficiency ratings, and such discrepancies themselves are implicit evidence of interpretation.
You may like Pomeroy, and I like that site too, but there are sites that present a less discouraging assessment of our defense. The issue at hand is whether we are actually as bad defensively as Pomeroy's rating suggests. It is legitimate to question Pomeroy's methods in comparison to Team Rankings' ones, which lead to a different picture of Pitt's performance. I was rather hoping that someone might know more about how Team Rankings (or another comparable system) derives its ratings, or as you might say, stirs its fudge.
Raw numbers are always raw, but a defensive efficiency rating is not a raw number. What you dismiss as "fudge factors" are explicitly interpretive. The discussion is about significant discrepancies among defensive efficiency ratings, and such discrepancies themselves are implicit evidence of interpretation.
You may like Pomeroy, and I like that site too, but there are sites that present a less discouraging assessment of our defense. The issue at hand is whether we are actually as bad defensively as Pomeroy's rating suggests. It is legitimate to question Pomeroy's methods in comparison to Team Rankings' ones, which lead to a different picture of Pitt's performance. I was rather hoping that someone might know more about how Team Rankings (or another comparable system) derives its ratings, or as you might say, stirs its fudge.
Defensive efficiency should be a pretty straight-forward number, a simple division of points allowed by possessions. It would be simpler to analyze if box scores listed possessions instead of that number requiring calculation. And, there is even some disagreement on the formula used to calculate posessions.
Obviously, the more factors each statistician uses in calculations affects the rankings and they do become subjective or interpretive. I don't know exactly what goes into Team Rankings's model but even using their figures, (doing the math from the ND boxscore indicates they do SOME adjusting because their "last game" DE number is different from the straight basic calculation) the DE from our last 3 games would rank us about #250 overall. That is slightly worse than they ranked us last season (245). Even though we are 3-0 in those games, it is very unlikey such defense will continue to thrive.
By EITHER system our defense has rapidly worsened since the ACC schedule has begun. Is that because a couple of opposing guards had great shooting nights or is the root cause deeper? My "eye test" would say, the problem is basically the same as last season, this group of players are simply not very good defensively.
Again, Pomeroy's figures are pretty much the standard that are used by the basketball community. It's easier if everyone is using the same figures for comparison.
Last edited:
