I remember the starting lineup with Bev was around mid-90's in defensive rating up to the end of November. A lot of times you have to break them up into sections just to have a better perspective. It would probably be eye popping if we can somehow extract defensive rating only for December. If I have to estimate, it's probably in mid 100's. I also remember one time that Twin Tower was just awful in both ratings but when you break them up into two halves, it really told a different story.
Here are the teams Lin / Beverley started against in with the standard starting lineup. Purpose is to show show the quality of competition in terms of the lineup data. Ordered from strongest to weakest: Spurs Spurs Pacers Thunder (No Westbrook) Warriors Dallas Denver Chicago Memphis Pelicans Knicks Knicks (No Chandler) Nets Kings Kings Celtics If we erase duplicates, ie Spurs/Kings where each Lin and Beverley started one game against: Pacers Thunder (No Westbrook) Warriors Dallas Denver Chicago Memphis Pelicans Knicks Knicks (No Chandler) Nets Celtics
Defensive efficiency through first 35 games, game by game. The pink line shows defensive efficiency over last 5 games. Asik's last game was against Utah. Beverley's last game was against Detroit. I don't know if either one is the main factor. Its the cumulative effect of injuries to both of them, and perhaps also the schedule where we had a bunch of games without much practice time in between.
Hard to get many takeaways from that, as the trends seem seasonal. There is so many things you could add to that. Opposing ortg, days rest, etc.
box out & p&r D, we were better last year with asik so you blame mchale ????? hes not on the court with his celtics uniform im sure he tells the guys just like he did last year when we didnt have this problem. & we have a 10year vet right in the middle burt we blame mchale
Don't we have one of the hardest schedules this year? I know that really shouldn't be an excuse, but having one of the most "back to back" games really drains players. Maybe someone could cross reference this with the stats on our Defensive ratings?
I think when we compare line ups and defensive ratings, there is an assumption that the whole team is playing relative consistent defense from game to game, thus we could lump sum all the player time together to compare them. Usually a mature team has a style of defense, so that assumption is taken. However, Rocket this season's defense are so different from game to game. From the TT experiment, to player's effort, they could be black and white and from one game to the other. James could vary from paying no attention to defense to playing great defense when he is focused, which would have a great impact on the player share the court with him. It is best shown in this graph. Just think about the team defense on the first Clipper's game vs. the one in first GSW game, that's a big 50 point difference. And how much impact of that in your defensive rating, and do you think that was due to one individual player's excellence defense? Now think that Beverley did not play in the first CLIPPR's game and Lin did not play in the first GSW game. Thus at this stage, I'd say these two games will give a much bigger error bar when you compare these two players head to head. Maybe when we have a bigger sample size, it could be evens out, but who knows. In terms of Casspi vs. TJ. Obviously Casspi plays more time during TT experiment, and TJ didn't. I do feel the defense was much more emphasized during the TT period, and offense was put on more weight later on as the whole team. Thus I think Casspi got the benefit there over TJ.
The one common denominator on this point is we all agree the defense needs to improve. Some will have you believe a 6 foot nothing guy is the key to your team defense. Others just want to see Lin play and can care less about what happens. I think both views are asinine. The key to playing good defense is the system. It all begins with the head coach, and there is no way around that. Absent this critical piece we are doomed to score 115ppg and lose by 2.
Here is the variable: If you replace TJones with Casspi, Bev with Lin (Lin-Harden-Parsons-Casspi-Howard lineup), that lineup allows 91.3 in 84 min over 16 games. If these stats really reflect what you claim, what we need to do is start Casspi and Lin. These stats are not good comparisons. You can't just use them as hard evidence to prove a point. Defense is a team thing, you can't just plug someone in to fix it.
... and the Big Three + Casspi + Beverley lineup allows 78.5 points per 100 possessions The bench lineups are another matter, but starting lineup to starting lineup comparison, over more than a dozen games, shows that SO FAR the four starters with Beverley not only has been significantly better defensively than the four starters with Lin, but has outproduced it offensively as well. I said it when Beverley went down: NET RATING. Lin still has time as the starter and more data will come in ... he can still prove he is the better option. And if he is, it will show in the numbers. Up to now, that has not been the case.
This has been addressed in earlier posts. What you basically did was add an additional variable (Casspi) in an apples-apples comparison (or closest possible incarnation of one). Now you made it an apples-oranges comparison. To make it back to an apples-apples comparison, you add the same additional variable (Casspi) to the Beverley Lineup, and the defense is actually 78.5 per 100 possession (Beverley, Harden, Parsons, Casspi, Howard) All the stats show in this case is the defense is better with Casspi instead of Jones.
This lineup rating is more an indication of how the team in general is playing at different TIMES than anything else. Harden and Howard are borderline superstars. The PG in this system has almost zero impact on their performance.
Hmmm, Based on similiar mins as the above line up (17mins), Brooks, Casspi, James, Dwight and Parsons has a rating of 76.6 per 100 possession! We should play Brooks more
TTNN and Durvasa, curious on whats your take on this. I am interested to see a break down of the top 3 lineup by each games. And also want to see the standard deviation between the games. As mentioned by TTNN's comments and durvasa's graph, there are some extremely high and low DRTG values for some of the games. Another factor I can think of is break down of Harden's points for those games. Since the offense runs through Harden, I would have thought he would have contributed the most differences for the offense numbers. We know Bev doesn't contribute as much offensively in the starting lineup comparing to Lin. So I was surprised to see the lineup with Bev has higher OffRtg number. With all the inconsistent defensive/offensive efforts, makes me wonder if this is really a good comparison between Lin vs Bev. If just looking at names in the lineups, we would think it does just because the other 4 players in the lineups are the same. But how much do those high/low games contribute to the differences? If we get rid of the data for the 3 high and 3 low games, would that make a big difference in the final values? How much was it really just Bev/Lin making the differences, and how much was it really the other 4 players making the differences?
Sometimes, the presence or absence of a player is a major factor in one of those really good or really bad defensive efforts, and that should be accounted for somehow. So I don't think just treating them as outliers to be ignored is the way to go. Bev/Lin is one of many factors, even when the other 4 players are the same. Matchups make a difference -- sometimes you just matchup better against certain types of lineups than othres. Differences in schedule which means some games players are more rested or better prepared because they had more practice time can be important. Then, on top of that, much of the volatility needs to just be chalked up to randomness. You could have 2 identical lineups facing identical opponents under identical situations, and there's still a decent chance of apparently big differences in the defensive numbers when we're talking about only a handful of games. So, even the +9 difference in defensive efficiency for the two lineups, which seems like a lot, is probably not statistically significant. While the +/- numbers might suggest that one lineup is better defensively than another, its important to realize that when looking at only a small chunk of the season you can't really draw strong conclusions from it. Its one piece of information to consider, but firm defensive assessments require much more.
The line up I gave which was 84 min over 16 games, was already small sample. But my point was these stats cannot be used as hard evidence because of the small sample and variances, so it was ok there. Clutch, the line up you gave, was in total of 17 min. If anything, it just shows that these stats cannot be trusted. Do we really think Casspi replacing TJones would be a big boost for our defense? Or maybe just when we play Casspi, it was against the backups? Or maybe the particular line up stats we used were just happens to be on the better or worse team match ups, since both Bev and Lin were injured for an extensive time? These stats don't show a lot. Personally I think one of the reasons Bev has better numbers, if anything meaningful, would be he's a good fit next to Harden, that is he has enough energy to help cover up Harden's lack of energy on court. Harden, despite being a great offensive player, lacks the energy when it comes to hustles, compete and defend consistently. His energy comes in bursts. Bev really shines here, he's active most of time, goes for off rebounds, great at 1on1 defense, he picks up the dirty work. But unfortunately he's not talented at offense end. That is where Harden comes in, but the problem is Harden's ISOs and lack of variety in our offense. When the playoff comes, some fouls won't be called, and if we match up with a team that has an elite wing stopper, Harden will be shut down, and then our chance of winning would be on...Dwight's FTs.