Slow Starts

2010 January 8
by durvasa

To date, the Rockets have arguably had the most difficult schedule in the league and have come out of it with a 20-16 record. That’s better than almost anyone would have predicted. But with the Rockets now experiencing their first 3-game losing streak after a tough road swing through New Orleans, LA, and Phoenix, it is difficult to look past what has, of late, been a glaring problem — slow starts. In a recent tweet, Daryl Morey expressed the concern:

We are just not playing well.Not getting good shots¬ making shots we get.TOs.Start of 1st&3rd needs new approach.AB/Carl kept us in it.

– Daryl Morey (twitter)

Charting the Rockets Starts Over the Course of the Season

How bad have the starters been performing of late? Below, I’ve charted their offensive and defensive performance in the 1st quarters for each game, looking only at possessions that came before the first Rockets substitution. The bars represent efficiency for each game, while the lines are a “last 4 games” average. Red color represents points scored, while blue color represents points allowed.  Note that there is a blank for the game played at Oklahoma City on 11/29, because that is the game in which starter Luis Scola was injured on the first possession.

There has indeed been a recent downturn in the 1st quarter starts, but of late the Rockets have only trailed the last two games  — a back-to-back on the road against the Lakers and Suns. It is troubling that the Rockets would have had such a difficult time scoring against the poor Phoenix Suns defense (they managed only 7 points in 12 possessions, while allowing 21 points). But should we be concerned with the larger trend? Before the Lakers game, the Rockets starters ran out to a favorable start in 4 of 5 games, and they played the Hornets evenly on the road. And for the first month of the season, the Rockets generally played well at the start of games.

But 1st quarters are only a part of the story. This is what the starters have done to begin 3rd quarters:

Now we see a much more troubling trend. In 7 of the last 9 games, Rockets starters have been outplayed to start the 3rd quarter (before their first second-half substitution). The offense has ranged from mediocre to anemic over that stretch, and the defense has struggled also on average (with particularly poor defensive efforts coming against Orlando, Cleveland, New Orleans, and Phoenix).

Below, I summarize the net results (that is, points scored minus points allowed) for the 1st and 3rd quarters. Here, the red color refers to the 1st quarter start, the blue color refers to the 3rd quarter start, and the green color refers to the combined net efficiency to start both halves. From this, the starters did their job pretty well through the first month of the season. At the beginning of December, the team had a string of poor first quarter starts, and over the last few weeks the team has struggled in their second half starts.

The following summary stats indicate how poorly the Rockets starters have sputtered since the first month:

Entire season ...
      start_1st      start_3rd  start_combined
OFF      107.0          98.1          102.7
DEF      105.1          99.3          102.4
NET      +2.1           -1.2          +0.3
Since beginning of December ...
      start_1st      start_3rd  start_combined
OFF      100.0          90.0           94.8
DEF      106.9         102.4          104.6
NET      -6.9          -12.4          -9.8

The defensive efficiency has been respectable, but we expect at least that with a unit consisting of Shane Battier, Chuck Hayes, and Trevor Ariza. It has not been close to good enough to overcome the awful offense.

Patterns in Starters’ Performance and Final Margin

Strangely, of late there has been a large divergence in the performance at the start of each half. An interesting pattern emerges when I plot net efficiency in the 1st quarter versus net efficiency in the 3rd quarter (below). The 14 times the Rockets starters have “lost” in the 1st quarter (the left quadrants), they’ve “lost” the 3rd quarter 9 times (the lower-left quadrant). But in the 20 times the Rockets have “won” the 1st quarter (the right quadrants), they’ve “won” the 3rd quarter only 9 times (the upper-right quadrant). This would seem to suggest that when the Rockets starters play poorly to start the game, its because they’re typically outmatched. But when the Rockets starters play well to start the game, it is much less an indicator of superiority — opposing starters will adjust to the Rockets more frequently than the Rockets can adjust to the opposition.

How are these slow starts impacting the Rockets ability to win games? Ultimately, this is the concern. Below, I have plotted the final margin against the combined performance to start the 1st and 3rd quarters (which will usually represent about 1/4th of the game).

Interestingly enough, these result show that the starters are not losing their portion of the game any more often than the team loses overall. The team’s overall win-loss record is 20-16, but if we only consider the start of the 1st and 2nd halves, they have a “win-loss” record of 21-15. If we only consider the start of the 1st half the record is 21-14 (very good), but considering only the start of the 2nd half the record is 14-20 (opponents are adjusting, as I said earlier).

Closing Remarks

Being a team of role players, there is much less of a gap (perhaps no gap?) between the players that are 1-5 in the rotation versus 6-10. Consequentially, it should come as no surprise that the starters would be outmatched on many nights by opposing, star-laden units. But because of the depth and talent on the bench, the Rockets have been well-equipped to overcome this. The numbers show that the starters have played pretty well when looking at the season as a whole, but they having been trending seriously southward since the beginning of December.

It should be fairly obvious that major improvement on the offensive end is necessary. The competition should ease up over the coming weeks. This could be an opportunity for the current starting group to regain some of the cohesiveness and chemistry they seemed to have in the month of November. I could also see a change in the starting lineup, with Carl Landry replacing Chuck Hayes seeming most likely.  I am hesitant to endorse that particular change, as the defensive performance with Scola and Landry on the floor has been consistently awful for years, but it is possible that Landry’s insertion would jump start the offense enough to offset this. Other suggestions I’ve read from other fans include replacing Chuck Hayes with David Andersen or replacing Trevor Ariza with Chase Budinger or Kyle Lowry. A combination I would like to see more of is Carl Landry and Chuck Hayes, but I understand the concern in using such a tandem. With Hayes having effectively no shooting range, the defense is free to load up on Carl Landry and (likely) reduce his effectiveness as an interior scorer.

Bookmark and Share
  • physicsgeekandrocketsfan
    rahat, great job with the site. a question or two for your stats:

    a) when you do these per game stats like above you should normalize to how good the opposing team is
    ie: the rockets should be getting off to better starts against worse teams. even though we are on a three game skid, those were three pretty tough teams we played

    b) if you're claiming theres a pattern or periodicity to their performance use fourier analysis on your data. it will make it very clear if the rockets offense or defensive performance goes up and down periodically

    c) where do you get numbers like first quarter performance and such? im guessing you dont have a stat sheet cause you jot notes during the game?

    anyway great job

    as for a line up change, im dying to see more lowry (hopefully not at the expense of brooks!)
  • durvasa
    Thanks for the feedback.

    a) The teams faced were listed in the charts so that the reader could get a sense of which teams the good/bad performances came against. It can be tricky to really gauge opponent strength accurately, and I didn't want to complicate the discussion too much by adjusting the raw performance numbers. My guess, also, is that it would not significantly change the shape of the curves. Because for each game we are only talking about maybe 20-30 possessions, the ORTG/DRTG will be spread out much more. The impact of any opponent adjustments would, as a result, be lessened.

    b) Not claiming periodicity. It was more just looking at the relationship between winning in the 1st vs winning in the 3rd, and then the relationship between winning the starts and winning the game overall. Though since December, I have noticed an inverse relationship between winning the 1st and winning the 3rd (see the first half of December, and the last 3 weeks). Not too sure if that's a reflection of anything significant, though.

    c) The source of the numbers is the play-by-plays. In particular, the excellent provides play-by-play data in Excel-friendly CSV files.
  • Stephen
    I too believe starting Budinger would be the better move as his passing into bigs,ability to shoot 3s and some demonstrated ability to drive are qualities that could help the starting unit. However,I'd move Battier to bench. And if Adelman had another SG who could actually shoot,I think he'd have done it a while ago. An Ariza/Bud,Battier/scorer-shooter rotation would be far more balanced and would help keep Shanefresh for Playoffs.
  • physicsgeekandrocketsfan
    Just out of curiosity, would landry be ok with seeing a rookie start when landry is averaging basically 17pts and 6 rebs a game? i know they have different roles but do you think landry wants more playing time?
  • durvasa
    Landry doesn't seem like a guy who'd complain about that. He's very humble about his role on the team, at least from his public comments.
  • luislandry
    The points you make about why not to pair Landry with Hayes seem correct. I don't know that they've played enough minutes to be significant statistically, but from observation, I'd think that Landry (a worse passer than Scola but better inside presence) would really suffer with a non-shooter on the floor. Scola is not the focus of the offense when he's on the floor, and is more capable of finding Chuck for a layup when he's in. Agree with Rahat, the change should be at a wing position.

    Has something been off for Chuck lately, though, or has he not been guarding the guys who lit us up? Minutes are down because of Carl, but David West score a bunch, Frye hit six 3s, which I would think a quicker smaller guy like Chuck could close out on better...
  • Hayesfan
    For some reason they have been defensively been switching Chuck and Shane off the best offensive players to allow them to "help" more. It's actually be counter intuitive though because instead the productive players get hot and then there is little Chuck or Shane can do. Plus coach has been willing in the last few games to give up points to get points (sacrificing the defensive play on the court to keep offensive players in the game.)
  • rahat_huq
    I would hesitate pulling Hayes from the starting lineup for Landry because, based on comments I have read, and basketball norms, the key in 1st quarters is to establish offensive flow and as much equity in usage as possible. having landry out there, i think you're wasting his minutes because i think his best usage right now is as a go-to option (the irony, considering my earlier remarks). you want to feed the hell out of carl every second he is on the floor.

    to me, the most obvious solution is starting budinger in place of ariza. while not nearly as "effective" as landry, there isn't a guy on the team who makes a more visually discernible change in the offensive flow than budinger. he's shane battier if shane had paid attention in ABCD camp and learned how to dribble. of course, this will never happen, but i do think this would be the best route.
  • bob schmidt
    I like the idea of inserting Bud for Ariza on an experimental basis. Bud's one start when Ariza was suspended was a good indicator that he could contribute in that role. I just wonder if Ariza is "coachable" enough to accept a role off of the bench.

    I'd love to know the current PER ratings from Hollinger of our present roster. Sorry to say I'm not an "insider", but hard to justify the expense.........Maybe someone who is could post the stats?
  • durvasa
  • bob schmidt
    Thanks! Fact-based opinions are always better than gut-feeling opinions. Keep up the interesting blogs.
  • Hayesfan
    durvasa stuff like this is why I love you! I have to believe that part of the downward trend can be attributed to the very tough schedule and sheer insane number of back to backs we have played in the month of December. The question then becomes.. is the trend something we worry about now, or do we see wait to see what a few days off and some practice time does to improve those trends.
  • durvasa
    Adelman usually has a lot of patience. I think he'll look at the schedule, see that's easing up, and stick with same rotation for the time being. If a change is made (and it could happen with a couple more bad outings like the Phoenix game), I think it will be Landry in for Hayes.
  • Hayesfan
    I think that's most likely, because it's "easiest" for coach. Chuck is less likely to be bothered by coming out of the starting lineup than anyone else not named Battier. Plus you can see it already happening.. when that lineup struggles, Chuck is the first one pulled.
blog comments powered by Disqus