The Diff: Byron Scott, blown leads and the second-half Cavs

Last week in The Diff, I brought you inside the numbers of the Sweet 16. With no more March and no more college basketball for Ohio sports fans, it’s time to move back to the pros. Although the current excitement surrounds the Tribe’s Opening Day win, I’m back for Cavs talk today.

The Diff

During the Cleveland Cavaliers’ ongoing season-high eight-game losing streak, much of the media talk has been directed toward head coach Byron Scott’s future with the organization. I covered the beginning of the talk about 10 days ago. Anonymous players then shared their frustrations with the Akron Beacon Journal’s Jason Lloyd over the weekend. And our very own Scott threw his hat into the ring on Monday. But looming large in these murmurs is this largely unreported fact: The 2012-13 Cavaliers are currently the worst third-quarter and second-half team in the National Basketball Association. By far.

In the usual narrative of the Cavs recent rebuild, a solid chapter or two is spent on the much-hyped, high-drama heroics of a certain Mr. Fourth Quarter. In just 101 (out of 144 possible) career games, Kyrie Irving has proven that he deserves to be on the elite short list of the best young talents in the NBA. He’s shocked a lot of folks that might not have expected him to be this good this early. And he’s single-handedly brought the team back from countless deficits in his short career. That’s certainly true.

But in these two seasons, for every one Cavs victory, on average, there have been about 2.3 Cavs losses. So for as many spectacular, incredible and incomprehensible fourth-quarter comebacks by Irving company, there have usually been at least 2.3 times as many bone-headed third quarter miscues, confusing rotation patterns or just downright awful second-half play — in other words, lots and lots of blown leads. A near-historic amount this season, in fact.

As you likely can tell, I’ve got some statistics to back up these lofty claims. For years and years, has kept track of team records based on leads at different intervals of the game. So, without further posturing, here’s your 2012-13 leaderboard in winning percentage when ahead at halftime and when ahead at the end of the third quarter:

RkTeamWLPct RkTeamWLPct
NBA SINCE ’0528.510.80.725 NBA SINCE ’0532.47.20.819


Take a look at my footnote for source information and a notable caveat1. But there’s one thing that should be most glaring to anyone from this chart: The Cleveland Cavaliers rank dead-last in both categories.

At some point, your mind likely will then wander to consider the effects of an already good team on these categories. Inherently, a very good team2 is great at maintaining a lead — or evaporating an opponent’s lead — while bad teams suffer the consequences. Indeed, the correlation r-values support that hypothesis too3.

The next question you might ask: How do the ’12-’13 Cavaliers compare to the rest of this data set over the last eight years? Well, not so good. And in a way, a bit too poorly compared to their overall record. Here are the bottom 13 teams in both categories:

RkTm82WS PctYear RkTm82WPctYear


Overall, the Cavs’ 19 losses when ahead at halftime are the fourth-most since the 2005-06 season. The record is 21. Additionally, the Cavs’ record when ahead after three quarters is proportionally worse, but 19 teams have had 12+ losses in this category over the past eight seasons. The record is 15.

For this historical data, it was now intriguing to compare the team’s split percentage to their per-82-game overall success. So yes, over the course of the last eight seasons, the current Cavaliers have been one of the worst in each of these categories. Only four other teams — the 2007-08 Miami Heat and three separate iterations of the Minnesota Timberwolves4 — place on both of these bottom 13 lists. Only that one Miami team is worse than Cleveland in both splits.

As I teased above, it does appear that per the overall team success, Cleveland’s ability to blow leads is a bit unprecedented. In the “Ahead at Halftime” list, only one other team had a winning percentage projected to over 24 wins in an 82-game season. In the “Ahead at end of Third Quarter” list, there were five such other teams along with the Cavs. In essence, I propose that the Cavs’ second-half failure is a bit strange considering they’re not that awful overall.

That hunch then led me to another data portal via efficiency differentials by quarter and by half. As a quick primer for those unfamiliar again with efficiency differentials, it measures the net points of a team compared to its opponent per 100 possessions on the court. Splitting the Cavs season into three distinct periods, I present this next table5:

SplitRec. 1Q2Q1H 3Q4Q2H ALL
Thru 12/215-23-4.1-9.0-6.5-10.8-14.1-12.4-9.5
Since 3/12-13-
SEASON22-51 -3.5-1.7-2.6 -11.8-7.4-9.6 -6.0


Fairly accurately, this chart above epitomizes the current Cavaliers season. In the first split, the Cavs were a very awful team, despite the health of Anderson Varejao. There was no bench to speak of during this 5-23 start to the season — players such as Samardo Samuels, Jon Leuer and Donald Sloan were receiving meaningful playing time. That’s certainly why, when Varejao, Irving and the other starters sat more in the second or fourth quarters, the bench struggled mightily. Overall, second halves were atrocious. This was a sign of a bad trend.

In the second split of the season, post-Varejao’s season-ending injury, the season started to turn around. The main highlight was the Leuer trade that brought a more complete bench to Cleveland in the form of Marreese Speights and Wayne Ellington, plus the signing of Shaun Livingston. This is also where the quarter-by-quarter efficiency flip-flopped: Now, the Cavs were most successful in the second and fourth quarters with this cohesive bench. The team was much more exciting to the tune of a 15-15 stretch and showcasing a top-10 offensive efficiency.

Since March 1st, however, the season has suddenly gotten ugly again. Just as many fans were truly starting to enjoy the year, the team fell apart again. And where has this destruction been most noticeable? In the third quarter and second half overall. Shockingly, the Cavs are not the worst third-quarter team since the start of March — that belongs to the recently-awful Detroit Pistons with a -32.5 mark. But they are the worst second-half team overall and it’s a stark contrast to the perfectly mediocre success (0.0 net rating) in the first halves of the last 15 games.

On the season, this is where the Cavaliers rank in the NBA by quarter and half in offensive efficiency, defensive efficiency and net efficiency differential:

SplitO EffRkD EffRkNetRk


All of the storylines above again fit this narrative. Now it’s spelled out in black-and-white for all to see: Cleveland is the worst third-quarter and second-half team in the league. The defects are seen both offensively and defensively. And for more perspective, keep in mind that the average NBA efficiency is about 103.0, per HoopData. So while the Cavs offense is still fairly bad overall, it’s the defense that is most off-putting6.

As one final topic, last month, the 7th annual Sloan Sports Analytics Conference took place at MIT. Our very own Craig attended the conference this year. Although Craig wrote a few articles (this and this) about how the math overload relates to the future of the Cleveland Browns, the bulk of Sloan still relates to basketball statistics. Guys like Daryl Morey are superstars at the conference and, as the NBA is the best sport to follow on social media, there were tons of other great commentators and writers at the conference.

One of the most hyped research papers out of Sloan this year was one untitled “Live by the Three, Die by the Three? The Price of Risk in the NBA“, written by Matthew Goldman and Justin M. Rao. The paper dealt with the risk-friendly or risk-averse nature (as measured by percentage of three-point attempts to overall field goal attempts) of NBA teams when leading and when trailing. Thus, there’s potential for a slight overlap with the current Cavs situation. It’s not a perfect fit, but it’s interesting nonetheless.

To a certain extent, perhaps the Cavaliers — and Byron Scott — aren’t making proper risk adjustments in the second half of games this season. The team is playing a slightly slower pace in the second half. Three-point frequency and other aspects of the Four Factors didn’t appear to change drastically, but it’s the entire mindset of the franchise that potentially needs to feel more comfortable to be risk-friendly when games are still within reach — and especially when the lead is actually in hand heading into the final 24 minutes.

Eventually, Byron Scott’s future will depend upon the team’s ability to hold onto leads and sustain its success throughout the second halves of contests. What is happening this season, and especially since the start of March, could be characterized as either extremely poor in-game coaching, fatigue or maybe just plain tanking. It’s not fair to make a one-sided judgement call and label it one without considering the other But this near-historic failure in the second half is a missing component of what is turning out to be another dark and dreary season at the Q.

  1. Source: 2012-13 NBA data, 2011-12 NBA data. You can then use last year’s to look at the previous years as well (replace 2011 with 2010, etc.). For the historical average, I’ve started with the 2005-06 season since the data from the season before that appears to be incomplete on the NBA website. Also for complete statistical accuracy: I’ve averaged the historical NBA records under these two splits to be per an 82-game season. Obviously, last year there were only 66 games and we’re not done yet this season, so this was a necessary adjustment. []
  2. Miami, Oklahoma City and the LA Clippers both appear in the top four of these categories. Those first two teams certainly are the odds-on favorites to meet again in the NBA Finals this season []
  3. R-values are .901 for when leading at halftime and .820 when leading at the end of the third. []
  4. At one point, I started looking at three-year or longer intervals of success or failures in these categories. Here’s your Minnesota fact du jour: In the six-season stretch 2005-2011, the Timberwolves went 84-95 (.469) when ahead at halftime. Look at the chart above to place that in historical context. It’s ugly. Really, really ugly. []
  5. The Cavaliers have only played in one overtime game this year so for the purpose of an easy table, that was generally overlooked below. []
  6. I’ll get to this another day, but I think Scott was right on the money a few weeks ago: A defensive presence on the interior is a huge need long-term for this franchise. The Cavs just can’t stop anyone in the paint right now. And even with Anderson Varejao, things weren’t that different. []

  • This is excellent.

  • Ezzie Goldish

    For conspiracy theorists’ sake… (and I’m not buying into this theory I’m completely making up, I’m just noting it for kicks since some friends and I have had fun with this on Facebook)

    A skeptic *might* say:

    The Cavs realistically need one more great pick to make this team a contender. But they also can’t alienate the fans completely by playing horrid ball all year, so… they start off the year playing exciting first halves – people see potential, etc. – but “odd substitution practices” lead to a lot of second half collapses. Didn’t the Cavs have one of the top lineups in the NBA, but never use it during the 4th in Oct-Dec?

    Then Andy goes down, and losing is more palatable to fans who understand they’re missing their MVP to that point, so they play the way they would, get bench pieces for the future (can’t pass up that deal)… but somehow, they start winning. Too much. Whoops!

    So Kyrie gets hurt… and then after some discussion, it turns out to be serious. Then Waiters pulls a great run, and they still win a few… and then of course he gets hurt. Meanwhile, with about 7 teams battling for that 3rd-4th spot in the lottery, the Cavs take no chances – second halves require some serious adjustments to make sure they have no chance. (I mean, come on – to play even in the first half but be insanely demolished in the second like that??)

    After all, if you’re Chris Grant, you can’t plan this all any better… Unless you can. 🙂

  • JacobWFNY

    Thanks much Tom. I had a feeling you’d enjoy.

  • JacobWFNY

    Absolutely, the whole “tanking” storyline is a fair point. It’s ideal for an NBA team to be awful as opposed to simply mediocre. And with the crowded nine-team group between 3-11 in the lottery order, the Cavs had been looking like a 6-8 team. Now they’re destined for 3-5.

    So yeah, I’d buy that theory. Just don’t know if it’s the actual reality. Regardless, still incredibly fun to see these shocking stats.

  • Ezzie Goldish

    Yup. Also – another excellent piece. One note on a number of these pieces – more explanation of some of the stats used either in-line or in footnotes (I know sometimes you have). It takes a bit to figure some out at first.

  • JacobWFNY

    Thanks Ezzie. I’ve been trying to do that a bit more as I know my mind works differently with numbers than most. Appreciate the feedback.

  • mgbode

    I don’t think they’d intentionally lose games. But, I think they’d intentionally sit “somewhat” healthy players. So, that doesn’t jive with the 4th quarter meltdowns.

    Overall, tanking is a stronghold of April in the NBA. Just take a look at the standings for the last 10 games. Orlando, Cleveland, and Phoenix all 1-9. Detroit is 2-8. Charlotte is a whole 3-7, but they now only sit 1.5 games up(back?) of Orlando, so expect them to not win many more down the stretch.

    New Orleans and Sacramento are still playing hard. Their reward? their lottery odds have dropped to 6th and 7th overall respectively. 10games ago, these teams were within a 1/2 game of the Cavs (now 3.5 and 4.5).

  • mgbode

    great piece as usual Jacob. the defensive efficiency of the 2nd half is especially apalling.

  • JacobWFNY

    Thanks mgbode.

    I think we all knew the defense was bad. But what I’ve been most surprised about is that the offense has struggled so mightily of late.

    Season efficiency: 101.1-107.1 (-6.1)
    Best efficiency: 106.4-107.0 (-0.6) (during all play of 15-15 stretch from 12/22-2/28)
    Worst efficiency: 92.8-114.1 (-21.3) (during second halves of 2-13 stretch since 3/1)

    Thus, the defense was quite similar during the best stretch and the season-long efficiency. It’s dropped by 7 points during this awful stretch.

    But the offense has been most variable. It jumped 5 points during the best stretch. Now, it’s down 8 points per the season average.

    So relatively speaking, for as bad as the defense has been, the offense has been more slightly abnormal of late — especially per the team’s height of offensive success and relatively poor defensive play all year.

  • mgbode

    but, are we really worried about the offensive efficiency faltering while we take away virtually all of our offensive weapons and force our 2nd unit to go against other teams starters?

    the poor defensive efficiency seems to be a more systematic failure within the system and falls on Byron as well. no matter the personnel, we stink on defense. that’s not good.

    is it easy to find an offensive efficiency w/ Kyrie and w/o Kyrie?

  • JacobWFNY

    Yup. Here ya go:

    With Kyrie: 1749 mins, 103.3 o eff, 107.7 d eff, -4.4 net
    W/o Kyrie: 1764 mins, 98.8 o eff, 106.6 d eff, -7.8 net

  • BenRM

    I would label this trend “disconcerting.”

  • mgbode

    thank you sir.

  • woofersus

    So, at the beginning of the year when we had no bench depth, the team sucked hard when the reserves were in the game despite being fairly good when the starters played. Then when we added Livingston, Ellington, and Speights (and Miles came out of his funk) and ditched Pargo and Sloan, and buried Gibson and Casspi deep on the bench, that flip-flopped with the reserve unit being impressively efficient. Now, without Kyrie and Dion, with Livingston and Ellington being forced in to starting roles we are back to having no bench depth and have reintroduced Gibson, Casspi, and a couple of D-league players to the mix and stunningly, we are again awful when the starters come out. Theories anyone? It couldn’t be because of bad players, right?

    Don’t get me wrong. I have concerns about Scott’s rotations too. I’m just saying if the talent isn’t there, the results will be predictable. I’d like to see the players on the court and the minutes they played from the peak of the lead over Miami until the game was tied back up. I bet you see a lot of names that wouldn’t make a playoff roster.

  • woofersus

    In this case, two of our most important offensive weapons are also defensive liabilities. It’s hard to underestimate the effect of youth on those defensive efficiency stats. Whether or not Byron is doing a sufficient job at teaching the defense is unclear to me, but it is clear that both Kyrie and Dion just get lost on defense pretty often, even when they are trying hard.

  • mgbode

    that’s my point too though. the offensive efficiency seems to fluctuate depending on players. the defense has been terrible no matter what.

  • JacobWFNY

    Just to add to the conversation:

    With Waiters: 1667 mins, 103.5 o eff, 109.3 d eff, -5.8 net
    W/o Waiters: 1846 mins, 98.9 o eff, 105.2 d eff, -6.3 net

    Relatively similar offensive effect as Irving. Worse defense here with Waiters.

    But worth reminding: Net efficiency ratings per player are relatively fickle; they’re practically plus/minus numbers, wrapped in a per-100 possession blanket.

  • JacobWFNY

    Agreed on lots of your points. As I mentioned in the post, the numbers do indeed seem to fit the season’s overall storyline. Yet the second-half numbers for the season still are relatively historic. It’s bad, and significantly worse than the first-half numbers. It’s not entirely explainable by bad players; otherwise, the third quarter would seem to mirror the first quarter more.

  • mgbode

    ok, that begs 1 more data mining exercise if it’s easy enough to compile:

    with Waiters and Kyrie on the floor
    without either on the floor

    since they both seem to have the same effect on the offensive efficiency in the above sets we can interpolate what happens when either is on the floor to about the same. but, does the offensive efficiency drop even more when neither is seeing floortime?

  • woofersus

    I agree, but my argument is that we have some players who are good at offense and not so good on defense, and some other players who are bad at both. I would be more concerned, except defensive struggles are normal for young players – especially scorers. I remember years ago when people were concerned about Lebron’s defense too.

    I recognize how bad it is, but I don’t currently see a reason to think it’s a scheme problem or that the development is abnormally slow. I know we sort of expected a step forward this year, but we did lose two starters (Jamison and Parker) and add a pair of rookies to the rotation. (both now starting)

  • mgbode

    if losing Jamison and Parker doesn’t help the defense, then I rest my case.

  • JacobWFNY

    Per again, here’s your stats of with BOTH Waiters and Kyrie on the floor: 970 mins, 104.1 o eff, 108.8 d eff, -4.7 net

    Then, since I couldn’t really easily find the answer via the website, I had to do some math. Here’s my estimate of the data with AT LEAST ONE on the floor: 2446 mins, 103.1 o eff, 106.4 d eff, -6.5 net

    Now finally, based on continued estimates, here’s your data with NEITHER on the floor: 1068 mins, 96.4 o eff, 104.2 d eff, -7.8 net

    A few quick related and additional statistical observations:

    — I was taken aback at first: How are the Cavs worse offensively with “at least one” than w/ Irving, w/ Waiters and w/ both? It just shows that w/ both impacts their overall play. The w/ both data skewed it overall.

    — Not to be forgotten, yes, the defense is quite bad with either on the court and is much better without them.

    — The Cavs do indeed play a much higher pace noticeably with Irving on the court (96.81) as opposed to without him (93.21). That could be a factor.

  • mgbode

    ok, i really need to spend a couple hours this week on and figure this out on my own. thank you very much for data mining all of this information. great tidbits in there including showing how much work they need to do on their defense in the coming years.

  • JacobWFNY

    Indeed, indeed. You’re welcome, sir. That’s why I’m here. I love, love, love getting research requests. If you ever have future ideas for The Diff, always let me know.

  • woofersus

    Replacing Parker with Waiters certainly hasn’t been a defensive upgrade. At least not yet. There are other factors though. Last year, until he got hurt, we saw a lot of Daniel Gibson, who was a decent defensive player, but not much of an offensive player. This year he’s on the bench and we’re playing Waiters, who is much better offensively, but still lost on defense. Varejao played more of the season last year too, and he’s one of their better defensive players. Another factor I’m suspicious about but haven’t had time to look up is the number of possessions per game. With the younger, more athletic players it would seem the pace is faster (particularly with the starting unit on the floor) and will result in more scoring for both teams.

    I could be wrong. Maybe Scott is a poor teacher of defense and the scheme is just not well organized. What I’m seeing though is a team that outscores opponents by a fair margin even with not-great defensive efficiency when certain guys are on the floor, and a team that just sucks when some other guys are on the floor.

  • mgbode

    all the stats Jacob provided indicate that we just get outscored by more when our best players are not on the floor (we score more, but the other team does too).

    he also showed that we play at a faster pace w/ Irving. this should largely be taken into account in the efficiency stats but it’s true that faster pace often results in more fastbreak opportunities (for both teams), which increase the Oeff and decrease the Deff.

    my main concern is that I cannot tell you what our system is on offense or defense, that it doesn’t seem like Scott is figuring out the best things that players do and getting them to do just that or that he’s able to mask any defensive deficiencies.

    now, he’s been dealt a poor hand to show some of these things, but, at this point, I think he’s going to be let go.

  • woofersus

    I agree with the concern over what appear to be offensive and defensive systems that aren’t developing well. I’m not sure which things are coaching failures and which things are player failures, but there’s probably a little bit of both. I wouldn’t be shocked either way with regards to his future.

  • woofersus

    Wow, that’s really interesting. Thanks for the research on that!