Turnovers: Do You Want Them Dead or Live? | Stat Geek Idol

March 22, 2012 - by Jay Cipoletti

This is a Sweet 16 submission in our inaugural Stat Geek Idol contest. It was conceived of and written by Jay Cipoletti of @Hoopalytics.

Last week I posed several possession-value questions. Two of them stuck in my head as I watched the rounds of 64 & 32 play out over the weekend:

What is the Floor Percentage margin between live ball and dead ball turnovers?

Is it worth taking more risks to create more of the former defensively and reducing risks to reduce live ball giveaways on offense?

As it turns out the answers are ‘Enormous, yes and yes’.

The findings also pointed out an upset that flew under the radar of Friday’s wave of surprises but may have produced the biggest headshaker of the weekend.

Turnovers: Kindred Spirits or Distant Cousins?

First, let’s take a snapshot of the relationship of turnovers to overall possessions, using last Thursday’s 16 NCAA tournament games. Over 2,080 possessions, the 32 teams combined to score 2,162 points on 1,060 scoring possessions, for an efficiency rating of 1.04ppp, a Floor Percentage of 51% and a points per scoring possession of 2.04.

On 346 of those possessions, teams committed a turnover. Teams on the receiving end of those turnovers scored 389 points on 186 possessions, an efficiency trio of 1.12ppp, 53.8%flr and 2.08ppsp. It would seem that while turnovers occur at a lower rate in the Dance than they do in the regular season – 16.6% to 20.3% – it is still advisable to force them as often as possible and to treasure offensive possessions.

The 346 turnovers were split fairly evenly between live ball (170) and dead ball (176). Once that distinction is made, two very contrasting views take shape.

Teams converted an anemic 46% of their 176 possessions following a dead ball turnover into scores, at a rate of 0.97ppp and 2.11ppsp. In comparison to Thursday’s averages, it would almost lead one to believe that drawing charges and the like isn’t worth the effort. Even compared to the 1,734 non-turnover created possessions (50.4% Floor, 1.02ppp, 2.02ppsp), dead ball turnovers produce substantially lower returns on the offensive end. This is not to say dead ball turnovers have no value – they are still defensive stops, and a team scoring at a somnambulant (read: South Floridian) rate of 0.97ppp is still preferable to a team unable to get shots off. What they are not, at least in this 16 game, 2,080 possession sample, are offensive amplifiers.

That title can intuitively be applied to live ball turnovers, and the numbers from last Thursday affirm those instincts. Teams transitioning off a live ball turnover scored on 105 of 170 possessions, a higher Floor Percentage (61.8%) than any team registered in the 16 games and significantly higher than the 51% overall conversion rate for the day. The 1.28ppp scored off live ball turnovers was exceeded only by Wisconsin’s 1.30ppp and matched by Gonzaga in their rout of WVU (an incredibly difficult game to endure for this Mountaineer). The 2.08ppsp was somewhat counterintuitive, which I will address later in a closer look at the Colorado vs UNLV game.

Before we get to the Runnin’ Rebs, a look at Thursday’s possessions off of live ball and dead ball turnovers, with non-turnover created possessions as the control:

Live Ball	Non-Turnover Created	Dead Ball
170 possessions	1,734 possessions	176 possessions
105 scoring poss.	874 scoring poss.	81 scoring poss.
218 points	1,773 points	171 points
61.8% Floor	50.4% Floor	46% Floor
1.28ppp	1.02ppp	0.97ppp
2.08ppsp	2.02ppsp	2.11ppsp

When you see the Points Off Turnovers graphic this weekend, and you surely will (likely in the Ohio – UNC game), understand that live ball turnovers are the favored son and his dead ball sibling is the black sheep of the family.

How To Kill The Live Ball Advantage

The last game to tip off Thursday was a 6-11 matchup between UNLV and surprise PAC-12 tournament champion Colorado. The Buffs entered the night 179^th in live ball giveaways at 9.7%, while UNLV was 30^th nationally in swipes at 11.8%. One could easily assume that live ball turnovers would be a problem for Colorado, and they were.

In the 74 possession game that ensued, Colorado coughed up live balls at 18.9%, nearly double their season average. In addition to their 14 steals, UNLV also benefited from 9 dead ball turnovers (Note: 2 steals were immediately followed by Colorado fouls and are considered dead ball turnovers for the purposes of this analysis). As a result, Colorado scored on only 32 of their 74 possessions, a 0.92ppp clip.

If all you knew about that game were the above two paragraphs, you would have been eagerly anticipating the matchup between Mike Moser and Quincy Acy on Saturday. Applying the averages from above, you could plug in 26 points for UNLV on 23 turnover-created possessions. All they would then need would be 43 points on their remaining 51 possessions, a 0.84ppp rate well below their season average of 1.08.

They lost 68-64.

In the first half, when Colorado jumped out early on their way to a 36-25 halftime lead, UNLV went scoreless on their first 6 turnover created possessions, scored only 1 point off 4 live ball turnovers and finished the half a combined 0.33ppp on 9 turnover created possessions on a Floor Percentage of 22.2%.

Their production improved somewhat in the second half, allowing them to cut the deficit to 57-55 with 4:19 left on a 3-pointer off of a steal. That happened to be their first made 3 off a steal after missing their first four attempts. While a number of factors contributed to UNLV’s loss, including but not limited to Colorado scoring on 32 of the 51 possessions in which they got a shot, those four empty transition trips stand out as game breakers.

Four steals. Four missed 3s. Zero points. Loss.

When the other 31 teams are converting live ball opportunities at 1.28ppp it has to be hard for @GregAnthony50 to stomach wasting 5 points in a 4 point loss.

Avert Your Eyes, Runnin’ Rebel Fans – UNLV Possessions

Live Ball	Non-Turnover Created	Dead Ball
12 possessions	51 possessions	11 possessions
5 scoring poss.	25 scoring poss.	3 scoring poss.
10 points	46 points	8 points
41.7% Floor	49% Floor	27.2% Floor
0.83ppp	0.90ppp	0.73ppp
2.0ppsp	1.84ppsp	2.7ppsp

If You Must Lose, Do Not Lose the Lesson

Fortunately not all of us had to endure that UNLV loss…some of us got our thrashing out of the way earlier in the evening. So what can we learn from UNLV’s pain?

Save the 3s For Later

The point in transition is to attack. The ridiculously high combination of 61.8% Floor and 2.08 points per scoring possession is more a function of getting to the rim and drawing and-1s than it is hitting 3s early in a possession. Although UNLV did not have any and-1s, they scored on 4 of the 5 live ball turnover possessions in which they got a shot at the rim.

Gimme That

Teams that do not get a shot off fail to score 100% of the time; in the limited sample above, teams converted live ball transitions at 1.28ppp. It doesn’t take too many of those possession couplings to create a sizable margin. That’s not to say teams should be toothless elsewhere defensively for the sake of forcing steals…surely Frank Haith would be happy to expound on that…only to say that playing a defensive style intended to create as many takeaway opportunities as possible the first 60-70 feet of a possession is a moderate risk, high reward approach, provided the last 30 feet or so are still dedicated to clogging passing lanes, contesting shots and rebounding misses. Rick Pitino would be the one to talk about that approach, minus the rebounding misses component.

If You Cannot Attack, Retreat

As I said, the point in transition is to attack the rim. If that option is not available, do not create the opportunity for your opponent. The impact of the possession increases the longer the ball is in your possession. Put another way – bad shots off a turnover are good shots, provided they come at the end of the shot clock, because it means you are actually getting a shot and not committing a live ball turnover. In the last 4 minutes of a game, it could even be argued (and I am more than a little intrigued by the argument), that an offensive set designed to create either a block/charge call or a shot clock violation is more valuable even than a scoring possession early in the shot clock.

That last point raises one more question that I asked last week – Are bad shots really the first pass in an opponent’s fast break?

I hope to answer that next week.