The curious case of “Italian statistics” (vids) – Part 1

{jathumbnail off}You have probably heard of the “Greek statistics” by now. It is a term used inside the European Union to describe “creative accounting” and it is, by now, an inside joke in Brussels and Strasbourg offices. But it seems that Euroleague got their own case of creative statistics over the past few seasons, in Italy.

Long life Euroleague fans have for sure noticed some irregularities and things not adding up in Euroleague games involving home teams from Italy. Infamous are Montepaschi Siena games where the home team would get credited with more steals than the number of their opponent’s total turnovers. And if you think we are exaggerating and that these were simply honest one-time-only mistakes here are more than enough examples: [example 1], [example 2], [example 3], [example 4], [example 5], [example 6] and so on.

Big deal you say? Maybe. But if we have a look at the 2007/08 stats, we will find a rather interesting pattern emerging.

And amazingly enough these statistics are always keen in improving the image of Italian teams and never the away teams.

The latest example of “Italian statistics” is last week’s AJ Milano – Panathinaikos game. The Greek side was credited with just seven assists in the game stats sheet, but a simple review of the game highlights proves that Panathinaikos players dished double the assists, fourteen. So we decided to run the story. It will be presented in two parts, in the first we will analyse and document the mistakes of the Milano statskeepers and have a general talk about assists. So here goes:

{youtube}hM-9ySPrtuI{/youtube}

Panathinaikos assists breakdown:

Assist #Time
stamp
Assist fromAssist toActionScore
10:00DiamantidisFotsisCaS0-2
20:46SatoTepicCaS12-17
31:13TepicNicholasCaS21-25
41:36TepicSatoCaS26-30
52:00TsartsarisMaricL/D26-34
62:44SatpNicholasCaS34-46
73:04SatoDiamantidisCaS38-51
83:13DiamantidisTepicCaS42-54
93:24NicholasMaricL/D44-56
104:23CalathesBatisteL/D57-67
114:34PerperoglouBatisteL/D64-71
124:44DiamantidisBatisteL/D *66-73
134:53DiamantidisBatisteCaS68-75
145:15DiamantidisFotsisCaS71-81

Glossary: CaS: Catch and shoot, L/D: Layup/Dunk
* slight hesitation

Here is the official Euroleague boxscore of the game:

Even if one can invokesl the different assist counting methods and philosophies, at best just 1 out of the aforementioned 7 assists could be judged as “questionable”. As you can see, Dimitris Diamantidis (averaging 5.5 assists per game before this game) got credited with just 1 assist, while in reality he dished 5. Moreover the total number of assists credited are just 50% of what Panathinaikos players dished out.

TalkBasket contacted Simon Jatsch, a real Euroleague statistics guru who is doing fantastic and ground breaking work at his website In The Game. He was kind enough to take us through rules and statistics  in order to help understand the situation better.

TB: Do you concur that Panathinaikos players dished 13-14 assists in this game?
S.J.:
Yes. In my opinion they had 14 assists. There’s a slight shot fake/hesitation by Batiste after Diamantidis’ pass (at 4:44 in the video above), but I think it should count as an assist. So, that’s 14 for me. The others are really not up for discussion: crystal-clear assists. Nine of them simple catch and shoot situations without any sort of hesitation, the others are passes that lead to immedeate easy layups or dunks.

TB: What is the exact method to determine whether a pass is an assist or not. Is it an exact science?
S.J.: I’ll cite from the official FIBA stats manual:

An assist is a pass that leads directly to a teammate scoring, if and only if the player scoring the goal responds by demonstrating immediate reaction towards the basket. Only one assist can be credited on any score. Even if the “second-to-last” pass set up the play, it is not an assist. An assist is credited when a pass is made to a teammate who shoots and scores-provided the shooter’s immediate intention, upon receiving the ball, was to shoot and that intention was main-tained until the shot was taken. It does not prohibit an assist where the shooter takes time to balance or makes a small play to score, provided the scorer always intended to shoot.

A pass to a player in good scoring position who considers other options before deciding to shoot and score is not an assist. The score is the result of the action by the shooter alone, not the passer.

The distance of the shot, the type of shot, and the ease with which the shooter makes the shot are not factors when considering if a pass is an assist. Similarly, the number of dribbles taken by the
player who scores is also not a factor, unless his efforts are such that you determine that he did the work to make the basket, rather than the pass.

A pass to a player at half court who dribbles directly to the basket for a successful lay-up is an assist. However if that player has to divert to dribble around a defensive player, no assist is given.

The statistician should bear in mind that the more the scoring player has to do in order to score, the less likely it is that the pass is an assist.

An assist may be credited on a pass to the pivot shooter (lowpost), provided there is an immediate reaction on the part of the pivot shooter in attempting to score.

Consistent with the definition above, an assist is not awarded simply when it is the last pass before a basket or because it was simply a “good pass.” The determining factor in awarding the assist must be the amount of work the scorer does and the immediacy of the shooter’s intention to score.

It is not a science, but clearly the assist is one of the more difficult actions to keep track of as a statkeeper. “The determining factor in awarding the assist must be the amount of work the scorer does and the immediacy of the shooter’s intention to score”. That’s the key phrase in the official definition. It allows room for interpretation. However, on 13 of those Panathinaikos assists, there was no hesitation, the shooter had the immediate intention to score, and the only work the scorer had to do was to make the shot. Number 14, Diamantidis to Batiste, is one too for me. Slight hesitation, but always intended to score, low amount of work to be done for the scorer before completing the shot.

TB: Why do you think there is such a vast range of criteria? Euroleague vs NBA criteria and even in different Euroleague arenas the criteria seem to be different.
S.J.:
Well, NBA is obviously very different. For example let’s take a look look at Steve Nash’s 23 assists from the 2007 playoffs, in the video below:

There are a couple of baskets by Stoudamire and Diaw that they really create for themselves, but Nash still gets the assist. Precisely, it’s the scenes at 1:07, 1:42, 1:53 and 2:01. 0:31 are debatable as well. Those don’t count as assists in Europe. Clear case of artificially pushing individual stats. Not sure about the intentions. Surely record-breaking (or -tying, like Nash’s) performances make headlines, but knowing that plenty of decisions of all sorts base on those stats, I think it is a problem.

I don’t think there are different criteria within the Euroleague. If you look at the Panathinaikos assists, identical situations are handled differently. The statkeepers simply weren’t consistent in awarding assists.

These are, according to play by play, the field goals that the statkeepers assigned assists to:


Let’s take the very first play in the video: Diamantidis to Fotsis. Simple catch and shoot, clearly an assist. But not according to the statkeepers. However, they give an assist to Sato for his pass to Tepic at 0:46 in the video. Where’s the difference? Identical situations, simple pass to catch and shoot without hesitation.

Maybe the statkeeping team simply wasn’t well-schooled?

TB: Do you think it is something that could and is happening in any Euroleague arena or that it is an Italian problem in particular (in the light of the infamous Siena “more steals than opponents’ turnovers” boxscores)?
S.J.:
Milan and Siena are the only cases known to me, but I’ve never taken a closer look at the statkeeping except for this Milano-Panathinaikos game where I was pointed at by a friend. In Siena’s case it was obvious, since having more steals than the opponent has turnovers is not possible. I have no reason to believe that it is an Italian problem and I don’t want make baseless accusations versus Euroleague, Italian clubs and even AJ Milano statskeeping. After all, it was only one game. But maybe we should look at statkeeping more closely in the future. Everybody is relying on these stats to be correct, not only fans, but decision-makers in basketball too. It is a sensitive topic.


TB: Have you spotted any other statistical irregularities in Euroleague?
S.J.:
Since this season, Euroleague is awarding assists for passes that lead to free throws, which can be seen in play by play data. I think that’s a very interesting stat to keep track of, but it should never be mixed with regular assists. An assist, by definition, leads to a field goal. And I don’t like the fact that nobody told us about this.

TB: What would you suggest to solve the problem and enforce uniform criteria in all Euroleague arenas when it comes to statskeeping?
S.J.: I don’t know how Euroleague operates when it comes to statkeeping. Statkeeping teams have to know what they are doing. I’m pretty surprised at what happened in Milano. Those weren’t close calls but crystal-clear assists. I wonder what’s going on there.


That concludes Part 1 of the story. In the second part we will talk about how could such errors affect one of the best parts of Euroleague, the Euroleague Fantasy Challenge, and we will be speaking to someone who knows the Fantasy business like no one else does.