• Finalizado
    FINAL
    2015-09-12
    Argentina 71
    Venezuela 76
  • Finalizado
    BRONCE
    2015-09-12
    Mexico 86
    Canada 87
  • Finalizado
    SEMIFINAL
    2015-09-11
    Argentina 78
    Mexico 70
  • Finalizado
    SEMIFINAL
    2015-09-11
    Canada 78
    Venezuela 79
  • Finalizado
    Ronda 2
    2015-09-09
    Mexico 95
    Argentina 83
  • Finalizado
    Ronda 2
    2015-09-09
    Uruguay 69
    Puerto Rico 80
  • Finalizado
    Ronda 2
    2015-09-09
    Dominicana 103
    Canada 120
  • Finalizado
    Ronda 2
    2015-09-09
    Panama 62
    Venezuela 75
  • Finalizado
    Ronda 2
    2015-09-08
    Mexico 73
    Canada 94
  • Finalizado
    Ronda 2
    2015-09-08
    Panama 71
    Puerto Rico 78
  • Finalizado
    Ronda 2
    2015-09-08
    Venezuela 75
    Uruguay 77
  • Finalizado
    Ronda 2
    2015-09-08
    Argentina 92
    Dominicana 84

Valuation of Statistical Criteria

The Polytechnic University of Cartagena (Spain) leaded in 2009 an investigation about the importance of statistics in basketball. Matias Barmat, today in WorldHoopstats.com, has contributed to the final result.
1. One of the key factors to develop an index which enables us to evaluate quantitatively the player’s value (best known as player performance) in a game is an accurate definition. Some experts think that the most correct definition is that who refers to "the player’s marginal contribution to the team’s victory". However, others believe that the best player has to be the most valued, i.e. the numerical evaluation index should be set perfectly to classify players according to their value. How do you think we should define the "player’s value" in a game? (We are not asking for a rating system, but a definition of a concept: the value)

Step by step. An evaluation index only serves to measure everything for a player is computable from a strictly numerical standpoint. Well, unfortunately this is not enough to measure the exact player’s performance.

If we're talking strictly about numbers, it means the same to score 16 points unmarked than 16 points with a static strong defense with ball congestion. Also it’s the same to score 16 points in any other league in the world regardless of their level. In the same way I can exemplify this with any other statistical section. Well, we know that it’s not true.

There is an extra variable that is very difficult (if not impossible) to measure, such as a brave man-to-man defense. In short: every player has a specific responsibility even without the ball -being in attack or defense- conditioned by a plethora of tactical drawings which meet each one with a different purpose. This explains a little why sometimes a player who averages only 8 points but plays 30 minutes a game could be so important for a team.

2. Specialized magazines and press reviews often establish "qualitative" player valuation in the games. For example, in Spain, the basketball magazine “Gigantes” chooses the “Gigante del partido” ("Giant of the match"), and newspapers like "As" and "Marca" assign with 0, 1, 2 and 3 points (fail, pass, good, excellent) the performance of each player in every game. Sometimes, there are players with "worst numbers" qualitatively better valued than those with "best numbers". Concerning to valuations in games that you've followed, are you agree normally with those assessments? Do you think that those qualitative assessments better reflect the player’s performance in a game rather than those based solely on a numerical weighting?

Normally I don’t agree with this, because they are based on very superficial -and strictly numerical- criteria. I would even suspect that there is an extra-sports component in the rating which is very subjective and only reflects perceptions from who qualifies. In the best case, they take the player’s valuation algorithm as a guideline. The only way to optimally qualify a player is to sit down and watch the 40 (or 48) full minutes of every game and only then you can draw conclusions, because of that non computable factor I mentioned above.


3. ¿What’s your opinion about the rating system currently in place in Spain’s ACB? Please comment openly and give your opinion about its reliability, its weak points, etc.

The ACB rating system has its pros and cons.


There’s an advantage in terms that every ball is counted as one, regardless of whether or not is reflected on the scoreboard and in which form. Being a field goal, a 3-pointer, an assist, a steal, a turnover or a block, all the balls are worth as one. And that's good up to a certain point, because we can compute how many balls have been taken into advantage, and therefore we’re able to deduce how many possessions were successful. In that sense it’s like the NBA’s Player Efficiency Rating, but with a difference: it takes into account the blocks against (as a negative value) and also takes into account the faults (adding received, subtracting committed).


However, the ACB rating system has a disadvantage which shares with the American league, very important in my opinion: the minutes played, therefore the rating average per minute. Another important factor to take into consideration, is the effective possessions per game, it means, the game pace, something that other algorithms like TENDEX (official in the Greek League) do.

4. ¿What’s your opinion about the different election systems for MVP in ACB? Remember that the weekly MVP is chosen strictly by valuation, the monthly MVP by an experts committee, and the season’s MVP by experts plus popular voting.

It is highly questionable, but the reason ACB has adopted this election system for the MVP is actually quite simple and it follows some logics.


In a standard FIBA league, every team plays only once per week, generally on Saturdays and Sundays, and relatively all at the same hour. This makes the audience follow the real-time statistics from all games and, by relying entirely on a statistical basis, it increases a lot the expectation about who will be the MVP of the day. In addition, in a single-game-per-team basis, there’s no room for any kind of speculation, which itself could happen if the MVP of one month (4 or 5 games) had been computed strictly by rating. In this approach, I find it logical that the ACB Board decided to combine statistical factors provided by the evaluation index with a heuristic component (for instance, the experts committee).

The fact of leaving aside the ratings to compute for the MVP of the league to focus specifically on heuristics as well, and especially in the popular vote, is a matter of marketing taken from the NBA All Star Weekend that is more a defense of specific vested interests from ACB than anything else.

5. Tell us, in your opinion, if there is any relationship between the ACB player’s value and its market value, that is, if you think the ACB player’s value revalues or devalues a player.

The relationship between valuation and market value exists, but is not a linear one. The main reason is that a professional league is, above all, a business. Therefore the value of the player’s “show business” is more important than the statistical value he can contribute to a team. When we talk about “to prioritize” the show, that is, to sell and / or to promote the league, every self-respecting numerical assessment has absolutely no value, to the detriment of the added value that a player can contribute to the league. Priority is given to a "showman" over a serious player, because he "sells" more; for the same reason a renowned or well-known player is preferable over a “rookie”, although everybody has the same points average.

For example: do you remember Andrea Pecile, and his alter-ego Captain Sunshine?

6. There are other player scoring systems, apart from that used in the ACB or Euroleague (which are identical). In the table below there are some of them. Please rate from 0 to 10 the degree to which you believe that these systems best reflect the performance (it means, the value) of a player. Put an X in the second column in case you don’t know enough the valuation system to issue a considered opinion.

Rating (from 0 to 10)

I don’t know it enough

Valuation system of ACB/Euroleague

7

SEDENA valuation system (created by Francisco Navarro)

X

Valuation system of the French League (created by Jean Luc Monscheau)

X

Tendex (created by Dave Heeren) (official in the Greek League)

7,50

NBA Efficiency

7

Player Efficiency Rating

7

Net worth plus/minus

4

Adjusted plus/minus

4,50

Manley credits

7

IBM award

X

Hoopstat grade

X

Diamond Rating

4

Steele value or Global Tendex (created by Doug Steele)

8

Points Created (created by Bob Belloti)

X

Clearbout Quality points (created by David Clearbout)

X

Mays Magic Metric

6

Total Performance Ratings (created by Joe Schaller)

X

Individual Wins/Wins Produced (created by David Berri)

5,50

Points Responsible (created by Larry Lindsay)

4,50

Efficiency through data envelopment analysis

5

 

7. Do you consider the ACB Player’s Value when performing your work as a journalist and as an expert in basketball? Do you use some other alternative valuation system?

I use different metrics that are specific to the analysis of cumulated basketball statistics. Basically my analysis consists in using different "ratios" such as points per minute, rebounds per minute, rebounds per block, the well-known assists per turnover, assists per steal ... points produced after offensive rebounds per offensive rebound, offensive rebounds per total rebounds and, in turn, all of these divided by minutes played....

…field goals per minute (2-pointers per minute, 3-pointers per minute), free throws per minute and also points scored per 3-pointer made. I can predict the player’s position only by watching his cumulative numbers. Normally a shooting guard scores around 1.5 to 2 triples per game, more than 2 he's a superstar player, a "crack". so we can deduce that a shooting guard scores about 8 points per triple scored. Well, a point guard uses to score only 6 points per triple, as it usually either dishes an assist or a steals when he hasn’t the ball,... or shooting from outside.

Let me give you another example. Esteban Batista playing for Atlanta Hawks in the NBA 2005-06 season averaged 1.8 points and 2.5 rebounds. Well, in the 57 games he played in that year, only played 8.1 minute average.  It means, that for every 48 minutes played he grabbed 15.2 rebounds...


Matias Barmat (@worldhoopstats)

Lecturas: 529 - 2011-08-02