Guest Post: Improving MLS Quality by Leveraging its Partnership with Opta

We are once again fortunate to have a guest post from Dave Laidig.  Dave is a corporate contracts attorney who occasionally submits posts about soccer.  He resides in Minnesota, and laments the Panama result.  Once again, thanks to Dave for his great insights. 

Soccer is known as the beautiful game, an implicit acknowledgment of the art underlying the sport.  And art frequently defies objective evaluation.  Can we describe Picasso with numbers, or affirm that Hemingway is 20% better than John Grisham?  We know such descriptions are futile.  And we carry that sense of futility over to the art of soccer.  We all appreciate a no-look heel pass, but how do we value the consistent vision of a midfielder?  And this is, quite literally, the million-dollar question. 

However, soccer has an objective end product – wins.  And wins are the result of objective events, goals scored or allowed.  The events of the game are defined, and their contribution to the ultimate result can be measured through performance metrics.  This data is valuable, as demonstrated by the recent partnership between Opta and MLS.  And as discussed in earlier commentary, some of this value is derived from the marketing benefits of advanced game information.[1]  But a large part of the value of better performance data is in the area of labor management.  Specifically, advanced performance metrics can adds accurate, objective information that can supplement talent evaluation, increase cost efficiencies and support pricing for transfer sales

Importantly, the performance metrics are valuable if they intrinsically describe the game; such as data of the fundamental elements of soccer like touches, runs, passes, player location, time, and goalie actions.  Further data can describe a pass (square, cross, forward lob) or a defensive touch (tackle, deflection, header) or similar actions. 

Using the elements of game action, derivative statistics can also be developed that convey more complex concepts in soccer.  For instance, upon merely counting the types of passes and whether they are successful, one can report the number and percentage of completed passes, or compare square passes and advancing passes.  Additionally, one can track linking “creative” passes, or advancing passes that allows the recipient to complete their pass (or score).  Thus, this creativity can be defined as putting the recipient in a position to advance the possession.  A measure such as “creative passes” allows a person to quickly understand the context of a game, and how a player impacted the result.  

And these measurements are just the initial, first-tier types of measurements.  With further analysis of a season’s worth of data, one can determine which game elements affect the game outcome, and by how much.  Additionally, this same analysis can be used to determine which game elements affect player salaries, and by how much.  Thus, by comparing the game data with the salary data, then one can identify inefficiencies in the labor market.[2]  Ultimately, a savvy manager would use the data to exploit the labor market.

Most businesses evaluate the performance of its employees with metrics.  A company may measure production rate, sales, or even customer service.  As a business, Major League Soccer is no different in this regard.  MLS can determine the value of its players through scouting players, film study, or evaluating some of the few objective statistics available, all labor-intensive methods.  Some MLS employees – the players – perform their job in a very public way, and success is easier to understand.  Consequently, MLS can identify the best players available (and evaluate their relative value) on the field, perhaps even determine which position on the field leads to the greatest success for a player by using performance metrics. 

As a pub mate recently explained to me, some are able to evaluate the players just by watching the match and feel that objective data is unnecessary.  And it’s certainly true that expert observers can meaningfully evaluate a game and its participants.  However, it takes staff time to watch every single game, which becomes a daunting task if one considers evaluating talent in numerous leagues overseas.  Using metrics can allow the same staff to cast a wider net.  And even in the video presence of a stellar 90 minutes, one cannot tell if the game was an aberration, or a representative display.  And for players with some weaknesses – which makes up almost all players not named Messi – how do we decide the relative value of strengths and weaknesses?  How will results in a different league translate to a current one?  While trained experts can provide answers, in the current environment it becomes increasingly difficult to offer more than mere speculation.

In deference to my drinking buddy, advanced metrics will not replace professional evaluation, made through training time or scouting, but should supplement professional evaluation with objective, relevant information.  Of course, no method of predicting future value is perfect – indeed, an injury can derail the best laid plans – but adding information can make the evaluation process incrementally better.  And with the size of contracts increasing, and the total number of evaluations made across the league, an increase is evaluation efficiency can yield strong dividends.

As nothing in life occurs without cost, advanced performance metrics need to be compared to other talent evaluation methods, such as extra scouting, team trials and additional film study.  Compared to the costs associated with other talent evaluation methods, the return on investment for advanced metrics is favorable.  For example, MLS had more than 70 contracts over $150,000 in 2010.  A certain percentage of these contracts will be “busts”; or players who do not perform to their expected contribution level.  Advanced performance metrics may pay for itself – independent of any marketing benefits – if the costs of advanced metrics are less than the reduction in “busts.”  In other words, the league may come out ahead if it spent $200,000 for its advanced metrics, and avoided 2-3 busts with top-heavy contracts.  Additional revenue may also be generated by identifying undervalued talent which can lead to a profit in the transfer market.  Further, more specific, objective information also can support the value of MLS players on the transfer market.

Fully incorporating data analysis into the labor management function, using data provided by Opta or within each team’s scouting apparatus, can lead to many benefits for MLS.  As early adopters, MLS will be in a better position to determine the value associated with a performance measure and exploit inefficiencies in the market.  Also, as a league with comprehensive statistics, MLS players will have the potential for higher demand as more information leads to more accurate predictions of performance – and serves as an equalizer for players that get less media attention.  And while in-depth analysis, repeated again and again, can indentify higher performing players, the development of objective performance data can only benefit decision-makers as they evaluate talent.  Further, because the metrics are objective and consistently applied, the empirical value to the MLS can be demonstrated and lead to even better performance metrics.  Investing in the quantitative analysis of advanced in performance metrics is in the league’s interests.  Ultimately, standardized, meaningful measurements of performance can lead to an improved product at a better value. 


[2] The same process was followed by the Oakland A’s, described in Moneyball, where the A’s management determined that on-base percentage (and walks) were important to wins but were undervalued in the market. 


7 Responses

  1. Hi,

    Nice piece and thanks for the Opta plug.

    One quick request – could you use our current logo? I hate that old one….

    Simon Banoub
    Director of Marketing, Opta.

  2. Simon:

    Happy to do it. If you can shoot me an e-mail: with the current logo, I will change them out.


  3. Are we not talking about the Capello Index?

  4. The Capello Index, and the MLS version – the Castrol Index, are publically available lists that assign a single number to a player that represents their overall ability or value. The explanations for the index assert that the number is an objective measure that values the most important events of the game and combines it all into a single number that ultimately relates to the final outcome. As an example, Rooney might be given a 94 while Nani might get an 87. And the interpretation would be that Rooney was more valuable to his team, contributed more to winning, than Nani.

    While I applaud efforts to quantify performance, these scales have limited utility. First, the specific data inputs, and the formula for combining them are not disclosed. Thus, one cannot verify that the index considers what a manager/team would consider relevant information (remember the stats should support expert opinion by objectively reporting items that experts would care about). For one specific example, I would value defenders that block shots in the penalty area – as it reduces the need for a keeper to make a save. I am unsure if this is considered in the Index, and if it is, what value is placed on it.

    And considering my current job as an attorney for a corporation that provides information to professionals, I don’t fault the Index for not disclosing its algorithm. However, there are statistical tests for the utility of the information. And the index should be statistically valid, i.e., it should measure what it is intended to measure. For example, the Index could report that its methodology explains 70% of game outcomes, which would means its measure is very highly correlated to wins, and that it’s player ratings are +/- 2.5 points. Accordingly, the Index can report its validity and demonstrate its utility without disclosing proprietary information. And that has not been done.

    Although I will refer to an index for conversation, my main issue with the Index is that it loses too much information in the process. Few players are top notch in every skill set, and the index does not share a player’s strengths and weaknesses. I would like to know how a defender fits into an offense, as well as the number of tackles and deflections. I want to know the success rate for crosses more than square passes. I would personally report whether Player A’s pass led to Player B making a successful pass (or shot attempt) as a measure of Player A’s ability to deliver a meaningful ball. The index sheds no light here.

    In short, using meaningful data, whether provided by Opta or Stats Inc. or other data providers, needs to go beyond what is reported in an Index in order to be useful for identifying talent.

  5. I appreciate the level of detail you have gone into to answering of my somewhat facetious question. As valuable as this measure would be, I could see that it would be difficult to put into use. How many statisticians would be needed at each game where you are tracking players.

    You stated “Specifically, advanced performance metrics can adds accurate, objective information that can supplement talent evaluation, increase cost efficiencies and support pricing for transfer sales”

    When tracking dozens if not hundreds of players, as wonderful as this system would be, the costs would be prohibited. Would you be evaluating everyone on the field or players in certain positions. If you are tracking certain players then you might miss the proverbial diamond in the rough.

    As much as I believe there are certain negatives with this, it a very interesting proposal. Too often, coaches and/or scouts are unduly influenced by a spectacular play for flub by an individual and that is what is remembered. Seldom is the that player that consistently makes the run off the ball to open space up for others ever looked at or that player that consistently does the small things that set up others for success on the field.

  6. Cost and feasibility are, of course, important concerns. But cost should be considered in relation to savings from avoiding bad contracts – especially DP contracts – and the value gained from being able to market players and achieve higher transfer fees. And for the sake of disclosure, when I originally drafted these arguments, I combined the value gained from using stats in marketing activities as well as labor management.

    In my opinion, the basic statistics should be procured at the league level; and leave it up to teams to determine the value of certain measures and how to use them. MLS, a single entity anyway, gets its data from Opta. And my understanding is that other US leagues centrally organize their statistics.

    Most importantly, the effort can be as large or small as resources allow. I have scored games by hand, just by watching a TV broadcast. It took a bit of effort to ID players; quick, tight camera angles make for good TV, but make it tough to tell who is contesting a header. Conservatively, it took me about 8 hours to score 90 minutes, and another hour to translate it to a spreadsheet. Thus, a week’s worth of MLS games (9) may take 81 man-hours, and some companies pay part-time scorers $13 an hour. Thus, the actual scoring and entering of the data can be done, with pen and paper, for a little over $1000 a week with some college students and a little time. And with some experience, and a wider camera view, the time and costs may come down from that. The result was data on passes by type (aggressive v. square, air v. ground delivery), whether the pass recipient did anything with the ball, possessions, tackles, deflections, headers, free kicks, and the usual stuff such as shots and saves.

    The main stats companies provide the stats in near real-time (which adds to the cost) by using technology that recognizes the location of all of the players on the pitch and records the movement. This system provides some complex data that my archaic method does not, such as a heat map and distance covered, but greater data permits greater flexibility for investigators. And with the larger data providers, one can get data from other leagues as well. Consequently, one can tailor their assessment process to their resources and time needs.

    And I agree that scouts can frequently give too much value to an infrequent play (stellar or horrendous). In other sports, players who “look the part” are generally paid more than those that don’t, in spite of performance, another Moneyball conclusion. (I think Ronaldo gets paid more than Messi, and that’s just wrong.) Similarly, in soccer, some nationalities command a higher premium than others for relatively equal performances. Better stats would help cut through that clutter, and could even allow a scout to justify a riskier pick with higher upside.

  7. You understand therefore significantly with regards to this topic, produced me for my part consider it from a lot of varied angles. Its like men and women aren’t fascinated unless it’s something to accomplish with Girl gaga! Your personal stuffs great. All the time take care of it up!

Comments are closed.