Soccer Power Index explained
ESPN's Soccer Power Index (SPI) is an international and club rating system designed to be the best possible representation of a team's current overall skill level. As opposed to other systems like the FIFA rankings, SPI is forward-looking and predictive. It is designed to project which teams will have the most success going forward, not rate the success of teams in the past - though the two are obviously related.
SPI differs from other rating systems in a number of ways.
• For international matches it weights each match based on the lineup each team is playing. This is often a problem for international ranking systems, as most teams do not put out their best lineup in every match. SPI recognizes that and weights the importance of each match based on how seriously both teams are taking the match
• Mexico is a great example of how this works. Coming out of the 2011 CONCACAF Gold Cup Mexico was 9th in the FIFA rankings and 10th in SPI. Then Mexico proceeded to lose all three matches in Copa America, which dropped its FIFA ranking to 20th. However, Mexico brought nothing more than a glorified U-23 squad to that tournament, which did not include any of its top players that had just competed in the Gold Cup. Undoubtedly the world learned very little about how Mexico would perform in matches it actually cares about based on that performance. SPI detected Mexico's lack of first teamers and de-weighted the matches such that Mexico did not move in the rankings at all.
• There is a time effect such that a match played yesterday is weighted more toward a team's rating than a match played two days ago, which is weighted higher than a match played three days ago, and so on. For clubs the cut off for the "assessment period" is four years. For international matches the cut off depends on how much data we have for the particular team. For countries that play a relatively large number of matches per year (say, Spain), the cut off will usually be somewhere between 4-5 years. For countries that almost never play (Belize, for example) the assessment period could go back six years or more.
• It weights the result of each match based on the quality of the opponent. Clearly, beating San Marino 1-0 is not the same as beating Germany 1-0. SPI identifies this and gives each team the appropriate amount of credit (or blame) based on how the team performed relative to how well it should have performed given its opponent.
• Home-field advantage is also factored into this "expected" performance for each team. France beating England is more likely to happen in Paris than in London. So France would be rewarded more heavily for an away win, and conversely, penalized more harshly for a loss at home. Based on historical evidence, home pitch advantage is slightly higher in international matches compared to major club competition. This is taken into account as well.
• It uses club play as a small indicator for international performance. SPI very carefully looks at how players in the top European leagues are performing and adjusts those players' international teams accordingly. For example, if Robin van Persie goes on a scoring spree for Manchester United in the Premier League we would expect him to at least somewhat continue that form for the Netherlands. Strong play for a club incrementally increases the rating for that player's country. Conversely, poor play for a club could adversely affect country ratings as well. Keep in mind that this is a small part of the rating. The vast majority of a country's rating comes from its performance on the field as a team. And for countries that have no players playing in the top European leagues, all of their rating comes from their international matches.
• It is based on goals scored and allowed, not wins, losses and draws. This goes back to SPI's definition of being a forward-looking system. Since the 1998-99 season, Premier League teams that enter a match with a better goal differential in league play, but fewer league points, have a record of 179 wins, 138 losses and 130 draws. Similar trends show up in other leagues like La Liga and Serie A as well as other team sports across the world. Because SPI is trying to predict future events, it uses the stat that correlates better future success - scoring margin. Other systems like the FIFA rankings and the ELO ratings are solely based on wins and losses, which is why SPI is better at projecting future success than these two methods.
The outputs of SPI are offensive and defensive ratings for all 217 international countries and every club that participates in the Barclays Premier League, La Liga, Serie A, Bundesliga, Ligue 1 and the Champions League. The offensive (or defensive) ratings are interpreted as the number of goals a team would be expected to score (or allow) against a league-average team at a neutral site. For example, if Manchester United had an offensive rating of 2.0, it would be expected to score 2 goals against a league average defense (think 10th-place team in the Premier League table). And if Chelsea had a defensive rating of 1.0 it would be expected to allow one goal against a league average offense.