Glossary

Player Measures 

 

A90: A90 is a player’s assist rate given as assists-per-90 minutes of playing time. (see Assists definition below).

Age: (see effective playing age below)

Age Profiles: Age Profiles are a useful tool to use when analysing players futures and can be used to get a feel for the distribution of any statistic by a player’s age.

Assists: Assists are given to the last two players to touch the ball on the attacking team when a goal is scored. The touch must be of an attacking nature and must be intended by the player. We believe that goals are scored by one person, but set up by many. A move scores a goal not just a single cross by the right winger onto the centre-forward’s head. By widening the criteria to the final two passes our “assist” gives a greater insight into who are the players most frequently involved at the ‘business end’ of goal scoring moves. 

Baseline Performance: A player’s Baseline Performance is generated in SIMprofilerTM and is used to calculate a player's forecast for the up coming season. The Baseline Performance of a player is a calculation based on the player's previous two seasons of performance. The reason for using the last two seasons of a players career is to remove luck or the error of a ‘fluke’ season where either the player’s current level of performance has been artificially too high or too low from good fortune or injury for example

CS: CS is the number of part and whole clean sheets a player has been a part of.

CS%: CS% is the percentage of games a player was present and the team kept a clean sheet.

Effective Playing Age: Effective Playing Age is the player’s age as of 1st January of that year for the entire season and forms part of a players profile when assessing past and present performance as well as playing a vital role in building a player’s forecast in SIMprofilerTM. 

Fixture Strength: Fixture Strengthis a statistic that gives the strength of opposition a player or team has faced throughout the season. 

FSE goals: FSE goals stands for Full Season Equivalent goals and is a measure that gives the total number of goals a player would score given his rate and if he played every minute of every game during the season.

G90: G90 is a player’s goal scoring rate given as goals-per-90 minutes of playing time.

GAA: Goals-against-average. This metric gives the team’s goal-against-average when the player was present in the line-up.

GFA: Goals-for-average. This metric gives the team’s goal-for-average when the player was present in the line-up.

Goals: Number of goals scored by the player

GS: The number of games started by a player

aGVA: Attacking goal value above-average is a single measure that is an estimate of a player’s total value in an attacking sense (measured in goals added) above that of an average performer that plays in the same position.

aGV: Attacking goal value is the precursor for aGVA but differs in that the players attacking performance is not compared to that of an average performer that plays in the same position.

dGVA: Defensive goal value above-average is a single measure that is an estimate of a total value from a defensive perspective (measured in goals prevented) above that of an average performer that plays in the same position.

dGV: Defensive goal value is the precursor for dGV but differs in that the player’s defensive performance is not compared to that of an average performer that plays in the same position.

tGVA: Total goal value above average is simply an addition of attacking goal value above average and defensive goal value above average. It is an individual goal difference, if you like, that measures the total goal value that a player adds or takes away from his team given his attacking production and defensive prevention compared with the average player at his position. 

tGV: Total goal value is the precursor for tGV but differs in that the players attacking and defensive performances are not compared to that of an average performer that plays in the same position.

GW goals: Game winning goals. This is a count of the number of goals a player scores that ended up being the deciding goal of the contest.

Height: Players height given in metres and centimetres.

M90: Mistakes-per-90 minutes, or its full version of mistakes that lead to goals per-90 minutes, is a player’s mistake rate. (see mistake definition below)

Major attrition: Major Attrition is the percentage chance of a 50% reduction relative to a players playing time baseline. The statistic is not only a good indicator of the risk of injury for a player in the upcoming season, but also capture playing time decreases due to poor performance and managerial decisions.

Minor attrition: Minor Attrition is the percentage chance of a 25% reduction relative to a players playing time baseline. The statistic is not only a good indicator of the risk of injury for a player in the upcoming season, but also capture playing time decreases due to poor performance and managerial decisions.

MINS: Minutes played.

Mistake: The Mistake or mistake leading to a goal is a count on the number of times a given player made a defensive error that lead to the opposition scoring a goal.

P90: P90 is a player’s attacking points rate given as points-per-90 minutes of playing time. (see points definition below)

P90 probability distribution: A players Points-per-90 minutes probability distribution is an estimated range for the player’s predicted performance in attacking points for the upcoming season generated from SIM profilerTM. The distribution gives the players expected performance in P90 at various level of probability based on the player’s baseline performance and set of most ‘similar’ players. For example, if a player’s 65th percentile prediction is a Points-per-90 minutes rate of 0.59, this indicates that he has a 65% chance of finishing the season with a P90 rate of 0.59 or lower, and a 35% chance of ending the season with a points-per-90 minutes rate higher than 0.59.

The best use of this distribution is to look at the percentage chances a player will be above-average quality playing in his primary playing position. For example, for the upcoming season Didier Drogba has a 70% chance of finishing the season with an above-average P90 rate as a Forward, whereas Freddie Ljungberg has a 10% chance he’ll finish the season with an above average P90 rate as an attacking midfielder.

Player Forecast: The standard Player Forecast (and the one given in the book) is the mean projection from a player’s baseline performance and set of similar players that incorporates all the potential values into one single forecast line. However, due to how the model is constructed using a set of ‘highly’ similar players we are able to produce a forecast distribution that gives a prediction of the player’s performance at various levels of probability. 

The model produces playing time and attacking rate statistics at percentiles ranging from the 10th to the 90th to give an indication as to the possible variability in a player’s performance. These can be useful in determining the chances of either outstanding or abysmal performances in the up coming season as well as the production levels one might have more confidence in occurring.

This output obviously has its benefits over a rigid single forecast line and gives a better feel for a player’s maximum and minimum level of performance that could realistically be expected. It can be useful in determining which players even if their performance collapses will still be an above-average performer and which players even if they serious improve will still be no better than average standard.

Player Profiles: Player Profiles are similar to age profiles, but are based on only one player. They detail the given players age profile for a chosen statistic and give a visual representation of a players performance throughout their career.

Player Tables: Player Tables display a player’s miscellaneous, usage, attacking, defensive, rate and team statistics for the last two years as well as a forecast for the upcoming season.

Points: Attacking points is another ‘count’ statistic, and is a measure of the overall attacking production of the player: it is a simple addition of the player’s goals and assists, combined to give a useful indication of his overall attacking effectiveness.

PPG: Points-per-game. This metric gives the team’s average points won when the player was present in the line-up.

aPVA:Attacking point value above average is the number of league points a player contributed, above what an average player would have done at the same position, given his attacking performance. Like aGVA this statistic is adjusted for the level of competition faced.

dPVA:Defensive point value above average is the number of league points a player contributed, above what an average player would have done at the same position, given his defensive performance. Like dGVA this statistic is adjusted for the level of competition faced.

tPVA:Total point value above average is the number of league points a player contributed, above what an average player would have done at the same position, given his attacking and defending performance. Like tGVA this statistic is adjusted for the level of competition faced.

Primary Playing Position: A player’s primary playing position is the position in which he plays 50% or more of his time playing during the course of the season.

Set Piece goals: The set piece goal statistic gives the number of penalty and free-kick shot goals a player has scored.

Similar Player: Historically similar players are determined using similarity scores (see below) which is a method used to determine how statistically ‘similar’ two lines of data are. When generating a player’s forecast SIMprofilerTM only considers players that are 95% ‘similar’. 

Similarity Scores:Similarity Scores are a method used to measure a player's comparability to other players using statistics and were first introduced by Baseball Analyst Bill James. Our similarity scores use on and off field measures to get a comparability percentage to use highly similar players in forecasting a player’s future performance. In theory a score of 100 would occur when two players are exactly the same and a score of 0 represents two players who are in no way similar.

Stops: Stops are a measure for the number of saves + other goal saving actions a goalkeeper makes. Standard saves only take into consideration ‘saves’ from opponents’ shots-on-target, although there are many other times when a goalkeeper’s actions prevent a goal. For example, when a goalkeeper smothers the ball at an attacker’s feet or punches the ball clear from a corner there has been no shot-on-target, but his actions have definitely prevented a goal. This metric gives a better feel for the total command a goalkeeper has for his penalty area and how regularly he intervenes on potential goal chances - not just his shot stopping prowess.

Stop %: Stop percentage is simply the number of stops a goalkeeper makes as a percentage of the number of goal scoring chances against.

Weight: A player’s weight given in kg

Manager Measures

Age: the age given for a manager is his age as of 1st January each year.

CPG: CPG or changes-per-game is the average number of changes made by the coach to the starting line-up on a game-to-game basis.

Doubles: Doublesgives the number of times the manager has used a double substitution.

Formation: Formationis the starting formation most often employed by the coach during the season.

Median Substitution times: The Median Substitution Timeis the time that sits in the middle of the range of times for each substitution. It is often given for each separate substitution.

Record: Recordis the team’s number of wins, draws and losses from the league campaign given in the form W-D-L.

Same: Sameis the number of times the coach used the same starting line-up in back-to-back games.

SPG: SPGor substitutions-per-gamegives the average number of substitutes used by the coach in each contest.

Starter GAA: Starter GAAis the goals-against averages (per-90 minutes) for the starting team put out onto the pitch by the coach.

Starter GD: Starter GD is the raw goal difference for each team’s starting line-ups.

Starter GFA: Starter GFAis the goals-for averages (per-90 minutes) for the starting team put out onto the pitch by the coach.

Total changes: Total Changesis a count statistic of the sum of player changes made to the starting line-up from game-to-game.

Total Players: Total Players gives the number of players used by the manager during a certain time period such as 10 games, 25 games, or the whole season.

Total Subs: Total Subsgives the total number of substitutions made by the manager over a certain time period such as 10 games, 25 games, or the whole season.

Win %: Win % is the manager or coaches winning percentage calculated as the number of wins divided by the number of games in charge.

Years: Years is the total number of years the manager has been in his current role.

Team Measures

Adjusted Estimated Points: Adjusted Estimated Points are created in exactly the same way as Estimated Points (see definition below) but are adjusted for opponent and the ground they were played in. To make these adjustments the point estimates for each game are adjusted by the overall estimated point quality of the opposition, in a home or away situation, depending on the game being adjusted. This ensures that all estimates are adjusted to be against league average opposition and in a neutral venue regardless of whether the game was played home or away and whether it was against, for example Chelsea or Watford. The reason we make these adjustments is because it is far easier to play better and produce a higher point estimate versus a lesser team at home, than it is to go to Stamford Bridge, and play Chelsea off the park. If left unadjusted by opponent and situation some point estimates would be artificially high and some would be artificially low. These adjustments allow us to assess a team’s overall performance game-by-game throughout the season, literally, on a level playing field.

Attacking Adjusted Estimated Points: Attacking Adjusted Estimated Pointsare similar to the overall adjusted estimated points, however, the initial estimates are based on attacking metrics alone (see attacking estimated points definition below) and the adjustments made are for venue and the defensive estimated points quality of the opposition. The reason for using the defensive estimated points for this adjustment and not the overall estimated points figures is because a team might have a high overall quality, but be unevenly formed by an outstanding attack and a poor defence, or vice versa. As we are assessing attacking quality here, we adjust the attacking estimated point scores by the defensive quality of the opposition.

Attacking Consistency: The Attacking Consistency metric is the percentage of games during the season that the team in question performed above average according to our attacking adjusted estimated points model and enables us to see which teams were more consistently playing above-average attacking football during the whole season.

Attacking Corner %: Attacking Corner % is the number of corners the team in question has scored from as a percentage of the number of corners taken by that team in league play. Different teams win and concede different numbers of corners so just giving the number of goals scored from corners, although useful, can be misleading.

Attacking Efficiency: Attacking Efficiency is a metric that fundamentally measures how effectively each team uses its possession to score goals. The measure gives the number of goals a team would score per-game had it had 100% possession. This measure essentially levels the time of possession differences between teams and gives a goal-per-game value per 100% possession.

Attacking Estimated Points: The theory and methodology behind the Attacking Estimated Pointsmetricis exactly the same as the Estimated Points measure (see definition below), but only attacking team stats such as goals-for, shots-for and other highly correlated metrics to goal scoring were used in the modelling process to predict points. The figure given is the estimated number of points a team should have collected based solely on their attacking team statistics and ignoring the effect of defensive and overall metrics. This enables us to determine the best and worst attacking teams in the league taking more than just goals scored into account. The model assesses, through attacking team statistics, whether the team did the things necessary to generate scoring chances, score goals and essentially win football matches.

Attacking Free-Kick Cross: Attacking Free-Kick Crossis the raw number of goals a team has scored from free-kick crosses during the season.

Attacking Free-Kick Shot: Attacking Free-Kick Shotgives the raw number of goals the team in question has scored from free-kick shot attempts during the season.

Attacking Total Set Pieces: Attacking Total Set Pieces includes partially and substantially cleared free-kick crosses and corners, long attacking throws, short attacking throws, free-kick shot rebounds, follow-ups to penalty misses and all the qualifying goals in the Corners % and Free-Kick Cross and Free-Kick Shot categories. This figure gives a good indication as to how well the team in question has used all its set pieces to generate goals.

Clean sheets: Clean Sheetsgives a count of the number of games the team in question prevented their opponents from scoring a single goal.

Consistency: The Consistency metric is the percentage of games during the season that the team in question performed above average according to our adjusted estimated points model and enables us to see which teams were more consistently playing above-average football during the whole season.

Defensive Adjusted Estimated Points: Defensive Adjusted Estimated Pointsare similar to the overall adjusted estimated points, however, the initial estimates are based on defensive metrics alone (see defensive estimated points definition below) and the adjustments made are for venue and the attacking estimated points quality of the opposition. The reason for using the attacking estimated points for this adjustment and not the overall estimated points figures is because a team might have a high overall quality, but be unevenly formed by an outstanding defence and a poor attack, or vice versa. As we are assessing defensive quality here, we adjust the defensive estimated point scores by the attacking quality of the opposition.

Defensive Consistency: The Defensive Consistency metric is the percentage of games during the season that the team in question performed above average according to our defensive adjusted estimated points model and enables us to see which teams were more consistently playing above-average attacking football during the whole season.

Defensive Corner %: Defensive Corner % is the number of corners the team in question has conceded from as a percentage of the number of corners that team has faced in league play. Different teams win and concede different numbers of corners so just giving the number of goals conceded from corners, although useful, can be misleading.

Defensive Efficiency: Defensive Efficiency is a metric that fundamentally measures how effectively each team defends when not in possession. The measure gives the number of goals a team would concede per-game had it given up 100% possession. This measure essentially levels the time of possession differences between teams and gives a goal-per-game value per 100% possession against.

Defensive Estimated Points: The theory and methodology behind the Defensive Estimated Pointsmetricare exactly the same as the Estimated Points measure (see definition below), but only defensive team stats such as goals-against, shots-against and other highly correlated metrics to goal conceding were used in the modelling process to predict points. The figure given is the estimated number of points a team should have collected based solely on their defensive team statistics and ignoring the effect of attacking and overall metrics. This enables us to determine the best and worst defences in the league taking more than just goals allowed into account. The model assesses, through defensive team statistics, whether the team did the things necessary to prevent scoring chances, prevent goals and essentially win football matches.

Defensive Free-Kick Cross: Defensive Free-Kick Crossis the raw number of goals a team has conceded from free-kick crosses during the season.

Defensive Free-Kick Shot: Defensive Free-Kick Shotgives the raw number of goals the team in question has conceded from free-kick shot attempts during the season.

Defensive Total Set Pieces: Defensive Total Set Pieces includes partially and substantially cleared free-kick crosses and corners, long attacking throws, short attacking throws, free-kick shot rebounds, follow-ups to penalty misses and all the qualifying goals in the Corners % and Free-Kick Cross and Free-Kick Shot categories. This figure gives a good indication as to how well the team in question has defended all set pieces it has faced during the season.

Drawn: Drawn is the number of games drawn by the team during the season

Estimated Points: Using a statistical technique called regression analysis we developed a model to estimate the value of points a team can be expected to win based on its statistical relationship with a number of team statistics. The Estimated Points metric, therefore,gives an estimate of the number of points a team should have collected based solely on their team statistics. This enables us to determine the best and worst teams in the league removing luck from the equation as well as being able to rate a team’s performance in an individual game better than the result alone. The model assesses, through team statistics, whether the team did the things necessary to win football matches and gives each team an estimated points figure based solely on how the team performed.

Failure to score: Failure to scoreis a count of the number of games the team in question failed to score a single goal.

GAA: GAAgives the team’s goals-against average which is the average number of goals the team concedes per-game during the season. 

GFA: GFA gives the team’s goals-for average which is the average number of goals the team scores per-game during the season.

Lost: Lost isthe number of games lost by the team during the season.

Points: Points is the number of points won by the team during the season.

Shots-against Efficiency: Shots-against Efficiency estimates the number of shots-on-target per-game a team would have given up had its opposition had 100% possession. This metric allows us to see which teams are better at preventing scoring chances whilst not in possession of the ball

Shots-for Efficiency: Shots-for Efficiency estimates the number of shots-on-target per-game a team would have created had it had 100% possession. This metric allows us to see which teams are better at generating scoring chances when in possession of the ball.

Set Piece classification: Judging what is considered a goal from a corner, free-kick cross or free-kick shot can be tricky, so we setup some rules to ensure comparability and remove bias. For corners, Free-kick crosses and Free-kick shots, possession must be held by the attacking team and the ball can not have been substantially cleared. A corner or free-kick cross that is flicked on by the defending team counts, but one that is headed out of the area by the defending team, recycled for another cross that results in a goal is not included (These are included in the Total Set Piecefigures). The goal must be scored within 4-5 touches by the attacking team. The only time that more touches are permitted is if all touches are of an attacking nature. The ball could be worked for a cross by three or more touches, flicked on and finished for example. Therefore, short corners that result in a goal from a cross or direct shot are included in the corner figure but if possession is lost and regained after a few touches this will not be included. For free-kick shots the ball may be worked for a better shot or scored from a direct attempt.

Won: Won is the number of games won by the team during the season.

Miscellaneous

Correlation Coefficients: Correlation coefficients are a statistical method to measure how closely associated two statistics are, on a scale of 1 to -1. A correlation approaching 1 means the two variables are very closely positively related. That means that an increase in one means there is an increase in the other. A correlation approaching -1 means the two variables are closely related but in opposition. An increase in one is associated with a decrease in the other. If the correlation between two variable approaches zero, however, there is no statistical relationship what-so-ever, and a change in one variable tells us nothing about a change in the other, and vice versa.

Counting Statistic: A counting statistic is a count of the number of times a player achieved a certain event. For example, total number of goals or total number of assists.

Rate Statistic: A rate statistic is the rate at which a player achieved a certain event. For example, goals-per-90 minutes or assists-per-90 minutes. 

Regression Analysis: Regression analysis is a statistical technique for estimating the value of one variable (the response variable) based on another one or a series of other variables (explanatory variables). Regression analysis can be used for forecasting and estimating outcomes as well as testing scientific hypotheses about relationships between variables. 

Statistical Significance/P-value: StatisticalSignificance shows you how likely a result is due to chance. In statistics a result is called significant if it is unlikely to have occurred by chance alone.The significance of a test is also called its p-value; the smaller the p-value, the more significant the result is said to be. The most common level, used to mean something is good enough to be believed, is .95. This means that the finding has a 95% chance of being true. Often this is referenced to being a testing for significance at the 5% level meaning that the finding has a five percent (.05) chance of not being true, which is the converse of a 95% chance of being true. The 95% level is thought to have come from academic publications, where a theory usually has to have at least a 95% chance of being true to be considered news worthy. In business, however, if something has a 90% chance of being true, it can't be considered proven, but it is probably better to act as if it were true rather than false.