Park Effect: Measurement and Impact on Team and Player Performance
Purpose of this article
This article answers the following questions:
- What is the Park Effect?
- Is it important?
- How is it measured?
- Do Park Factors affect team performance?
- How do player batting, pitching and overall performance ratings change when Park Effects are included?
What is the Park Effect?
In Major League Baseball each team plays its home games in a unique stadium/park. These home field venues differ in many respects including: the dimensions of the playing field, height of fences, symmetry, roof/no roof & sun position, altitude, turf or synthetic surface, infield properties, prevailing wind direction, separation of the seats from the playing surface, size and location of in-play foul ball areas. Compared to the uniformity of fields, courts and rinks in other sports, these differences add to the richness, interest and variety of game events in baseball. A good description of various aspects of park factors can be found in Fangraphs Park Factors.
Is the Park Effect Significant?
The attributes of a park can have a significant effect on game outcomes. Consider the following examples: a home run in one park may be a long fly ball out in another park; sun, shadows, altitude, lighting and wind factors may contribute to fielding errors; ground rule doubles may be more common in one park than another; pop flys may be in the stands or caught depending on the venue, successful bunts may be problematic; knowledge of park characteristics can affect the success of outfielders; roster strategies may be influenced by the park. Park characteristics can definitely affect game results and individual player performances.
When comparing team and player performances, one would like to remove biases due to game locations. In this article, team and player performances are compared — with park effect and no park effect — using the RunPlusMinus methodology for measuring on-field performance.
How Is Park Effect Measured?
In theory, a park-independent measure of player performance would require simulating replays of every game in a park-factor-neutral park. Of course, this is infeasible because many game-time decisions are park-dependent and therefore not transferable to a different venue. There are many lists of “park factor” values that typically use a value of 1 for the theoretically-average park. The value is greater than 1 for “batter-friendly” parks and less than 1 for “pitcher-friendly” parks. When calculating a park factor the number of games played by each team in each park is an important input. This data is necessary because teams play half their games in their home park. All published methods include the number of home runs and runs scored/allowed when calculating park factors.
Different proposals have been made for measuring park effects but there is no one method that is accepted by everyone as being the best. Baseball Reference provides a formula for calculating park factors. Reddit users have provided tables of factors that differ for batters and pitchers. The formula used by Fangraphs is described in Fangraphs Park Factor Calculation. RunPlusMinus™ uses park factors published by ESPN at ESPN Park Factors. The ESPN values are obtained using the formula:
Park Factor = ((homeRS + homeRA)/(homeG)) / ((roadRS + roadRA)/(roadG))
‘RS’ = runs scored, ‘RA’ = runs allowed, ‘G’ = games
The formula states that the park factor is the ratio of total runs in home games to total runs in road games. In 2018, the highest (year-to-date) factor is 1.411 for Colorado’s Coors Field and the lowest is .731 for the Oakland Coliseum meaning that Coors Field is the most batter-friendly park and the Oakland Coliseum is the least batter-friendly park. The table of the ESPN Park Factor values used by RunPlusMinus appears later in this article.
Ideally, methods of calculating player performance that depend on Expected Runs as the basis for performance measurement should use a table of Expected Runs values for each park. Although such tables can be calculated, the sparseness of the state-to-state transition counts would mean significant statistical variance is associated with the point estimates.
How Do Park Factors Affect Team and Player Performance?
Because MLB schedules are unbalanced (opponent counts are different for each team), the number of runs scored/allowed is affected by the park factors associated with each team’s games. This makes it difficult to answer the question “How would each team have done if all games had been played in the same (hypothetical) park?”. Proof that park factors do influence team performance vis-à-vis runs scored/allowed is found in the next section of this article.
Park factor influence on player performance measurement is a different story. Baseball, being a zero-sum game, has created many different offensive and defensive statistics to measure player performance. How do park factors modify a pitcher’s, ERA, Save percentage, opponent Batting Average, etc. when comparing two pitchers? How should a batter’s OPS, RBIs, Home Run count, etc. incorporate park factors in order to provide park-independent comparisons of batters? Even if a generally accepted method existed for offense and defense, there would remain the need for combining offense and defense values to obtain an overall player performance measurement. The RunPlusMinus Rating statistic is a composite measure of player performance that includes park factor values when calculating the batting and pitching components of player performance. Examples are provided in the following section.
The RunPlusMinus Solution
RunPlusMinus™ provides a single statistic that measures the on-field performance of players and teams. Information, FAQs and references to detailed descriptions of RunPlusMinus methods, applications and results are found at RunPlusMinus. Fundamental to the value of RunPlusMinus analysis are: 1) The inputs satisfy the 5 ‘CRAZI’ axiomatic criteria necessary for a composite statistic, and 2) The 6 challenges required to incorporate fairness are met. There are important applications of RunPlusMinus in addition to measuring individual player performance.
Park factors are incorporated into the calculation of player ratings by multiplying the state-to-state Transition values by the park factor. Transition values involving events such as a strikeout, caught stealing and catcher interference are assumed to be independent of the park factor. On the other hand a home run, or ground rule double, has more or less “runs value” depending on the home team’s park. Since Transitions are the atomic unit of progress in a game and player ratings are fractions of Transition values, player ratings are influenced by park factors.
Park Factors
The chart below shows park factor values published at ESPN park factors These values are used in the subsequent calculations showing the team and player effect of park factors.
The chart shows that Colorado is the most batter-friendly park and that the Oakland Coliseum is the park in which it is most difficult to score.
Proof of Park Factor Significance
The RunPlusMinus (RPM) statistic is based on the values of state-to-state Transitions in a game. There are 24 possible states ranging from 0 out with bases empty to 2 out with bases full. The value of a transition = (runs scored + the change in expected runs in the remainder of the half inning). Expected Runs values are published by ESPN and other sources. Since you can’t undo an out or “run backwards” only a subset of state-to-state Transitions are feasible.
How do park factors affect transition values? Simply put, batter-friendly parks will average more total runs scored/allowed per game. Because Runs Scored is a term in the Transition formula above, one would expect that the magnitude of Transition values would be larger in parks with higher park factors. This is indeed the case as shown in the pair of charts below. Chart data is based on all 2018 MLB games through June 30, 2018. The “RPM Magnitude” values are the sums of the absolute values of offense Transition RPMs. Only offense Transition values are needed since defense Transition values are equal and opposite in every play.
The sum of the absolute Transition RPM values for Colorado home games decreases by 163 when RPM calculations use park factors. Note also the effect is small for parks such as the Blue Jays Rogers Center that have factor values close to 1. This gives credibility to both the ESPN values and the RunPlusMinus methodology. The following chart shows the results for pitcher-friendly parks.
The data shows that pitchers benefit the most when playing in the Oakland Coliseum.
How Park Factors Affect Player Performance
The three pairs of charts which follow show the changes in players’ batting, pitching, and overall performance when calculated with and without park factors. These results are obtainable because RunPlusMinus calculations include a park effect parameter which can be assigned a value between 0 (exclude park effect) and 1 (full park effect).
Batting Performance
The RunPlusMinus RPM statistic claims to be a fair measure of on-field player performance for each of the four components. For games played in batter-friendly parks, RPM batting values are higher because the RPM statistic is run-based. Players having at least 200 plate appearances in games up to June 30, 2018 were analyzed.
All players above with the exception of Freddy Galvis play on teams that have home fields that are not batter-friendly. The converse is true for players in the chart below — with one exception, the batters’ home fields are batter friendly. Their batting performances would decrease if playing in a “neutral” park.
Pitching Performance
The charts below supports the claim that RPM values for pitchers are affected by the home-team park. Pitchers with at least 200 batters faced were included in the analysis.
Chad Bettis has the largest improved pitcher ranking when park factors are used. Note that four Colorado pitchers would benefit the most if playing in batter-friendly parks.
All of the pitchers in the chart above play in home field pitcher-friendly parks and would have their performance ratings decreased by the amounts shown if playing in parks with a park factor of 1.
Total Player Performance
RunPlusMinus was developed to measure total player performance using a single index of performance. Although each of the four components (batting, running, pitching and fielding) can be measured using the RPM statistic, they cannot be simply added as explained in the article The Best Baseball Statistic. The RunPlusMinus Rating statistic combines the four component RPM values to remove biases that would otherwise skew the results in favor of one or more of the components. As with the component RPM values, the Rating values are additive and the average performance Rating is zero.
The two charts below show players with the largest benefits and the greatest decreases in total performance when park factors are used in the calculations. The players include pitchers with at least 200 batters faced and/or batters with at least 200 plate appearances.
Because what’s good for pitchers is bad for batters and vice-versa, there is no concentration of teams in this chart or in the “biggest decrease” chart below. The “Rank” values in the “No Park Factor” and “With Park Factor” are based on all players participating in 2018 MLB season year-to-date — not just the batters and pitchers satisfying the 200 minimum participation criteria. Consider Chad Bettis: he has the most to gain in measuring overall performance using park factors with an increase of 27.5 Rating units. On the other hand, the chart below shows that Trevor Story who plays in the batter-friendly park of the Rockies drops 76.5 rating points when park factors are used.
The data presented previously provides numeric values showing the park effect on team and player performances. In all cases, the high correlations between park factor values and changes in performance measurements support the use of RunPlusMinus as a tool for measuring the impact of park factors.
Conclusions
The main ideas in this article are:
- Park Factors should be included when measuring offense and defense player performance
- Different formulas exist for calculating Park Factors
- RunPlusMinus analysis can be used to calculate the effect of Park Factors on team and player ratings
A Conjecture. Of great interest to the author of this report is the possibility of reverse-engineering the values of park factors beyond the methods that are used today. That is, given a widely-accepted measure of component and composite player performances, it is possible to calculate park factors based on comprehensive performance data — not a limited subset of data that is currently used. That is rather than modifying performance data by park factors, one would derive park factors based on player performance data. Because of season-to-season changes in player competencies, rule changes and park replacements/modifications, park factors will change over time.
While You Are Here…
If you want to be reminded whenever we release new content, please subscribe to our mailing list to be kept up to date!
If you have any questions, comments, requests or complaints, please feel free to add them in the comments below or to email us at info@runplusminus.com
You can learn more about the RunPlusMinus™️ statistic at RunPlusMinus.com
