Skip to content
CricketTaken

Analysis

Park factors in baseball: why no two ballparks are equal

How park factors are built, what actually varies between ballparks, why a single season of the number is useless, and what the adjustment still cannot fix.

By CricketTaken EditorialPublished Analysis20 min read

How this is written and checkedReport an error

Two hitters put up the same batting line. Same home runs, same doubles, same walks, same strikeouts, same number of trips to the plate. One of them was substantially better at baseball than the other, and nothing in the line tells you which. That is the problem park factors exist to solve, and baseball is the only major sport that has it in this form, because baseball is the only major sport that never standardised the field.

Every other big team sport fixed its playing surface and moved on. A basketball court is a basketball court. A hockey rink varies by a few feet at most and the goals are where the goals are. Football pitches are permitted a range of sizes and the range is narrow enough that nobody adjusts for it. Baseball fixed the infield to the inch and then let the outfield do whatever the neighbourhood allowed, which is why a fly ball to left field is a home run in one city and an out three hundred miles away, hit at the same speed and the same angle by the same man.

What the rulebook fixes, and the enormous amount it leaves alone

Rule 2.01 of the Official Baseball Rules is precise where it wants to be. The bases are ninety feet apart. The pitcher's plate sits sixty feet six inches from the back point of home plate, and ten inches above the level of the base paths. The foul lines run from home plate through first and third and onwards. All of that is specified to a degree that would satisfy a surveyor.

Then the rule reaches the outfield and gets vague on purpose. The distance from home plate to the nearest fence, stand or other obstruction in fair territory must be two hundred and fifty feet or more. Any field built by a professional club after 1 June 1958 must provide at least three hundred and twenty-five feet down each foul line and at least four hundred feet to the centre field fence. That is the whole of it.

Read what is missing. There is no maximum distance. There is no rule at all about the height of the fence, so a wall can be a low railing or it can be tall enough that a ball off the top of it is still in play. There is no rule about the shape of the outfield between the foul lines and centre, so gaps can be square, angled, rounded or kinked. There is nothing about the amount of foul territory, which means a park can put its seats almost on the first-base line or leave enough space out there to hold a second infield. Nothing about elevation. Nothing about whether the park has a roof, and if it does, nothing about when it has to be shut.

And parks built before June 1958 were never required to comply with the post-1958 minimums, which is how the sport ended up with grandfathered geometry that no architect would design today and no club would be permitted to build now.

What the rulebook actually specifies about the field
  • 90Feet between the bases
  • 60.5Feet from the pitcher's plate to home
  • 250Minimum feet to the nearest fence in fair territory
  • 0Rules governing the height of that fence

Distances set by Rule 2.01 of the Official Baseball Rules, in feet, except the last.

That last figure is not a joke. The height of an outfield wall is one of the largest single influences on how a park plays and the rulebook has nothing whatever to say about it.

What actually varies, in rough order of how much it matters

Outfield distance, but only in combination with wall height. These two are always discussed separately and always act together. A fence three hundred and thirty feet from home plate with a wall five feet high is a home run park. The same distance with a wall thirty feet high is not, because the trajectory that clears a low fence at that range is a very common one and the trajectory that clears a high one is not. What you want to know about any wall is not its distance and not its height but the height of the ball as it arrives there, which depends on both. A park with deep fences and low walls and a park with short fences and high walls can produce similar home run rates by entirely different routes, and they will differ sharply on doubles.

Outfield shape. The gaps in left-centre and right-centre are where the distance from home plate is greatest in most parks, and they are the least standardised part of the field. An outfield that runs in a smooth arc plays very differently from one with a sharp angle in the corner, because the angle creates a place where the ball caroms unpredictably and the fielder cannot cut the throw off. Triples are almost entirely a function of outfield geometry rather than of the batter's speed, which is why triples factors are the most extreme component factors in the sport.

Foul territory. The most consistently underrated variable, and the only one that suppresses offence without touching the ball's flight over the fence. Large foul ground means more pop-ups and slices that would land in the seats are caught instead. Every one of those is an out that would otherwise have been a fresh pitch with the count unchanged. The effect works on batting average and on runs at the same time, and it is invisible in a home run factor. Small foul ground does the opposite, and it also puts the crowd close enough to the field to change how a fielder plays a ball he has to chase.

The backstop distance. How far the wall behind home plate sits from the plate itself sets how far a passed ball or a wild pitch travels before it can be retrieved. A close backstop kills the running game on balls in the dirt. A distant one turns a catcher's mistake into a base. This is a small effect and it is a real one, and it never appears in any published park factor because nobody publishes a park factor for wild pitches.

The surface. Artificial turf produces faster ground balls and truer bounces than natural grass, which lifts the rate at which hard-hit grounders get through the infield. That is a hit-type effect, not a home run effect, and it will show up in a doubles or singles factor while leaving the overall run factor closer to neutral than people expect. If you want to understand why a hitter's batting average on balls in play swings around more in some venues than others, the surface is usually part of it.

Sightlines and the batter's eye. The dark backdrop in centre field behind the pitcher exists so the batter can pick the ball up out of the pitcher's hand. Parks differ in how well that backdrop works, in how much light and movement sits behind it, and in how the shadows fall in late afternoon. A park where the sun cuts a hard line across the infield in the seventh inning is a genuinely harder place to hit in for an hour and a normal place to hit in for the rest of the game. None of this is measurable from a box score and all of it is inside the park factor.

Whether the roof is open. Retractable-roof parks are two parks. Closed, they are climate-controlled boxes with no wind. Open, they are ordinary outdoor stadiums subject to whatever the evening is doing. The decision is made by the home club under league rules, and it is made partly on comfort and partly on nobody's business but theirs. Public park factor tools now let you split by roof state for exactly this reason.

Air density, done properly

This is the part that almost every explanation gets half right, and the half it misses is the interesting half.

The force that slows a batted ball down is drag, and drag is proportional to the density of the air it moves through. Thin the air and the ball keeps more of its speed over the same distance, so it lands further away. That is the familiar half, and it is why a mile of elevation is worth so much carry.

The half that gets dropped: the Magnus force, which is what makes a spinning ball curve, is also proportional to air density. A curveball breaks because the air pushes on it. Thin the air and you weaken the push. A pitcher at altitude is throwing into an atmosphere that has stopped cooperating with him twice over, once by refusing to slow the batted ball down and once by refusing to bend the pitch he threw.

That second effect changes behaviour rather than just outcomes. A breaking ball that arrives with less break than the pitcher's body expects is a breaking ball he has to throw differently or stop throwing, and pitching staffs adjust their pitch mix at high-altitude venues in ways they do not need to adjust anywhere else. The park is not simply adding runs to a fixed process. It is changing the process, and it does so before contact rather than after it. Anyone thinking about how a pitcher's arsenal is designed has to treat altitude as a constraint on which pitches are even available.

Three things move air density, and one of them runs backwards from intuition.

Elevation. Higher means less atmosphere overhead, so lower pressure and lower density. This is the largest and most stable of the three, because a stadium's altitude does not change.

Temperature. Warmer air is less dense. A hot evening carries the ball further than a cold one at the same venue, and it also makes the ball itself livelier, because a warmer ball is slightly bouncier off the bat. Those two effects point the same way and compound. This is a real and substantial source of within-season variation in how a park plays, and it is one reason a park factor computed over a full season is an average across weather that was not uniform.

Humidity. Humid air is less dense than dry air at the same temperature and pressure, because a water molecule is lighter than the average molecule in dry air, and adding water vapour displaces heavier nitrogen and oxygen. Muggy air therefore offers slightly less resistance to a batted ball, which is the reverse of what most people assume when they say the ball does not carry on a heavy night. The effect on the air is small. The effect of humidity on the ball is not small at all, and that is a separate mechanism entirely.

A baseball absorbs moisture. A ball stored in a dry environment loses water, shrinks fractionally, gets harder and comes off the bat faster. A ball stored damp does the opposite. This is a property of the leather and the wool winding, not of the air the ball flies through, and it is the reason the humidor exists.

Humidors and the ball moved the whole league's baseline

The Colorado club used a humidor for years before anyone else, for the obvious reason that Denver's dry air was drying out its baseballs on top of everything else altitude was already doing. From 2022 all thirty clubs store their game balls in humidors, at a controlled humidity, which is a league-wide standardisation of the one part of the equipment that was previously varying by climate.

The effect is not the same everywhere, and this is what makes it worth understanding rather than just noting. In a dry city, the humidor adds moisture the balls would otherwise have lost, so it works against offence. In a genuinely humid city, the humidor can hold balls drier than the outside air would have, which works mildly the other way. It compresses the range across the league rather than pushing every park in one direction.

There is a wider point buried here about what a park factor can and cannot see. Park factors are relative measurements. They compare a park with the league average of parks in the same period. So a change that hits the whole league equally, a different ball with different drag, a humidor rollout that touches all thirty venues, a rule change that alters how the game is played, mostly cancels out of park factors while doing violence to raw statistics. If home runs across the sport fall because the ball has changed, every park's home run factor can stay roughly where it was while every hitter's total drops.

This is the single most common misreading of the whole subject. A park factor tells you how a stadium compares with its contemporaries. It tells you nothing about whether its contemporaries are collectively a good place to hit this year, and the league baseline moves more often than the parks do.

How a park factor is actually computed

The classical method is a ratio, and its logic is honest. A team plays roughly half its games at home with the same roster it takes on the road, so compare the rate of some event in its home games against the rate in its road games. The roster is a constant across the two halves, so anything left over is the venue. Index it so that 100 means neutral, and a value of 110 means the park raised the rate of that event by about a tenth.

Two problems sit inside that comparison and both are worse than they look.

The reference set is not neutral. A team's road games are not played in an average sample of the league's parks. They are played in the other parks, which is to say in every park except its own, weighted by an unbalanced schedule that sends it to divisional rivals far more often than to the other side of the league. A club in a division full of hitters' parks is measured against a tougher road baseline than a club in a division full of pitchers' parks, and its home park will look more pitcher-friendly than it is. Any serious implementation corrects for the actual road schedule rather than assuming the road is neutral, and the correction is not small.

The park is being judged by the people who chose to play there. This is the circularity, and it is the deepest problem in the field. The hitters generating the home half of the sample were signed by a club that knew what its own park does, and so were the pitchers. A club with a short right field porch has been buying left-handed pull hitters for years. Their home performance therefore reflects both the park and the roster the park caused, and the naive home-versus-road ratio attributes the whole difference to the building.

The better public methods attack the circularity at the level of the individual. Rather than comparing team totals, compare each batter's rate in this park against that same batter's rate in every other park he played in, controlled for his handedness, and do the same for each pitcher. A player is then his own control, which removes the roster-selection bias almost entirely, because a left-handed pull hitter is being compared against himself elsewhere rather than against a league average he was never a member of. This is how the league's own public leaderboard is built, and it is a real improvement on the ratio method rather than a cosmetic one.

What it does not remove is the effect of the park on how players behave. If a hitter changes his swing plane because he plays half his games in a park that rewards fly balls, his own road numbers are contaminated by the habit his home park gave him. There is no clean way out of that, and nobody claims there is.

How a park factor gets built, step by step
  1. Pick the event you are measuringNot just runs. Home runs, doubles, triples, singles, strikeouts and walks all have separate factors, and they do not move together.
  2. Gather the park's events, not the team'sEvery plate appearance staged in the venue, by both clubs, over the window you have chosen. The home side is only half of the sample.
  3. Give every player his own baselineFor each batter and pitcher, work out his rate of that event in every other park he appeared in during the same period. That is the comparison, not the league average.
  4. Control for handednessLeft-handed and right-handed batters do not experience an asymmetric park the same way, and comparing them against a pooled baseline throws away the difference you are trying to measure.
  5. Correct for the road scheduleA club's away games are played in the other parks under an unbalanced schedule, so the road half is not a neutral reference and has to be adjusted towards one.
  6. Turn the comparison into an indexDivide the park's rate by the comparison rate and multiply by a hundred. An invented worked example: 0.038 home runs per plate appearance in the park against 0.034 elsewhere gives 1.118, published as 112.
  7. Regress it towards a hundredHalf a season of home games is a small sample for a rare event. Pull the raw estimate towards neutral by an amount set by how much data stands behind it. The rarer the event, the harder the pull.
  8. Average across several seasonsCombine three, four or five years so that one freak summer of wind does not define the park. This buys stability and pays for it in staleness, because the window includes the stadium as it used to be.
  9. Halve it before applying it to a playerA hitter plays only half his games at home, so a park rated 112 is applied to his season line at roughly 106. Skipping this step double-counts the correction and is a common error in amateur analysis.

The sequence used by the better public implementations. The arithmetic in step six is an invented example.

That last step catches more people than any other. A park factor describes the park. A player did not spend his season in the park. He spent half of it there and the other half in a scattered set of other venues, so the correction applied to his line is roughly half the size of the park's own number. Published tools differ in whether the figure on screen has already been halved, and reading a halved factor as an unhalved one will make you wrong about every player in the league in the same direction.

Why one season of park factor tells you almost nothing

Eighty-one home games sounds like a lot until you count the events you actually care about. Home runs are rare per plate appearance. Triples are much rarer. Split those by handedness, as you must, and the sample behind a single-season handedness-specific triples factor is small enough that the number is mostly noise wearing a decimal point.

The consequence is that a stadium which has not changed by a single brick will publish visibly different one-year factors in consecutive seasons. Weather varies. The set of pitchers who happened to work there varies. A hot dry August with the wind out of the south-west will move a whole season's figure.

Here is an invented illustration of the pattern, with round numbers chosen so the point is legible rather than to describe any real venue.

Worked example: one-year park factors bounce, multi-year ones do not
  • One-year estimate
  • Three-year rolling average
9097.5105112.5120One-year estimate — Season 1: 114One-year estimate — Season 2: 101One-year estimate — Season 3: 110One-year estimate — Season 4: 96One-year estimate — Season 5: 112Three-year rolling average — Season 1: 108Three-year rolling average — Season 2: 107Three-year rolling average — Season 3: 108Three-year rolling average — Season 4: 102Three-year rolling average — Season 5: 106Season 1Season 2Season 3Season 4Season 5

Invented figures for an imaginary stadium that did not physically change across the five seasons. Index where 100 is a neutral park.

Show the numbers
Worked example: one-year park factors bounce, multi-year ones do not
ItemOne-year estimateThree-year rolling average
Season 1114108
Season 2101107
Season 3110108
Season 496102
Season 5112106

The one-year series swings by eighteen points across five years for a building that never moved. The rolling average sits in a band of six. If you were arguing about whether a hitter's season was inflated by his home park, the first series would let you argue either way depending on which year you picked, and that is exactly why nobody serious uses it on its own.

The cost of the multi-year window is real, though, and it should be stated plainly rather than waved away. A five-year regressed factor in use today is partly a description of the stadium as it was five years ago. If a club moved its fences in, raised a wall or rebuilt a section of seating in the interval, the published factor is an average of two different buildings and is wrong about both. There is no clean answer to this. Every implementation is choosing a point on a trade-off between noise and staleness, and the honest ones tell you where they chose.

The practical rule that follows: when a park has physically changed, throw out the history and start again with a short, heavily regressed window, and accept that you will not have a trustworthy figure for two or three seasons. Analysts hate saying this because it means admitting a gap, and the gap is real.

Component factors, and the park that giveth and taketh away

A single overall run factor is the least informative thing published about a park, because a park can be neutral in aggregate while being extreme about the specific things it does.

The mechanism is easiest to see with a tall wall. A fly ball hit at a certain speed and angle is a home run over a five-foot fence and a ball off the wall over a twenty-five-foot one. The park has not reduced the number of well-struck fly balls. It has converted a slice of them from one outcome into another. Home runs down, doubles up, total bases per contact down but not by nearly as much as the home run drop implies, runs down modestly. Judge that park by its run factor and you will conclude it is roughly average. Judge it by its home run factor and you will conclude it is one of the hardest places in the league to hit for power. Both are true. Only one of them matters if you are valuing a slugger.

Deep outfield gaps work similarly on triples, converting balls that would be caught in a small outfield into extra bases, and a large foul territory works in the opposite direction on everything at once by turning would-be foul balls into outs.

Worked example: a park with a tall wall and deep gaps
Home runs86
Doubles118
Triples128
Singles102
Runs99

Invented component factors for an imaginary stadium, index where 100 is neutral. The point is the pattern, not the values.

Show the numbers
Worked example: a park with a tall wall and deep gaps
ItemComponent factor
Home runs86
Doubles118
Triples128
Singles102
Runs99

The run factor in that invented park is a point below neutral. A casual reader concludes the stadium is irrelevant. In reality it is punishing exactly one skill and rewarding two others, and the two players it treats most differently are both on the home roster.

This is why the honest way to use park factors is to pick the component that matches the question. Valuing a power hitter, use the home run factor for his handedness. Valuing a pitcher who gives up fly balls, the same. Valuing a contact hitter with gap power, the doubles factor tells you more than the run factor does. Understanding why two players with similar raw lines have very different park-adjusted offensive numbers usually means finding which component the adjustment is loaded onto.

Handedness is not a refinement, it is the main event

Almost every ballpark in the sport is asymmetric, because almost every ballpark was fitted into a city block rather than laid out on open ground. Asymmetry means the park treats left-handed and right-handed batters differently, and it treats them differently by more than the overall factor suggests, because the overall factor is an average of two populations that are having opposite experiences.

Consider what a short porch in right field does. A left-handed pull hitter aims for it, and his home run factor at that park is well above neutral. A right-handed hitter almost never reaches it, because his pull field is the deep one, and his factor is neutral or below. Average the two and you get a number a little above 100, which describes neither player.

The consequence for reading any published table: a park factor without a handedness split is an aggregate over a population you may not care about. If the hitter you are evaluating is a left-handed pull hitter and the park's asymmetry runs the other way, the pooled figure is not merely imprecise. It points in the wrong direction.

Handedness splits also carry the least reliable numbers on the page, because splitting the sample by handedness halves it, and the sample was already small. This is the point in the whole exercise where noise and importance are highest at the same time, which is an uncomfortable place to be and is why these figures need the longest windows and the hardest regression.

Building a roster to fit a park, and why it is harder than it sounds

The logic is obvious enough that every fan has proposed it. Play in a park that suppresses home runs, so sign fly-ball pitchers whose mistakes will die on the warning track. Play in a park with a short porch in right, so sign left-handed pull hitters who can use it.

Clubs do this. It works, within limits, and the limits are the interesting part.

You are optimising half the schedule. A fly-ball pitcher who fits your park perfectly takes that same fly-ball tendency to every road start, and half his innings are thrown in parks that punish it. The gain is real and it is roughly half the size of the naive calculation.

The market has already priced it. If a park makes a certain kind of player more valuable to one club than to the other twenty-nine, that club is bidding against a market that knows this. The advantage available is the difference between the player's value to you and his value to the next-highest bidder, which is smaller than the difference between his value to you and his league-average value. Roster fit is a genuine edge and it is a thin one, and it competes for the same money as everything else on the payroll, which is where the tax on high payrolls starts to bite on which edges a club can afford to chase at all.

The park can change and the contract cannot. A club that builds a roster around its fences and then moves the fences, or that signs a seven-year deal predicated on a wall that gets rebuilt in year three, has bought an asset whose thesis expired. Fences move more often than people realise, usually because the club decided its park was playing wrong.

It narrows the roster. A team stacked with left-handed pull hitters is a team that a manager with a left-handed bullpen can attack, in a sport where the postseason is played against opponents who get to choose their pitchers. Optimising for eighty-one games against the league can leave you badly shaped for seven games against one opponent.

The version of this that works best is quiet and marginal: preferring the fly-ball pitcher at equal price rather than paying up for him, running platoon splits that lean on the park's asymmetry, and adjusting how a batter is coached to use a specific wall. Those are cheap. Building a whole roster around a building is expensive and it commits you.

What park factors cannot do, and this is the section people skip

A park factor is an average correction applied to an individual, and every gap between those two words is a place it will let you down.

It cannot tell you what happened to a particular ball. A hitter who lost four fly balls to a tall wall in one season and a hitter who lost none get the same adjustment, because the adjustment is applied to a season line rather than to events. The correction is right on average across many players and it is not right about either of these two. That is not a flaw in the method. It is the definition of an average, and it is worth stating out loud because people cite park-adjusted figures as though they had audited each swing.

It cannot separate the park from the schedule. The set of pitchers a hitter faced at home is not the set he faced on the road, and no published park factor knows anything about opponent quality. That correction lives in different statistics.

It cannot handle behaviour it caused. If a park changes how its home players hit and pitch, some of the park's effect has already migrated into the players' road performance, which is the baseline. The estimate is biased towards neutral by an amount nobody can measure.

It has an error bar and nobody prints it. Every published park factor is an estimate with real uncertainty, and it is invariably displayed as a clean three-digit integer. A handedness-specific component factor from a short window can be uncertain by a great many points, and it will be quoted to the unit as if it were a measurement. A statistic that inherits a park adjustment, as WAR does at several points in its construction, inherits that uncertainty silently and reports a tenth of a win.

It does not travel between leagues or levels. A minor league park factor is computed against a different population, a different ball in some leagues, and a different set of parks. Importing one is not a correction, it is a guess with a decimal point.

The new park problem, which is not hypothetical

Everything above assumes a stadium with history. The last few seasons have removed that assumption twice.

The Athletics moved out of Oakland after the 2024 season and began playing home games at Sutter Health Park in Sacramento, a Triple-A ballpark, while a new stadium is built. The Rays played their 2025 home schedule at George M. Steinbrenner Field in Tampa, an open-air spring training ground, after Hurricane Milton tore the roof off Tropicana Field in October 2024, and returned to the repaired Tropicana Field for 2026.

Two clubs, in consecutive seasons, playing major league games in venues with no major league history at all. A club moving from a permanently enclosed dome to an open-air Florida park in summer, and a club moving from a coastal night-time climate to an inland valley one. There is no multi-year regressed park factor for either situation and there cannot be, because the prerequisite is years of data from the same building.

What people do instead is instructive about the whole field. They fall back on the physics: measure the actual dimensions, take the elevation, pull the historical temperature and wind records for that site at those times of day, and model the ball flight forward. They use minor league data from the venue with heavy caveats about the different population playing there. They watch the batted-ball data from the first weeks and update aggressively, accepting that early estimates will be badly wrong. And they say out loud that the number is provisional, which is more honesty than the mature figures usually get.

The Sacramento case makes one point particularly cleanly. An inland valley in high summer is much hotter in the evening than a coastal city, warm air is thinner, thinner air carries the ball, and the difference is not marginal. That is a physical prediction available before a single game was played, derived from the same mechanism that makes altitude matter, and it did not require a park factor to see. Which is a useful reminder that the park factor is a measurement of a physical process, not a substitute for understanding it. This is one of the clearer cases in the wider study of how venues shape results where the physics is knowable in advance and the statistics arrive late.

The same logic generalises past baseball. Every sport has some version of the home venue affecting outcomes, and most of them treat it as a single lump called the advantage of playing at home. Baseball is unusual in that a large, measurable, purely physical component of that advantage can be separated out from crowd, travel and familiarity, because the field itself is different. Baseball is the only one where the pitch is a variable rather than a constant.

How to read a park factor without being fooled

Five checks, in order, and they will keep you out of nearly every trap above.

Check which component you are looking at. Overall runs is the least informative number on the page. If your question is about power, you want the home run factor. If it is about a gap hitter, doubles. A park can be neutral for runs while being extreme for the thing you actually care about.

Check the handedness. A pooled factor for an asymmetric park is an average of two opposite experiences. If your player is left-handed and the park's short field is in right, the pooled figure understates the effect on him, possibly by a lot.

Check the window and whether the park changed inside it. A five-year factor spanning a fence relocation describes a building that no longer exists. Any park that has moved a wall, raised one, or altered its seating in the window needs its history discarded and its current estimate treated as provisional.

Check whether the number has been halved. If a tool shows a park at 112 and you apply the full twelve per cent to a player's line, you have roughly doubled the correction he deserves. Half of his games were somewhere else.

Check what the league baseline was doing. Park factors are relative. A league-wide change to the ball, the strike zone or the equipment moves everybody's raw numbers and leaves the park factors roughly where they were. If a hitter's totals fell and his park's factor did not move, the park is not the explanation and you need to look at the season, not the stadium.

Do those five and you will be reading park factors the way the people who compute them read them, which is with a good deal more caution than the confident three-digit integers on the screen invite.

Common questions

What is a park factor in baseball?

A park factor is an estimate of how much a particular ballpark raises or lowers the rate of some baseball event compared with an average park, expressed as an index where 100 is neutral. A figure of 112 for home runs means the park produced home runs about twelve per cent more often than a neutral one would have, given the same players. It is a correction applied to a player's line, not a description of the player.

How are park factors calculated?

The oldest method compares what happens in a team's home games with what happens in the same team's road games, on the reasoning that the roster is the same in both halves so any difference is the venue. Better methods compare each individual batter and pitcher against their own performance in other parks, controlled for handedness, which removes most of the bias from a team's roster and its particular road schedule. Both are then regressed towards neutral and averaged over several seasons, because one season of home games is far too small a sample to trust.

Why is a single-season park factor unreliable?

A team plays about half its games at home, so one season of park data is a few hundred games' worth of events split across every outcome type you want to measure, and home run rates in particular are noisy at that sample size. The published figure will therefore move by several points from year to year for a stadium that has not physically changed at all. Multi-year windows and regression towards 100 are the standard fixes, and both trade freshness for stability.

Can a ballpark help hitters and pitchers at the same time?

Yes, and this is the most useful thing component factors show. A tall outfield wall turns fly balls that would clear a normal fence into balls off the wall, which cuts the home run factor and raises the doubles factor at once, so the park can look neutral for total runs while being severely unfriendly to a specific kind of hitter. Deep gaps do something similar for triples. Judging a park on its overall run factor alone hides all of this.

Do park factors account for altitude and weather?

Only in the sense that they measure the result. A park factor is an empirical estimate of what happened, so the effects of thin air, heat and prevailing wind are already inside the number without being separated out. That is fine for correcting a season line and useless for predicting a park with no history, which is why analysts fall back on the underlying physics whenever a club moves into a venue that has never staged major league games.

Filed under Baseball·baseball · mlb · baseball statistics · analytics · ballparks