About donged

Every home run is a home run in the park it was hit in. The interesting question is what it would have been anywhere else — and because no two outfields in Major League Baseball are the same shape, the answer is very often “a flyout”.

What the site is for

A 375-foot fly ball to left-centre is a souvenir at Fenway Park and the second out of the inning at Comerica Park. Broadcast graphics will tell you a ball travelled 412 feet; they will not tell you that it only cleared the wall because it was hit in Cincinnati. donged closes that gap. It takes each tracked MLB home run, measures the ball against all thirty outfield walls at the angle it was actually hit to, and tells you how many of the thirty it would have left the yard in.

The number that comes out is the point of the site. A homer that clears all thirty is a no-doubter in the literal sense — there is no address in the league where it stays in play. A homer that clears eight is a park effect wearing a home run’s clothes. Both count the same in the box score, and they are not the same swing.

Where the data comes from

Home runs, batters, pitchers, game state and batted-ball measurements come from the public MLB Stats API, which carries Statcast tracking for every ball put in play. The fields used here are projected distance, exit velocity, launch angle and spray angle. New home runs are ingested through the day, so a ball hit tonight normally appears on the site within a few minutes of the swing.

Projected distance is the important one and it is worth being precise about what it means: it is how far the ball would have travelled had nothing been in its way, not where it physically landed. A ball that hits a facade twenty rows up is credited with the distance it was still carrying, which is exactly the measure needed to ask what another park would have done with it.

Wall distances for the thirty parks are held here as a profile of seven measured points across each outfield — the left-field line, left field, left-centre, straightaway centre, right-centre, right field and the right-field line. Between those points the wall is interpolated linearly.

How a home run is scored

Spray angle is expressed in degrees, with −45° the left-field line, 0° straightaway centre and +45° the right-field line, normalised so that the geometry reads the same regardless of which side of the plate the batter stood on.

For each of the thirty parks, the wall distance at that exact spray angle is interpolated from the two nearest measured points. If the ball’s projected distance is greater than or equal to that wall distance, it is scored as a home run in that park; otherwise it stays in play. Do that thirty times and you have the number you see on every home run page — and the park-by-park table underneath it shows the working, park by park, so you can see which walls it beat and by how much.

The same arithmetic run the other way is what powers the ballpark pages: take a week of tracked home runs from around the league, run all of them against one park’s wall, and the share that clear it is a straightforward measure of how much that park is helping. Ranking the thirty by that share is what the site means when it calls a park hitter-friendly.

What the model leaves out, on purpose

Distance and direction, and nothing else. Three things a complete physical model would include are deliberately absent, and it is better to say so plainly than to imply a precision the numbers do not have.

Wall height is not modelled. Every wall is treated as a line on the ground at its measured distance. A ball with the distance to clear the Green Monster’s 310 feet is scored as a home run at Fenway even though a 37-foot wall might have played it off the boards. This is the largest single simplification in the model and it makes the site slightly generous to the handful of parks with unusually tall fences.

Trajectory arc is not modelled. Two balls with the same projected distance can arrive at the wall at very different heights — a towering fly and a line drive land in the same spot having spent their flight completely differently. Only the landing point is compared here.

Air is not re-simulated. Projected distance already reflects the conditions the ball was actually hit in — the altitude, temperature and air density of that park on that night. It is not re-derived for the other twenty-nine. So a ball hit at Coors Field carries its Denver distance when it is measured against a sea-level wall, which flatters Coors home runs slightly, and the reverse understates what a sea-level ball would have done in thin air.

None of these change the shape of the answer — the difference between a park’s 330-foot line and another’s 400-foot alley dwarfs all three — but they are the reason the number is best read as a good comparison rather than a physics verdict.

What is on the site

Recent home runs is every tracked homer with its park count. Each has its own page with a field diagram, the Statcast measurements and the full thirty-park table. Player pages aggregate a hitter’s tracked home runs and break down which parks would have held them. Ballpark pages cover the wall profile, the park’s current home-run friendliness ranking and what makes each one play the way it does. Dates groups everything by the day it was hit.

The site is free, has no accounts and no paywall, and is supported by advertising. It is not affiliated with, endorsed by or sponsored by Major League Baseball or any of its clubs; all club names and marks belong to their owners, and the underlying data is consumed from MLB’s own public API.

Who makes it

donged is built and run by Haynes Lab, an independent software studio. Corrections to a wall profile, a bad measurement, or anything else that looks wrong are genuinely welcome — mail [email protected]. See also the privacy policy.