This report is generated from a local JSON file. The chart, summary cards, and sortable tables run in your browser. No data is sent to a server, your usage is not tracked, and this page will never have ads.

As the 2026 Major League Baseball season progresses and more details are emerging of what the owners want in the upcoming collective bargaining agreement, it's starting to look like there may not be any baseball in 2027. The owners and players are far apart with owners indicating that they want a salary cap for competitive balance reasons.

With the current labour problems, I am thinking back a lot to 1994 when I was still in high school and the owners last proposed a salary cap. Baseball stopped in August... totally stopped and there was no World Series. There was a World Series all throughout the second world war, but a labour stoppage in 1994 resulted in the entire season including the playoffs being cancelled.

But there's a bit of a problem with my memory. It all seems romanticized and in my mind, baseball was experiencing a serious offensive heyday in the months leading up to the end of baseball as we knew. I remember Tony Gwynn flirting with .400, Matt Williams chasing Roger Maris, Jeff Bagwell and Frank Thomas having monster years, and Paul O'Neill really coming into his own in New York.

Then the strike stopped everything in August. It's strange because I usually remember pitching and pitchers but in my memory I can only remember the offensive accomplishments. And so that got me thinking, asking myself some questions... and then digging into Retrosheet data. :)

This report is the result of that digging. It starts with a simple question - did the league-wide offensive environment itself look unusual before the strike? Or is that just my carefree teenaged years adding a lustre to otherwise typical performances? The compares regular-season production through August 11 in every season from 1989 through 1998.

- 1994 Offensive Environment -

The dashboard describes the production pattern. It does not make or support any claims about a different baseball or come to any conclusions on why the environment moved. It does not incorporate umpiring patterns into the analysis and only looks at the overall offensive environment in Major League Baseball in 1994, compared to years immediately before and after.

What counts as a signal?

A high rank or a sharp year-over-year move can show that an offensive environment was unusual. It cannot independently identify causes. Maybe I'll write a report that digs into that later.

Loading 1994 offensive environment report data.
— 1994 rank in runs per team-game
— 1994 rank in home runs per team-game
— 1994 rank in OPS
— 1994 home-run rate change from 1993
— 1994 home-run rate change after removing Mile High games
— complete 1994 regular-season games analyzed

The chart always uses MLB-wide totals through August 11 of each season. Rates are recalculated from aggregate totals rather than averaged from game-level percentages.

  • MLB season rate
  • 1994 report year
MLB production through August 11, 1989–1998. The 1994 point is highlighted. 1995 is a valid rate-stat comparison but has a later opening date because of the strike, so it should not be treated as an ordinary calendar baseline.

1994 was not ordinary. It ranked near the top of this ten-season window in almost every core measure of offensive production. But the data only supports one conclusion - there was an offensive surge that began in 1993, remained high in 1994 and peaked again in 1996.

Then all those records started getting creamed. The conclusions are fairly obvious.

- Four Seasons of Context -

The 1994 result is easier to read beside the low point before the surge, the first move upward, and the high point that followed.

Loading season context

The report will add 1992, 1993, 1994, and 1996 after the local data file loads.

- Season Production Table -

One row per season and scope. The default is MLB-wide production, while American League and National League rows retain the host-league scope used by the underlying environment analysis.

Waiting for season data.
League-wide offensive production through August 11 in each season.
Loading season production data.

- 1994 Leave-One-Park-Out Test -

Each row removes every 1994 game played at one park, then recalculates the remaining MLB rate. A near-zero change means that one physical park does not control the league-wide conclusion.

Waiting for park sensitivity data.
1994 park sensitivity for home runs, runs, and OPS.
Loading park sensitivity data.

About this report

As I mentioned previously, 1994 has a weird and blurry place in my memory. I was young, obsessed with baseball, software and other teenaged things. When I think back to the season before the strike, it all seemed bigger - like offensive production hit an absolutely massive peak with some massive records (like Roger Maris' 61 home run season) seeming poised to fall and Tony Gwynn making a serious run for .400.

Weirdly, I do not remember any pitching performances whatsoever and in my old memories, 1994 seems like the year of the bat. And so while digging into Retrosheet data, I started to come up with an idea. And so I wrote some code, analyzed the data and started trying to answer a simple question.

Did MLB's offensive environment shift in 1994 in a way that exceeded the normal seasonal and park variation visible in nearby years?

Well, not so simple. This report is a first-pass answer. It uses Retrosheet team-game batting totals and game context from 1989 through 1998. Every season is cut off after August 11, matching the final day of the 1994 regular season. That gives 1994 a fair comparison window instead of comparing a 113-game player season or a strike-shortened league total to a full 162-game season. It creates some weird data and I don't think the methodology was right, but we'll file this one away as an exploratory pass. And, it shows some league wide trends that have been reported before, so the methodology is at least close.

What the data says

1994 was an unusual offensive season within the ten-season window that I analyzed.

Through August 11, the 1994 season ranked second out of ten seasons in runs per team-game, home runs per team-game, batting average, slugging, OPS, ISO, and home runs per plate appearance. It ranked third in BABIP. Compared with the same point in 1993:

  • runs per team-game increased by 6.6%;
  • home runs per team-game increased by 16.3%;
  • ISO increased by .0158;
  • home runs per plate appearance increased by 14.6%.

That is strong evidence that the 1994 offensive environment was not a memory trick constructed from Gwynn, Bagwell, Thomas, Williams, a very good Yankees outfield and those little distortions caused by 32 years of time.

But it is also not any kind of indication that something special happened leading up to the strike. Instead, it's just part of a wider and more interesting trend that culminates with a lot of the big years in the 'steroid era'.

The offensive rise already began in 1993. That year had the largest year-over-year increase in runs and home-run rate in this window. Then 1994 stayed high, and 1996 exceeded 1994 across most of the same measures. The cleanest summary is not "something suddenly changed in 1994." It is:

MLB entered a high-offense stretch beginning in 1993. 1994 was a major part of it, but not an isolated outlier.

And so while there were some amazing individual seasons cut off in 1994, it was just part of a broader trend that continued with things like Brady Anderson's 50 home runs in 1996 and the Sammy Sosa/Mark McGwire race to chase down Roger Maris' record.

The Coors question

The 1994 Colorado Rockies played their first season at Mile High Stadium. I wrote about it previously using a different way to analyze the contributions from a stadium. That is a very large potential confounder, especially for a report that asks questions about league-wide offense.

So the report runs a test where I leave one park out. For each 1994 park, it removes every game played there and recalculates the remaining MLB rate.

Mile High clearly raised scoring. Its 57 games produced 5.90 runs per team-game, well above the league rate. But removing those games changes the MLB home run rate from 1.0331 to 1.0324 home runs per team-game. That is a decline of about 0.07%.

In other words, Mile High mattered to the 1994 scoring context, but it does not explain the league-wide home run signal. And so while the Coors effect is real whether you analyze individual performance or overall performance, it's not the only thing going on here.

What this report cannot tell us

This is descriptive baseline analysis, not a causal model.

It cannot determine whether:

  • performance enhancing drugs were more or less widespread in 1994;
  • baseball construction changed (the balls got juiced);
  • any change was intentional;
  • ownership, players, umpires, or anyone else made a decision;
  • the rise came from player talent, training, pitching quality, expansion, park changes, weather, rules, scoring conventions, or a combination of several factors;
  • 1994 was unique relative to a wider historical sample.

The weather fields in the Retrosheet game context are useful for future work, but their coverage is inconsistent in early seasons. And so this report describes the fields but does not try to build any kind of weather-adjusted model to find a fit.

1995 also deserves caution. Its rate statistics are useful, but the season opened late because of the strike and so the games played are notably lower. It shows rates of production, but over a season that looked very different.

The connection to the strike attendance report

The MLB Attendance and the 1994 Strike report looks at the same historical moment from a completely different direction.

That report shows that raw 1994 attendance is a bad measure because the season stopped. Attendance per opening actually remained strong in 1994, while the more obvious attendance wound appears in 1995 when baseball returned.

This report adds a small piece of on-field context. Fans were treated to a very exciting game in 1994. The game was producing unusually high offense before the strike stopped it. That is interesting historical context, but it is not proof that offense drove attendance or that the strike damaged the sport more because it interrupted an exciting season. The two reports should sit beside each other because they look at the same era through different questions.

In the future, I think I should write another report that looks at attendance versus offensive performance.

Methodology

The source data comes from Retrosheet:

  1. simplified teamstats.csv for value-only team-game batting totals;
  2. main gameinfo.csv for game date, park, home/away context, and available environmental fields.

The analysis includes regular-season games from 1989 through 1998, through August 11 in every season.

For the first version, the main MLB rates are:

  • runs per team-game;
  • home runs per team-game;
  • batting average;
  • on-base percentage;
  • slugging percentage;
  • OPS;
  • ISO;
  • home runs per plate appearance;
  • BABIP.

The pipeline aggregates the raw totals first and then calculates each rate from the aggregated denominator. It does not average game-level OPS, batting average, or home-run rates. Rolling values are calendar-day windows calculated from rolling totals.

For the American League and National League output, the scope follows the home team's league. That made both batting lines in an interleague game belong to the host environment after interleague play began in 1997. MLB rows contain every included game.

My data quality check didn't find anything missing or any malformed rows because of team names, etc. In other words, me do good but would be good if no had check work.

Accessibility notes

This report is designed around text and tables first.

The line chart is there to make the season pattern easier to scan, but the season table contains the same numbers. Both tables use captions, column headers, and regular buttons for sorting. The chart metric selector is a normal form control, and the result counts announce updates through a live region.

This page was scanned for accessibility with Siteimp and tested with NVDA.

I am still learning accessibility, so I make mistakes. You would maybe think that after 16 years I could have stopped learning by now, but thus far it's been a constant learning process. But... any day now!!?? :) If something does not work with your screen reader, keyboard, zoom settings, or browser setup, please contact me so I can fix it.

What did we learn?

The 1994 season really was a high-offense season. That is visible in the league-wide data, not just in the famous names that survived in my teenaged brain.

It was high enough to deserve investigation. It was not clean enough to make any deeper claims about a juiced baseball or even any steroid use, though the numbers sure line up with the steroid era.

The strongest conclusion is that 1994 is part of an offensive surge from 1993 through 1996. The same surge lead to the famed Sammy Sosa/Mark McGwire race to beat Roger Maris' old famous home run record. The next useful version of this report would widen the historical window, add an expansion-era sensitivity pass that removes Rockies and Marlins games, and decide whether the available weather coverage can support a properly limited model.

For now, I am happy with the small and and am especially happy that I have at least three more reports to write! Baseball rules.

Related Links

  • MLB Attendance and the 1994 Strike

    A companion report on the same baseball moment. It shows why 1994 was schedule-distorted for attendance and why 1995 is the clearer year for measuring the post-strike attendance wound.

  • Coors Field and Fantasy Baseball

    A separate report on player movement into and out of Colorado. This report uses a different league-wide park sensitivity test to make sure Mile High Stadium does not explain the 1994 result by itself.

  • Retrosheet CSV Downloads

    Retrosheet provides the team-game batting totals and game context used to build the local data for this report.

  • MLB Ball Juiced Again? 2026 Data Shows Mid-Season Change

    On May 25, 2026 baseballs started carrying like they did before Covid. This report digs into that data.

  • Doping in Baseball

    A Wikipedia article on the use of performance enhancing drugs in Major League Baseball.

About Baseball Reports

These reports are small data projects built around practical baseball questions. The goal is to make the data readable, useful, and accessible while having some fun with Python and JavaScript. If my fantasy team sucks less as a result, that is a bonus.

About This Data

This report combines Retrosheet simplified value-only team-game batting totals with Retrosheet game context. It covers regular-season games through August 11 for each season from 1989 through 1998. It is descriptive analysis, not a model that can determine intent or baseball construction.