How 2017 Compares to the Steroid Era: Part I
The 2017 season has seen offensive levels rise to a height unmatched in major-league baseball for quite some time. Overall this year, teams are averaging 4.65 runs per game, the highest mark since 2007 — though not quite the five runs per game teams averaged in 1999 and 2000. Most of the offensive increase can be traced to a juiced ball. There’s also been a lot of talk about the role of a fly-ball revolution of some sort or another in the establishment of a new league-wide seasonal home-run record.
An increase in PED use has now been raised as an issue, as well. MLB has administered both PED testing and PED-related suspensions since 2004; both have existed in the minors since 2001. Even with those measures in place, however, power continues to be associated with steroid use, and unfounded rumors have hounded the authors of every breakout season over the last decade. With the rise of power in recent years, the whole league is under suspicion. But how similar is this version of the league to the one now known as the “steroid era”? Let’s take a look at what the latter actually looked like and how it compares to now.
Our split tools are very expansive going back to 2002. This is convenient because 2002 was the last season that lacked PED testing of any kind. It might not have been quite the height of that period now regarded as the “steroid era” — that was probably 1999 and 2000 — but the league looked quite different before testing and suspensions were permitted.
Notably, scoring looked a whole lot like it does now. For some perspective, here are some general statistics that illustrate some similarities and a few differences between the two seasons.
| Season | R/PG | HR | BB% | K% | GB% | HR/FB | ISO | BABIP | AVG | OBP | SLG | wOBA |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 2002 | 4.62 | 5059 | 8.7 % | 16.8 % | 43.5 % | 10.7 % | 0.155 | 0.293 | 0.261 | 0.331 | 0.417 | 0.326 |
| 2017 | 4.65 | 5882 | 8.5 % | 21.6 % | 44.1 % | 13.7 % | 0.171 | 0.299 | 0.255 | 0.324 | 0.426 | 0.321 |
In terms of run-scoring, the 2002 and -17 seasons are almost identical. The manner in which those runs have been produced, however, has changed. This season has been responsible for many more homers, for example, and while the ground-ball rate is similar between the two different campaigns, the number of homers per fly ball is much higher in 2017. The walk rates are the almost identical, while strikeouts have increased markedly since 2002. As a result, both batting average and on-base percentage have declined considerably — a substantial enough difference for the batters of 2002 to have actually recorded a higher wOBA despite scoring slightly fewer runs.
The starkest contrasts occur with strikeout rate and HR/FB ratio, however. More fly balls are leaving the ballpark this year. The two most logical explanations are those invoked above: the juiced ball and an effort among certain batters to get more loft with their swings. (I would recommend reading the links in the opening paragraph for more information on these issues.)
What about the demographics of the league? By looking at performance by position, it might be possible to identify a systemic difference between the 2002 and -17 season. The chart below shows the output by position in 2002 and 2017, sorted by wRC+. (The chart is sortable.)
| Pos | Season | PA | HR | WAR | wRC+ |
|---|---|---|---|---|---|
| 1B | 2017 | 19997 | 933 | 69.9 | 118 |
| 1B | 2002 | 20766 | 780 | 71.8 | 117 |
| RF | 2002 | 20842 | 733 | 86.8 | 115 |
| LF | 2002 | 20922 | 733 | 81.9 | 113 |
| RF | 2017 | 20096 | 778 | 65.9 | 108 |
| DH | 2002 | 9622 | 332 | 13.1 | 108 |
| 3B | 2017 | 19800 | 742 | 80.0 | 103 |
| CF | 2017 | 19979 | 595 | 88.0 | 102 |
| 2B | 2017 | 20066 | 541 | 72.1 | 99 |
| CF | 2002 | 21300 | 573 | 83.9 | 99 |
| LF | 2017 | 19879 | 659 | 43.7 | 98 |
| DH | 2017 | 9884 | 382 | -3.9 | 95 |
| 3B | 2002 | 20614 | 594 | 63.7 | 95 |
| SS | 2017 | 19552 | 516 | 70.9 | 92 |
| C | 2017 | 18517 | 602 | 63.7 | 90 |
| 2B | 2002 | 20967 | 352 | 56.0 | 90 |
| SS | 2002 | 20786 | 423 | 67.5 | 89 |
| C | 2002 | 19178 | 408 | 47.9 | 83 |
In both cases, first base rules supreme. That’s probably not a surprise. Eyeballing the 2002 numbers, we see the traditional power sources like corner outfield and designated hitter up near the top, with the weaker-hitting positions like catcher, shortstop, and second base down at the bottom. In 2017, everything is bunched more toward the middle. To provide a bit more clarity, we can see how the positions have changed in a more direct comparison.
| Pos | 2017 wRC+ | 2002 wRC+ | Change |
|---|---|---|---|
| C | 90 | 83 | 7 |
| 1B | 118 | 117 | -1 |
| 2B | 99 | 90 | 9 |
| SS | 92 | 89 | 3 |
| 3B | 103 | 95 | 8 |
| RF | 108 | 115 | -7 |
| CF | 102 | 99 | 3 |
| LF | 98 | 113 | -15 |
| DH | 95 | 108 | -13 |
Teams still seem able to find those big-hitting first baseman, but in the corner outfield, despite what Giancarlo Stanton and Aaron Judge have been able to do, hitters aren’t providing nearly as much offense as they used to. Relative to the rest of the league, we see an increase at the defensive-first positions like catcher and shortstop with second base and third base also moving up the ladder. Now let’s compare just homers.
| Pos | 2017 % of HR | 2002 % of HR | Change |
|---|---|---|---|
| C | 10.5% | 8.3% | 2.2% |
| 1B | 16.2% | 15.8% | 0.4% |
| 2B | 9.4% | 7.1% | 2.3% |
| SS | 9.0% | 8.6% | 0.4% |
| 3B | 12.9% | 12.1% | 0.9% |
| RF | 13.5% | 14.9% | -1.3% |
| CF | 10.4% | 11.6% | -1.3% |
| LF | 11.5% | 14.9% | -3.4% |
| DH | 6.6% | 6.7% | -0.1% |
The trend remains here, as well, with outfielders accounting for fewer homers relative to their infielder counterparts. Why? It could be that, across the league, teams are putting an emphasis on offense at the expense of defense. It’s possible that shifts have allowed clubs to deploy less talented defenders at premium positions. It could be random.
We can’t say for sure, but the WAR totals mirror the change in offense.
| 2017 % of WAR | 2002 % of WAR | Change | |
|---|---|---|---|
| C | 11.6% | 8.4% | 3.2% |
| 1B | 12.7% | 12.5% | 0.2% |
| 2B | 13.1% | 9.8% | 3.3% |
| SS | 12.9% | 11.8% | 1.1% |
| 3B | 14.5% | 11.1% | 3.4% |
| RF | 12.0% | 15.2% | -3.2% |
| CF | 16.0% | 14.7% | 1.3% |
| LF | 7.9% | 14.3% | -6.4% |
| DH | -0.7% | 2.3% | -3.0% |
If there were rampant PED use in the same way there was during the steroid era, we might expect to see the same types of wide gulfs that we saw in 2002. Instead, we see a leveling out. If we assume that PED is one of the main causes of increased power, the data suggest that absolutely everyone is on PEDs and that they’re all using those PEDs to power up in such a manner that corner outfielders no longer have a big advantage over second or third basemen. While we commonly talk about the PED era affecting everyone, the era was still one of extremes.
Consider that, in 2017, the top-30 home-run hitters make up roughly 18% of the total home runs and the middle-100 players (with at least 300 PA) make up 27% of the home runs. Back in 2002, the top-30 home run hitters — led by Alex Rodriguez (with 57) and Jim Thome (52) — made up 22% of the home-run totals, with the middle-100 players accounting for 24% of them. Things were more extreme in 2002, which suggests that the changes in 2017 are more likely a product of something that’s affecting the entire league as opposed to one subset of players. We will look further into this potential affect looking at player age, as well as the role pitching has played, in Part II.
Craig Edwards can be found on twitter @craigjedwards.
I am not entirely convinced that those numbers re: “leveling out” are particularly significant. It seems entirely plausible to me that the “steroid era” was, like the current era, actually caused by a juiced ball.
This is entirely possible. 15 years has passed. In the world of sports that is literally a lifetime. I’d argue that athletes today are stronger, faster and more explosive than most or all 2002 athletes regardless of PED usage. Science of sport, nutrition and recovery has moved almost exponentially.
BP did a lot of analysis back in around 2002-2004 that suggested the primary factor was neither the ball or the players but the rash of new ballparks that were a lot smaller than the parks they replaced. Parks adjusted a few times in the interim, and more have sought to get closer to average since the introduction of park factors to general understanding. In the 1990s, there wasn’t any such concept in numerical form.
The final assessment at the time I believe was that the increase was about 60% due to changes in park dimensions with the remaining 40% unexplained, whether it was due to changes in player ability/strength (eg PEDs) or changes in approach (more guys trying to hit home runs), changes to the ball (which they didn’t have any data to analyze), or random variance.
“The trend remains here, as well, with outfielders accounting for fewer homers relative to their infielder counterparts. Why? It could be that, across the league, teams are putting an emphasis on offense at the expense of defense. It’s possible that shifts have allowed clubs to deploy less talented defenders at premium positions. It could be random.”
I think the game has changed in a way that has increased its emphasis on outfield defense. I would couple that with the idea that shifting reduces the need for top defenders in the infield. For example, Byron Buxton bailing out the Twins pitching staff 3-4 times a week by saving extra base hits in key moments arguably justified him spending an entire 1/2 of a season completely lost at the plate. If he’s a shortstop, the data technology of shifting makes continuing to give him plate appearances in the name of premium defense less justifiable.
Watching Corey Seager play above average SS defense all year just due to superior positioning and data analysis has me agreeing with you. He’s got sure hands and a great arm, but he’s not as quick as you’d normally expect as SS to be.
There’s always been players who were better at positioning than others, but now every team has the benefit of being positioned as well as the best guy they can find can determine with the assistance of data. The Dodgers are positioned from pitch to pitch based on signals from their IF coach from the dugout, and the outfielders use those cards and are also responding to signals from inside the dugout.
The difference between teams is now about who has better analysis of the available data and an infield guru capable of understanding it, and that’s still a gulf, but it’s not as large as when it was an individual player skill.
One game recently the Dodgers trotted Logan Forsythe out as a shortstop for a game and he was fine, every play he made was routine, because he was standing in the right place every time.
How much of the decline in LF production from 2002 to 2017 is a result of no Barry Bonds and his 12.7 WAR/244 wRC+ season?
2017 Mike Trout – 309/446/629 – 18.7 BB% – 17.5 K%, .320 ISO, 182 wRC+
2002 Barry Bonds – 370/582/799 – 32.4 BB% – 7.7 K%, .429 ISO, 244 wRC+
Oh my…
So I had to think about this for a second, and then concluded that pitchers’ strand rate must have been higher then — to record a higher wOBA despite scoring fewer runs, you’d have to have more players getting on base and then not getting driven in, correct? But I can’t find strand rate in the splits tool, and when I compare the batting lines using the “with runners on” filter, the two years look almost identical. So.. I think I’m using the splits tool incorrectly, or my basic assumption is wrong. Can somebody clue me in?
If PED use was back to a large degree, we would anticipate changes in aging patterns like last time, too, right? We’d be seeing players extend their peaks further into their 30s, with a spike in the number of slugging players who fend off retirement until their late 30s or early 40s. That’s what we had last time, right? Are we seeing that again?
This is the thing about the steroid era, you had unremarkable hitters like Brett Boone and Rich Aurilia all of a sudden put up monster numbers adn then drop off a cliff. I don’t see that happen much in today’s era.
This is the key right here:
“If there were rampant PED use in the same way there was during the steroid era, we might expect to see the same types of wide gulfs that we saw in 2002. Instead, we see a leveling out. If we assume that PED is one of the main causes of increased power, the data suggest that absolutely everyone is on PEDs and that they’re all using those PEDs to power up in such a manner that corner outfielders no longer have a big advantage over second or third basemen. While we commonly talk about the PED era affecting everyone, the era was still one of extremes.”
But this is assuming PEDs help only batters, not pitchers. If, e.g., more pitchers relative to batters use PEDs now, that would also result in a leveling. That wouldn’t account for the increase in absolute numbers like HR and scoring, but there are certainly other factors like strike zone size that could be at play.
Either way, it sucked to be hitter from 2005-2014. I think there will be quite a few HoF snubs due to the lack of HRs.
It did? PED testing didn’t actually much affect offense.
If you sort all of baseball history by ISO, 13 of the top 20 years start with a 2…so 13 of the 18 years since 1999 are among the top 20 years for power. If you sort by PA/HR, 14 of the highest HR rates have been since 1999….the “worst year” to hit since 1999 was 2014, with the 33rd highest HR rate ever.
This entire century has been a good time to be a hitter compared to historical norms. The HOF snubs aren’t because guys hit few HRs, but because so many guys hit so many and so many voters are suspicious moralizers.
You should really be using Non-pitchers only. There’s one more DH team now than there was in 2002.
There’s no doubt that HRs are more broadly distributed now than in 2002, by any measure you like, and probably more so than ever before.
I tried to see if playing time now favors HR ability more than before — i.e., are teams selecting for power, as many folks think? Or is it more about self-selection, i.e., there just aren’t many powerless guys in the pool? So I ran this little study:
I took the top 300 in PAs for 2002 and 2017, split each set in half by PAs, found each group’s HR% (per PA), and compared this year’s rates to 2002.
Both 2017 groups had significantly higher HR% than its 2002 counterpart, but the semi-regulars had MUCH bigger gains:
— Full-timers: 2002, 3.38% … 2017, 3.81% … Net +0.43%
— Semi-regulars: 2002, 2.23% … 2017, 3.27% … Net +1.04%
That is suggestive, but it could still be self-selection. So I ran it again, taking the top 400 in PAs for each year, splitting each into 4 equal groups. Now the HR% gains are biggest for groups 2 and 3, double those for groups 1 and 4:
— 1) PA rank #1-100: Net +0.42% (2017 over 2002)
— 2) PA rank 101-200: Net +0.87%
— 3) PA rank 201-300: Net +0.83%
— 4) PA rank 301-400: Net +0.44%
I *think* this shows teams selecting for power. My interpretation:
— Group 1 skews towards players whose non-HR skills dictate full-time play, while Group 4 skews towards (a) bench types and (b) guys getting a chance but not earning more PAs. Gains in HR% by those groups would tend to reflect mainly the overall context.
— Groups 2-3 would tend more towards guys earning more PAs via performance. And the fact that their HR% gains are the biggest suggests that is what teams are valuing most.
— In 2002, groups 3 and 4 had the same HR rate, 2.2%. In 2017, group 3 has a solid edge on group 4, 3.00% to 2.64%. This also suggests that power is now a bigger factor in getting more PAs.
Anyway, that’s what I think. Other interpretations are welcome.
If you’ll pardon a tangent … In the long view, extra-base hitting has mainly risen throughout modern history, measured either by percentage of all hits or by total bases per hit. Even during the dead-ball era, 10-year averages for those rates rose steadily. It seems like a pretty strong trend.
For a long-range comparison, take any 10-year average of total bases per hit, and compare to the same figure from 30 years prior. (E.g., the current period of 2008-17 averages 1.59 TB/H, while 1978-87 averaged 1.50, so the current net is +0.09 TB/H.)
Out of 78 such comparisons, 70 are positive. All 8 negatives were very small (-0.02 or less), and came consecutively in the periods ending 1988 to 1995.
The current net of +0.09 TB/H is about the average for all such comparisons. The peaks of the PED era averaged +0.12 by this method.
‘Twas ever thus.
The choice of year matters. If there were data for 2001, I imagine Bret Boone alone would skew the HR numbers for 2B.