Power/Speed Teams

I’m going to violate my new rule and do this as a stand-alone post, even though I just came to it today. I own it will be approximate, somewhat silly, and inevitably unpolished.

I can’t recall seeing Bill James’ formula for Power/Speed¹, which rewards those players who not only have high totals of home runs and steals, but have good balance between the numbers, applied to teams. I’m not sure why it isn’t, commonly. It seems to me that the formula is essentially as well-suited to measuring the trait in teams as players, and there is the commensurate interest in teams that display both power and speed. The residue of the number crunching, of course, would not really be to get insight into winning teams, but more a means of highlighting some talented offenses, or a source of compelling trivia, if “compelling trivia” isn’t an oxymoron.

Stathead doesn’t seem to allow you to extract the highest Power/Speed numbers for teams, so I had to settle for subsetting by both home runs and steals, and taking it from there.

I first asked for teams that had had 200 home runs and 175 steals and only got two, the 1996 Rockies, and the 1998 Blue Jays.

It turns out the Rockies are the only 200/200 team in big league history.

Then I asked for 250-steal, 175-home run teams. To my surprise, there hadn’t been any of those.

Clearly, I was setting the bar too high, so I decided I’d just work with all of the 200-HR teams and rank them by stolen bases. After the 1996 Rockies (201 steals) and 1998 Blue Jays (184 steals), it is a gap back to the 1999 Reds (164 steals).

The way that the file emerged, team record was one of the first columns visible, and as I had rather randomly referenced that at the beginning of my abandoned note, I happened to see that I was ironically staring at a lot of winning records. In fact, it turns out that the top 24 teams² in stolen bases among 200-HR teams all had winning records. So, perhaps, so much for stolen bases not really mattering.

But then I thought more critically. It occurred to me that maybe a team really wasn’t going to rack up a lot of stolen bases unless it played before home runs became more commonplace, and if a team hit 200 home runs in that game, that was a real power-hitting team, and a team that probably won a lot of games.

Or maybe, to rack up those steals, a team has to have its share of hits and walks. A team that can do that, and hit a lot of home runs, has a hell of an offense.

And while we’re all wise to conditions where home runs can be cheap, a la 2019, there was also the matter perhaps of 200 home run teams just generally being good. So I had to check the rest of the file, not just the teams with the most steals, and see how successful they had been.

To correct for overall winning percentage, the most straightforward intervention was to correlate steals and winning percentage in the group (there have been 235 200-home run teams, by the way). The correlation was .13, and the interpretation of that, I would say subjective. It rated as small, but I didn’t think it was realistic to expect it to be high. To hypothesize a large role for stealing in winning would be extravagant, and what could be found figured to be diluted by the challenge of differentiating teams in the middle of the two categories. The p-value was also .06, and if you made it one-tailed (certainly a reasonable position), then the p-value was .03, and statistically significant.

I initially had it in my mind that I would do a fine-grained analysis of winning among many segments of stolen base teams, but when I actually looked at the distribution for steals, I realized that would be overkill. Dividing the group into 9 tiers, for instance, to have groups about the size of the one that had unfailingly shown winning teams, would have been wrongheaded, because in truth so many of the 200-home runs teams didn’t steal at all. About two-thirds had 100 steals or less. Going with nine groups would have roughly meant segmenting out teams that had stolen base totals in the 70s from teams that had steals in the 80s, for instance, and why would those teams magically have had different winning profiles?

So I tried to be practical instead, attending to possible impact and to variation in the data, and where I came out instead was the creation of four groups. These were defined as follows.

Group 1: less than or equal to 71 steals (80 teams, 34% of the sample)

Group 2: 72-100 steals (77 teams, 32.8% of the sample)

Group 3: 101-131 steals (55 teams, 23.4% of the sample)

Group 4: 132+ Steals (23 teams, 9.8% of the sample)

While using winning percentage is technically better, 200 home runs seems to be basically a post-154-game-schedule phenomenon (the teams played an average of 161.7 games). As I thought performance was more intuitive to absorb using wins, that is what I will use in the analysis.

What I found was that the group-4 teams won an average of 92.9 games, and the other three groups had average wins of 87.4, 87.9, and 87.3, respectively.

So we learn, first, that the high-steal teams did, in fact, play at a level well above that of the other 200-home run teams.

Second, we learn there really wasn’t an association between steals and wins throughout the dataset. If steals made a difference, it was only when they were particularly high.

I also did the analysis using median wins. Group 4 was additionally flattered by doing this, as their median win total was 95. Group 3 also did better by median than mean, although only moving up to 89 wins. Group 2 and Group 1, for their parts, presented almost the same picture as before, with 87 median wins apiece.

I hate to narrative bust, but I also have my credibility to consider, and I must admit that I am still doubtful of the association between steals and wins. I tend to think what I found coincidental. It’s just that my analytical prejudice against steals is not so easily vanquished, you understand

That .13 correlation between winning percentage and steals used the dataset as a whole, and so doesn’t represent the highest estimate for the role of stolen bases, which only applied to the top stolen base teams. That said, skepticism of the causal role of stolen bases grows when we find that the correlation between steals and on-base percentage among the teams in the database was higher, at .28. Were the teams really winning more games because of their steals, or possibility the team speed revealed by those steals? Or were they winning more games because they got on base more, and maybe then were able to steal more? Adding to this idea, the correlation between OBP and winning percentage was highest of all, .38.³ So there is a very good prospect that if we formally carried out the necessary steps,⁴ the relationship between steals and wins would be found to be “mediated” by OBP, to invoke the proper academic term.

One thought I had was that there have been all of these 200-HR teams, and all of these 200-steal teams (282 of them), but only one time have they come together. Emerging from my reverie, I generated all 200-steal teams, thinking maybe I would search for relatively high HR teams from that group to supply the other side of the 200-HR list.

I aim to be a defender and preserver of old-time baseball, but I have to say that I found myself drowning in pre-1920 teams from the 200-steal list, and thought of maybe employing a time cutoff. As I got more serious about doing this and homed in, I realized I didn’t understand what a feat it has been to steal 200 bases post-1919; it’s only been done 30 times. This goes a ways towards explaining why there hasn’t been a 200/200 team, outside of the 1996 Rockies.

I pivoted to asking Stathead for 175/175 teams, in addition to the ones I’d run into from 200-home run teams, the 1996 Rockies and 1998 Blue Jays. I found four more: 2009 Rays (199 HR, 194 Steals, 84 wins); 2023 Reds (198 HR, 190 SB, 82 wins); 2007 Mets (177 HR, 200 SB, 88 wins); 2016 Brewers (194 HR, 181 SB, 73 wins).

Unearthing the 2023 Reds, who flirted with 200/200, made me grapple with just how unmistakably impressive home run/stolen base combinations increased last season. I realized that that group of the 24 highest stolen base 200+ HR teams I had analyzed, the group that all coalesced in having winning records, was made up of four teams from last season (the Rays, Phillies, Padres, and Braves⁵).

This year, however, with home runs in remission, the sequel is not quite living up to the original. The best candidate to finish with interesting numbers is the Brewers, exciting in the same vein that the Reds are, to be sure. They are on pace for 166 HR and 227 steals. They need to up their power game a little bit, for my amusement, if not necessarily to win the World Series. With a big second half, perhaps the Phillies could also make some of these lists; they are on pace for 191 HR, 164 SB.

I admit the idea probably just substantiates the thought that the significance of power/speed data doesn’t go much beyond statistics, but a look under the hood at the 83-79 ‘96 Rockies reassures you that they are the team that should be our lodestar. First and foremost, they had two 30/30 players, Ellis Burks and Dante Bichette. Then, with 47 home runs and 18 steals, Andres Galarraga had a 26.0 power/speed number. If Larry Walker had been able to play a full season, he probably would have been 30/30, too. As it was, he had perfect balance, hitting 18 home runs and stealing 18 bases in 83 games. And just about every manager will welcome home runs, but with Don Baylor at the helm, who was four times a 20-20 guy and nearly a 300-300 guy for his career, the Rockies were represented perfectly in the power/speed cause. I also don’t think it’s ridiculous to speculate that someone who played the style that Baylor did was likely to let his players follow his path.

The 1998 Blue Jays had two notable power/speed players. Shawn Green reached 35 in both homers and steals. Jose Canseco was one bag short of 30-30 (no tears of regret for that liar’s loss), but hit 46 home runs.

While I do think this group of current Reds is exciting, and in the spirit of a power/speed team, it is also true that they are a thoroughly weird team playing in the dawn of a new era where team steals can be broad-based and truly a team effort. Could for the team bottom line, but maybe less exciting to the statistician.

We know now that Elly De La Cruz can be prolific, but not reaching the major leagues until June 6 of last year, he led the 2023 Reds with just 35 steals.

Only three Reds played 120 games, although Spencer Speer did play 156. He led the team with 23 home runs, the only Red to hit 20, and also stole 15 bases.

But the team leader in power/speed, at 21.6, was TJ Friedl.

The 2007 Mets, one of the six 175/175 teams, had David Wright go 30/30, and Carlos Beltran 20/20. Jose Reyes was the key to their 200 steals; he swiped 78 bags. He hit only 12 home runs, and no one on the team outside of Wright, Beltran, and Reyes stole more than 12 bases.

Jonathan Villar (62 steals, 19 home runs) played the Reyes role on the 2016 Brewers. He also led the National League in power/speed, if you can believe that (Jonathan Villar!). The top five in the NL were completed by Wil Myers, Paul Goldschmidt, Jean Segura, and Bryce Harper. None really toolsy, traditional power/speed guys.

Baseball is certainly in a better place for the relaxed stolen base conditions. Last year, 36 National League players had power/speed numbers of 15 or higher. In 2016, 15 players did.

But even with the conditions that prevailed in the year Trump first ran for president, the Brewers somehow show up on a historical list of excellence defined in absolute terms, not relative to the league. With 30 home runs and 16 steals, Ryan Braun rated 7th in the league in power/speed. Super utility man Hernan Perez (oh where have you gone?) had 13 home runs and 34 steals; that was a combination enough to place him 10th, and tempting me to invoke Ripley again.

That we have a Rays team to discuss here, the 2009 Rays, should come as no surprise to anyone. Jose Reyes and Jonathan Villar were played on that team by Carl Crawford. He’s actually the only one of those three not to lead the league in steals in the featured season, but his 60 was a career high, and he hit 15 home runs. Ben Zobrist muscled up with 27 home runs, his career high, and stole 17 bases.

But for the most part, this Rays team brought home to me something I should have realized earlier: a team can reach heights in both home runs and stolen bases with independent contributions. The team power/speed number is not the sum of the individual power/speed numbers, in other words. If a team had Pete Alonso at first, Dan Uggla at second, and Vince Coleman and Juan Pierre in the outfield, while the best scenario for power/speed would be if everyone contributed in both places, the team might still have a good home run total, and would certainly have a good stolen base total.

The 2009 Rays got 33 and 39 home runs from Evan Longoria and Carlos Pena, but they combined for just 12 steals. Shortstop Jason Bartlett (14 home runs) and center fielder B.J. Upton (11 home runs) brought a bit more balance, but were much longer in speed than power, combining for 72 steals.

So, let’s just pretend this is a Note, and that I don’t have to conclude it. Given its overall function, by the way, it’s odd that a conclusion can be faulted for being too short. It’s also odd that I trip up by not writing enough at any juncture of the process, given my general tendency.

1. (HRxSBx2)/(HR+SB)

2. Well, actually, I saw later that teams 24 through 26 were all tied with 132 steals, so it was just the order of the sort that put team 24 above the other two. So we can say that 25 of the top 26 teams in steals among 200-HR teams have had winning records.

3. A summary of this for you. Among 200+ HR teams

1. The correlation between W_PCT + SB is .13.

2. The correlation between OBP + SB is .28.

3. The correlation between OBP + W_PCT is .38.

4. And I remembered how to do that.

5. The Braves had 132 stolen bases, dead last of the 24, but since they laughed at 200 home runs on their way to 300, you had might ask where their combination of home runs and steals ranks by James’ system. I told you at the beginning of this piece that the piece is approximate and short on precise answers, so I don’t know, but the Braves power/speed score was below the 1996 Rockies, at least. It comes in at 184.6; the Rockies, with 201 steals and 221 home runs, score 210.5. Balance, baby. Note that the Braves SB+HR total of 439 is higher than the Rockies total of 422.