“This is not just a day off. Good baserunning is just as important as good hitting to win baseball games.”–Lou Piniella, Mariners manager, on sitting infielder Jeff Cirillo following baserunning errors in 2002
When last we were together, we added up the various baserunning metrics we’ve been formulating all summer to come up with a total number of theoretical runs contributed on the bases for individual players. This included runs from advancing on ground and air outs, advancing on hits, and runs contributed from stolen base attempts (and pickoffs).
What we found was that in a single season a baserunner may contribute from 8-10 runs over and above what would be expected, or may forfeit an equal number, meaning that the spread between the best and worst runners was on the order of two wins. Over the course of the six seasons that were examined (2000-2005), a runner may contribute up to 25 runs and give up about 20. Using those raw numbers, we christened Carlos Beltran the best baserunner of the past six years at +25.44 runs, and Jorge Posada as the worst at -20.39.
However, we left unexamined two things-team-level performance, and what these numbers tell us about baserunning in general. As Socrates taught us, “the unexamined life is not worth living”, so we’ll delve into those topics this time around.
Pieces of the Pie
All fans love to grouse at the tube when their team has runners gunned down on the bases, picked off, doubled up, and thrown out stealing. But which fans have suffered the most (and the least)? The following table lists the top and bottom 20 teams in total number of runs contributed on the bases over the past six seasons.
Year Team Opps EqGAR Opps EqAAR Opps PO CS EqSBR Opps OA EqHAR Total 2004 SLN 319 2.59 282 4.46 159 4 51 -5.76 447 6 9.19 10.47 2000 KCA 326 2.68 307 4.33 156 4 39 -0.68 461 8 3.48 9.81 2001 SEA 297 -1.06 320 2.80 212 7 49 -0.83 421 6 8.67 9.57 2005 NYN 289 2.40 246 -2.57 194 8 48 2.60 372 5 6.86 9.28 2005 ATL 313 -1.32 269 0.61 125 3 35 -0.09 392 4 7.99 7.19 2000 CHA 330 0.10 286 2.23 163 7 49 -3.35 390 6 7.70 6.67 2004 ANA 325 5.08 268 0.57 187 6 52 -0.98 489 13 1.96 6.63 2001 COL 354 3.13 251 0.74 188 10 64 -9.03 379 8 11.47 6.31 2000 COL 315 5.35 299 1.93 191 7 68 -12.45 445 7 11.28 6.12 2004 MON 316 2.54 236 -5.13 147 3 41 -0.50 373 4 9.16 6.07 2001 TEX 252 -1.33 305 4.24 130 8 40 -1.20 372 9 4.34 6.05 2005 TEX 244 -1.71 276 3.58 84 3 18 1.38 402 6 2.68 5.94 2003 OAK 289 -2.49 299 2.26 61 2 16 -0.25 401 11 5.23 4.75 2005 TBA 283 -1.66 277 -2.49 201 7 56 -5.33 447 6 12.42 2.93 2005 SLN 308 3.94 253 4.46 118 4 40 -8.83 448 6 2.62 2.19 2003 SLN 325 2.96 301 2.39 107 3 35 -6.60 478 11 2.74 1.50 2001 OAK 238 -0.95 275 2.58 98 4 33 -5.41 341 4 4.93 1.15 2004 ATL 299 -1.35 228 4.52 119 1 33 -1.72 428 8 -0.51 0.94 2003 NYN 322 2.14 240 3.29 105 6 37 -7.22 379 7 2.68 0.89 2000 CIN 304 2.43 295 4.49 139 7 45 -5.91 382 7 -0.45 0.56 ------------------------------------------------------------------------------------------------------ 2001 NYN 259 -1.16 297 -8.34 120 7 55 -20.72 328 12 -2.42 -32.64 2001 SFN 267 0.57 281 -4.68 103 5 47 -13.65 332 11 -12.23 -29.99 2002 MIL 339 -1.21 241 -4.92 155 13 63 -17.51 317 12 -4.51 -28.15 2001 BOS 266 -1.07 249 -3.03 83 4 39 -12.77 356 13 -10.39 -27.27 2005 WAS 326 -0.18 237 0.58 97 10 55 -22.49 392 7 -4.42 -26.50 2004 BOS 255 -4.67 299 1.12 100 4 34 -6.60 484 12 -14.80 -24.95 2001 MON 325 0.08 252 -2.74 154 8 59 -16.85 310 6 -3.74 -23.24 2001 PIT 299 3.61 255 -4.43 170 9 82 -26.71 285 8 5.72 -21.81 2005 LAN 312 -3.00 255 0.19 95 4 39 -9.71 396 12 -9.25 -21.78 2000 BOS 282 -0.96 324 -3.86 78 5 35 -9.97 381 10 -6.73 -21.52 2001 MIL 269 2.41 222 0.97 108 8 44 -12.72 311 13 -12.07 -21.41 2005 SDN 298 0.03 297 -1.23 153 12 56 -12.36 427 13 -6.51 -20.07 2002 CHA 254 -1.48 307 -3.35 116 13 44 -12.54 329 12 -2.60 -19.98 2002 SDN 312 -0.96 254 -3.91 114 5 49 -15.11 337 10 0.00 -19.96 2001 CIN 302 -1.09 231 1.04 161 8 62 -16.18 331 12 -2.22 -18.45 2005 PHI 302 -1.65 304 -2.37 144 7 34 2.17 447 18 -16.35 -18.20 2003 CHN 301 -2.22 273 1.50 110 8 39 -6.97 409 15 -10.33 -18.02 2002 ARI 304 -1.13 281 1.25 142 6 52 -12.35 340 11 -5.57 -17.79 2002 ATL 320 -2.81 272 -1.57 118 4 43 -11.67 337 8 -1.41 -17.45 2004 SEA 333 0.00 260 0.46 152 8 50 -8.23 488 10 -9.06 -16.84
First, it’s apparent that looked at this way, teams don’t really gain much from all their work on the bases. The top team, the 2004 Cardinals, came out with just ten and half runs to the good, or the equivalent of about a win. In fact, of the 180 teams in the study, only 23 accumulated positive totals. A quick glance at the list reveals that this is the case because the aggregate values for EqSBR are almost always negative and can be quite large in comparison to the other metrics. The reason? As a general-purpose strategy, stealing bases turns out to be a high-risk and low-reward endeavor as evidenced by the matrix showing the positive and negative run values associated with stolen base attempts in the previous column. This means that in order to accumulate runs, an individual or team is required to be successful a very high percentage of the time. Only 7 of the 180 teams came out on the plus side in EqSBR (as shown below), with the average team losing a bit more than eight runs.
Year Team Opps EqGAR Opps EqAAR Opps PO CS EqSBR Opps OA EqHAR Total 2004 NYN 320 -2.19 240 -1.79 132 3 26 4.47 360 11 -5.79 -5.30 2005 NYN 289 2.40 246 -2.57 194 8 48 2.60 372 5 6.86 9.28 2005 PHI 302 -1.65 304 -2.37 144 7 34 2.17 447 18 -16.35 -18.20 2004 PHI 315 0.77 269 -1.23 129 3 30 1.95 425 12 -9.30 -7.81 2005 TEX 244 -1.71 276 3.58 84 3 18 1.38 402 6 2.68 5.94 2000 CLE 280 -1.67 290 -6.41 143 2 36 0.44 395 11 -2.43 -10.07 2004 TBA 279 -0.30 293 0.31 175 5 47 0.40 398 6 0.06 0.47
The recent versions of the Mets and Phillies, along with these other three clubs, are the only ones to keep their heads above water in this regard, by minimizing their caught stealing and pickoffs, and by selecting higher percentage running opportunities. Teams that run a great deal, like the Marlins of 2000-2003, usually do poorly and often come out with EqSBR values in the -5.00 to -10.00 range or worse. The poorest team was the 2001 Pirates, who in 170 opportunities were caught stealing a remarkable 82 times, and picked off another 9 times, which amounted to an EqSBR of -26.71.
That doesn’t mean that stealing bases is always a bad idea. On the contrary it’s a strategy that is sometimes the best option. It’s knowing when to take those opportunities and with whom to take them that is the trick. In fairness (and as pointed out in a previous column), these raw stolen base numbers also include busted hit-and-runs, since there is no way to differentiate based on the play-by-play data, and that automatically drags down every team’s numbers.
Getting back to the first table above, on the negative side, the 2001 Mets were at -32.64 runs on the strength of an EqSBR of -20.72. This points out that the range for EqSBR is on the order of +10 to -30 runs, repeating the point that the upside is clearly not as high as the downside is low. At first glance, the implication here is that even if a team works to inculcate good baserunning in general-as suggested by Lou Piniella at the start, or the attempt to do so earlier this season when the Rockies benched Matt Holliday and Jason Smith after carelessness on the bases-the advantage can easily be wiped away by attempting to steal bases in low percentage and poor leverage situations.
Attentive readers may have noticed that EqSBR is calculated slightly differently than the other metrics here. Remember that EqGAR, EqAAR, and EqHAR all are measured in terms of runs contributed above what would be expected, given the specific opportunities that each team encountered. In contrast, EqSBR is simply an accounting of total runs, and is not base-lined by looking at what the average team would have done with the same opportunities. For most of the analysis, this is exactly what we want, since stolen base attempts are purely at the discretion of the offense. But if we wish to get a feel for the total magnitude of these metrics, we can instead simply show the total number of runs contributed in each area, and reformulate our top and bottom 20 list in the following table where, for example, GAR simply stands for Ground Advancement Runs.
Year Team Opps GAR Opps AAR Opps PO CS EqSBR Opps OA HAR Total 2000 KCA 326 26.75 307 46.66 156 4 39 -0.68 461 8 76.54 149.28 2001 SEA 297 23.04 320 46.11 212 7 49 -0.83 421 6 78.83 147.16 2000 COL 315 29.32 299 45.68 191 7 68 -12.45 445 7 74.96 137.51 2004 SLN 319 26.17 282 40.09 159 4 51 -5.76 447 6 75.75 136.25 2004 SDN 315 26.10 333 39.12 78 3 28 -6.16 490 9 74.70 133.77 2004 ANA 325 30.19 268 28.44 187 6 52 -0.98 489 13 74.57 132.22 2003 BOS 260 15.49 318 42.96 126 4 39 -4.76 475 7 76.75 130.43 2003 KCA 304 22.54 284 35.62 165 9 51 -3.54 490 10 72.26 126.88 2005 TBA 283 18.26 277 32.05 201 7 56 -5.33 447 6 81.32 126.31 2003 OAK 289 20.32 299 33.76 61 2 16 -0.25 401 11 72.16 125.99 2001 COL 354 31.04 251 30.64 188 10 64 -9.03 379 8 73.30 125.95 2003 TOR 283 19.52 324 35.86 63 2 27 -7.63 463 9 76.84 124.58 2000 CHA 330 20.81 286 37.57 163 7 49 -3.35 390 6 69.27 124.30 2005 OAK 259 14.38 320 28.15 54 1 23 -5.79 478 3 86.70 123.44 2005 ANA 321 33.55 265 24.34 218 5 62 -10.44 474 10 75.66 123.11 2004 COL 351 26.83 249 24.28 81 4 37 -10.34 445 10 82.27 123.03 2003 SLN 325 25.08 301 37.03 107 3 35 -6.60 478 11 67.34 122.86 2003 MIN 338 25.46 280 32.64 145 10 54 -8.88 461 8 72.78 121.99 2002 ANA 306 26.93 357 42.25 169 5 56 -8.71 437 8 60.10 120.57 2004 PHI 315 23.93 269 32.27 129 3 30 1.95 425 12 61.89 120.04 ------------------------------------------------------------------------------------------------------ 2002 MIL 339 24.05 241 21.53 155 13 63 -17.51 317 12 36.63 64.71 2001 NYN 259 16.21 297 19.67 120 7 55 -20.72 328 12 51.66 66.82 2001 MIL 269 20.84 222 25.44 108 8 44 -12.72 311 13 35.43 68.99 2001 MON 325 21.49 252 28.00 154 8 59 -16.85 310 6 37.80 70.44 2001 BOS 266 18.03 249 27.47 83 4 39 -12.77 356 13 37.72 70.45 2001 PIT 299 24.37 255 19.77 170 9 82 -26.71 285 8 53.90 71.32 2001 LAN 274 16.64 257 25.14 134 6 48 -8.60 292 13 38.40 71.59 2002 PIT 272 19.35 267 25.24 132 5 54 -12.50 294 5 39.83 71.92 2002 CLE 292 16.73 228 25.89 91 2 39 -11.20 319 8 41.02 72.44 2002 OAK 247 13.46 273 24.00 67 4 24 -5.24 351 7 41.57 73.79 2005 LAN 312 19.30 255 22.32 95 4 39 -9.71 396 12 44.75 76.66 2001 SFN 267 18.98 281 32.20 103 5 47 -13.65 332 11 40.70 78.23 2002 SDN 312 21.26 254 21.81 114 5 49 -15.11 337 10 50.27 78.23 2001 CIN 302 23.06 231 24.51 161 8 62 -16.18 331 12 46.86 78.24 2002 CHA 254 17.61 307 32.58 116 13 44 -12.54 329 12 42.25 79.90 2002 BAL 285 19.81 266 30.96 161 5 53 -7.11 297 11 37.39 81.04 2002 HOU 268 14.73 235 22.62 102 6 33 -7.63 365 13 51.80 81.52 2000 TBA 292 19.25 282 26.82 135 2 48 -7.65 352 11 43.95 82.36 2002 ATL 320 19.94 272 29.72 118 4 43 -11.67 337 8 44.57 82.56 2000 TOR 274 15.82 267 23.83 128 6 40 -4.01 385 10 47.29 82.94
One could interpret the table above as indicating that the 2001 Royals cashed in approximately 150 runs with their baserunning, while the 2002 Brewers got only about 65. Unfortunately, that’s not quite right. A better interpretation is that the aggregate of the actions by Royals baserunners in 2001 put their team in a position to score 150 more runs than they would have otherwise had they never advanced on the bases, tagged up, attempted a stolen base, or been picked-off. In other words, had they played absolute station-to-station baseball. The actual number of runs individual teams did gain from their baserunning is not what is being measured since the run values used in the calculations are taken from the overall Run Expectancy matrix for the period and therefore simply help us model the impact of baserunning actions.
This does, however, give us the opportunity to sum each run value across all teams to get a feel for how important each metric is in its contribution to the running game. When this is done, you can chart the results to show the contribution that each makes to the total picture:
EqSBR is not in the total pie, since it turns out to be negative to the tune of 7%. The interesting aspect of this graphic is that while air advancement makes up a greater percentage of the pie than does ground advancement, air advancement is both more variable (and hence less impacted by skill) and has a smaller range which makes it more difficult for teams to capitalize on their opportunities. Advancing on hits is where the real opportunity lies, since it is a repeatable skill to a larger degree, and it provides the chance for teams to increase their run-scoring. That said, it should be remembered that the contribution of the running game pales in comparison to hits and walks, Piniella’s quote above notwithstanding, and according to the analysis above, using the totals across all teams accounts for only 13% of the offensive output over the past six years.
Getting Older and Smarter?
Throughout this series, we’ve seen that players that do better in these metrics are generally faster than those who don’t. During my chat session earlier this week one reader suggested correlating these metrics with BP’s Speed Scores in order to find the speed threshold at which risk-aversion becomes the optimal strategy. Although I didn’t appreciate it at the time, doing so would allow us to see which metrics are most dependent on pure speed, and which on other attributes such as judgment. Alas, time pressures did not permit that analysis to be done this week.
Another way to do this perhaps, albeit a little less directly, is to look at how aging affects the metrics based primarily on the assumption that speed decreases with age. To do this, we can simply group all the players by age and compare the rate statistics for each measure. This was done for ages 21 through 39 (those are the ages between which there were more than 1,200 opportunities), and is presented in the following three graphs with the best-fit linear regression line thrown in for good measure.
As you can see, all three of these metrics show a downward slope, so indicating that the younger a player is, generally speaking, the better they’ll do. Looking more closely, you’ll also notice that the slope of the line for EgHAR (advancing on hits) is greater than that for the other two, with EqGAR (advancing on grounders) coming in second followed by EqAAR (advancing on flyouts). In fact, the correlation coefficients with age for the three measures pictured above are:
Metric r EqHAR 0.86 EqGAR 0.80 EqAAR 0.48
This provides support for the view that speed is more important when advancing on hits than when advancing on ground and air outs. This makes sense intuitively, since players have more opportunities to take multiple bases on hits, which can better exploit their speed. As mentioned previously, the correlation between EqAAR and age is lower because there’s simply more variability inherent in that metric. On the other side, it might also indicate that judgment is more important when advancing on ground and air outs, and is therefore a skill less affected by the affects of aging.
We don’t see quite the same kind of smooth trend for EqSBR as shown below using EqSBR per opportunity (since we have no rate statistic for EqSBR):
The likely reason for this bumpy track is that the combination of running only when the odds of success are good and running in higher leverage situations play a larger role as a player ages. The increase after age 32 likely reflects a selection bias, as some players that remain in the league past that age continue to be productive (at least in terms of the league as a whole), if not prolific, base stealers, while the guys that can’t run, don’t.
Under the Tag
In an early article in this series, we began with a quote from Bill James from the 1984 Baseball Abstract, where he lamented the fact that baserunning would be perfectly measurable if we simply tracked game events differently from the start. The last 20 years have seen revolutions in the collection of play-by-play data and in our ability to analyze that data. The happy result is that baserunning can be explored in more depth than ever before.
Thank you for reading
This is a free article. If you enjoyed it, consider subscribing to Baseball Prospectus. Subscriptions support ongoing public baseball research and analysis in an increasingly proprietary environment.
Subscribe now