How about some relief around here?

As every St. Louis Cardinals fan already knows so well, one of the major disappointments of the 2008 club was its relief corps.

There are plenty of reasons assigned to the problem, including the collapse of Jason Isringhausen, the struggles of the left-handers, relievers being used out of their roles, late-season fatigue, rookie inconsistency and many more.

I have no reason to argue about any of them, yet I wondered whether the starters might have been putting more pressure on the bullpen by not going as deep into games. Here are basic numbers for the starting pitchers over the last six seasons.

Games Total IP Starters IP IP/start
2008 162 1454 955 5.9
2007 162 1435.7 889 5.5
2006 161 1429.7 941.7 5.8
2005 162 1445.7 1048 6.5
2004 162 1453.7 996.3 6.2
2003 162 1463.7 979.3 6.0

While the Cardinals starters last season fell just short of six innings per outing, that mark was actually better than achieved in either of the prior two years. Not surprisingly, the high water marks in recent years of well over six innings were established by the triple-digit win clubs of 2004 and 2005.

There seems nothing obvious that can be pinned on the starters so let’s look at the relievers themselves. The most frequently-quoted measurement is their 31 blown saves, tied for worst in the majors. Here are the other traditional measures.

Save opps Saves Blown Holds
2008 73 42 31 115
2007 45 34 11 82
2006 57 38 19 90
2005 65 48 17 86
2004 73 57 16 102
2003 71 44 27 77

Those looking for the bright side in the pen’s performance can point to the 115 holds, which is the top number in recent years. That offers the major difference between 2008 and 2003, the season closest to last in terms of bullpen ineffectiveness.

Despite all the troubles, the 2008 bullpen was presented with 73 save opportunities, tied for the most during this period. (Though not material, I couldn’t help but notice that 2008 and 2004 would have been almost identical if 15 blown saves last season could have been moved to the saves column instead. Of course, had that happened, the team could have won as many as 101 games and everything would have been different!)

Here is where I decided to dig a bit deeper with the help of the stats available from Baseball-Reference.com.

One of the many splits available is starters vs. relievers. Every pitch, every at-bat all season long was made by one or the other, right? The next split is between relievers in a save situation (SS) and those not in a save situation (NSS). Again, every relief appearance has to be one or the other.

To look at the difference in results achieved between these two situations, I selected one traditional measurement, ERA, and another newer one, OPS+ of the opposing hitters facing the Cardinals relievers.

Save sit ERA SS OPS+ Non-save sit ERA NSS OPS+
2008 5.01 132 3.83 101
2007 3.28 88 4.25 95
2006 3.94 92 4.09 96
2005 2.59 98 3.41 90
2004 1.96 55 3.48 81
2003 5.20 132 4.65 115

A couple of things stand out. Most obvious is the increase in ERA and OPS+ in save situations in 2008 over previous seasons and the disparity between the 2008 results in the non-save situations vs. the save situations. In only 2003 and 2008 were the numbers worse when there was a small lead to protect.

Another observation is that the size of the 2008 gap based on the situation is larger than in any of the other years in this study. Why was that? Why did this pen do ok except when the game was on the line?

Relief IP Rel appearances Save situations SS IP Non-save situations NSS IP
2008 499 506 189 156.3 317 342.7
2007 546.7 516 113 104.3 403 442.3
2006 488 469 127 109.7 342 378.3
2005 397.7 436 158 118 278 279.7
2004 457.3 469 164 142 305 315.3
2003 484.3 461 146 128 315 356.3

The 2008 bullpen made over 500 total appearances for the second consecutive season, but what really catches my eye is the extraordinarily high number of save situations presented the pen at 189.

I initially wondered if that 189 number could have been increased by many short, ineffective appearances that required additional relievers to be deployed. Yet, the high number of innings pitched under those conditions, 156 1/3, tends to discount that.

Next, let’s look at the percent of the save situations and save situation innings compared to the total workload of the bullpen.

Relief IP Rel appearances Save sits SS % appearances SS IP SS % IP
2008 499 506 189 37% 156.3 31%
2007 546.7 516 113 22% 104.3 19%
2006 488 469 127 27% 109.7 22%
2005 397.7 436 158 36% 118 30%
2004 457.3 469 164 35% 142 31%
2003 484.3 461 146 32% 128 26%

Not only were the absolute numbers large, but the 37% of pen appearances occurring in save situations was substantially greater than in either 2006 or 2007. Same with innings pitched.

Could that mean that the 2008 relievers just weren’t able to handle the frequent pressure situations they were placed in? As crazy as it first sounds, would their overall results have been better if they had been presented with fewer save situations to protect?

The 2008 percentages at least did not vary much from 2004 and 2005, though the results in the win column between 2008 and those seasons were quite different.


My final step is to put this together. Remember that inherent in the definition of a save is that at most only one can be had per win. For example, during last season, Cardinals relievers pitched 189 times in save situations over no more than 73 different games. Of those 73, 42 saves were converted and 31 were blown (42%).

(Not to overly complicate, but there could be multiple blown saves in a single game if the bullpen loses the lead more than once.)

That 42% blown save rate in 2008 is the highest in this six-year period. Yet when considering the total number of pitchers that were put in a save situation, the blown save rate of 16% across all of them does not stand out nearly as much compared to other years.

Of course, none of these gymnastics change the bottom line.

Save situations Save opps (~games) Blown saves Blown save % – opps Blown save % – situations
2008 189 73 31 42% 16%
2007 113 45 11 24% 10%
2006 127 57 19 33% 15%
2005 158 65 17 26% 11%
2004 164 73 16 22% 10%
2003 146 71 27 38% 18%

I am going to stop there and ask you, the readers, what you think. Does this work say anything to you? What do you conclude? What if anything might be missing?


Addendum:
At the request of DD17 in the comments that follow, I factored out Jason Isringhausen’s 2008 stats. Interestingly, he appeared in an equal number of save and non-save situations. I guess it is not surprising that his removal helped the club’s save situation ERA, but he actually had a positive impact on the non-save ERA.

What I did find interesting is that Izzy’s save conversion rate of 63%, while not very good, was still better than the rest of the club with him excluded (56%).

In save sits G W L S BS Sv % IP ERA
Team 2008 189 5 13 42 31 58% 156.3 5.01
Izzy 21 1 4 12 7 63% 17.7 10.70
Team w/o Izzy 168 4 9 30 24 56% 138.7 4.28
In non SS G W L S BS Sv % IP ERA
Team 2008 317 17 18 0 0 0 342.7 3.83
Izzy 21 0 1 0 0 0 24 2.16
Team w/o Izzy 296 17 17 0 0 0 318.7 3.95

While I redid the team totals, I am not comfortable with them. To do it properly, it would have required me to make an assumption about whether or not another reliever could have converted the saves that Izzy couldn’t. Instead, I just took out his numbers.

Save sits Save opps (G) Blown Blown Sv % Blown save sit %
2008 total 189 73 31 42% 16%
2008 w/o Izzy 168 73 24 33% 14%