Let’s talk about the pitchers who do a great job of holding runners on

AKA the Mark Buehrle award. 59 SB against, 81 CS...in 16 years. The man was the master of holding guys on.

The man was the master of holding guys on.

In Terry Mulholland's 20-year career, only 85 men even tried to steal on him. 35 of them were successful.

Opposing basestealers had a 34% success rate against Kirk Rueter (34/99).

When statcast provides a number like Griffin Canning has an 8 in this metric, is that the same as 10 runs is about 1 WAR? How do the statcast rankings translate into WAR? Are they all in the same scale ?

AKA the Mark Buehrle award. 59 SB against, 81 CS...in 16 years. The man was the master of holding guys on.

It's amazing that you could figure this out without the benefit of this advanced metric.

Per the article, the metric is "base advances allowed or prvented." They say a base advance is worth only 0.2 of a run (seems a bit low) so Canning prevented a couple of runs. Elsewhere they seem to suggest a base prevented is about 1/3 of a run which is what I had thought. Either way, best to non-Noah/Ottavino worst looks to be a range of about 25 bases or 5-8 runs.

Frankly the conversion numbers are all over the place so maybe it's very contextual? In discussing Cueto (a RHP as best ever?), they say he has prevented 30 based, equating to about 8 runs or 0.27 runs/base advanced. But in discussing Jon Lester, they say he gave up 29 more bases which is about 4 runs so about 0.14 per base. I'll guess that's about Cueto also getting credit for some outs -- i.e. the break-even point between SB and CS.

There's no reason to have epxected this to be a big deal. In general, what Cueto (and Rueter and Mulholland and ...) accomplish is helpful but it doesn't produce a lot of outs because runners become conservative very quickly.

Cueto's 2012 must be in the running for greatest of all-time. Teams tried 10 steals, successful just once (a steal of home oddly enough) plus he picked them off 9 times. In three of Cueto's first 5 seasons, he had more pickoffs than SBs allowed. He aded 4 more such seasons. For his career, 36 pickoffs (plus 9 PoCS) vs 41 steals. I am stunned a RHP can do that.

They say a base advance is worth only 0.2 of a run (seems a bit low)

Stealing 3B is lower value than 2B, maybe some of the discrepancy is 2B steals vs all steals. Also a 2B steal has some value in avoiding a DP, maybe the context of that (not applicable with 2 outs) is factoring in there somewhere.

You can do better and simpler. Here's a method adapted from my Stolen Bases Profit method from my books in th e 1990s.

There is a break-even point in base stealing, usually figured to be about 65.5% or so. Decide on one, let's say that 65.5%. Compute 1 - 65.5 = 34.5% You break even if the successes are 65.5 / 34.5 of the failures. So, you can get Stolen Bases Profit expressed in bases, by SB - 65.5 / 34.5 CS = SBProfit. That answers the question of how good you were at foiling the attempts that were made. You can then do the work to figure out what the expected value of attempts made against you should have been and compare that to actual attempts. That gives you the volume component.

You then have to figure out some way of assigning percentages of the profit to the pitcher and to the catcher. But you have to do that regardless of what method you use.

This is simple and get the job done without any additional steps.

You can convert to run value instead of base value at any point in the process.

Where do pickoffs show up? If they are under "caught stealing" how does one break that out from a catcher's throw?

Traderdave: as I recall if the runner on first is going back to first on a pickoff it doesnt show as a CS. If the runner breaks for second then its considered a CS. For a runner like Maury Wills you might want to add 5% to his official SB rate to account for that some of those CS are really pickoffs (see link to Wills below adjusting Wills SB rate of 74% to effectively 79% by subtracting the picksoffs). But you might to do more research on that, I know they changed the rule on that about 1970 so I maybe missing something.

Walt: Hi Walt.

They say a base advance is worth only 0.2 of a run (seems a bit low) so Canning prevented a couple of runs. Elsewhere they seem to suggest a base prevented is about 1/3 of a run which is what I had thought.

0.2 is probably as good an estimate as we can make given that there are issues with high leverage situations, where we are in the batting order, possible throwing errors (probably under emphasized) etc. You can look at tango's base/out run expectancy matrix here and do some quick calculations:

http://www.tangotiger.net/re24.html

I would look at those tables dealing with the chance of scoring a run. You can see the gain of 2b is on the low side. Assuming base stealing occurs at the same rate as 0, 1 or 2 outs (it does vary but not a lot as I recall), the gain of 2b is less than 0.2 runs. However stealing 3rd with 0 or 1 out is worth 0.25 runs so you do have that.

this guy: Rybarcyk did some more work on this and also comes up with similar values near about 0.2. He also comes up with a break even rate of 73% or so which seems a bit high. But a lot of interesting stuff about pitcher times and catcher pop times and allocating value which some primates are asking about. Good comments too:

https://tht.fangraphs.com/stolen-base-attempts-an-algorithm-for-allocating-run-value/

on the more positive side Dave Smith did a deep dive on Maury Wills's 3 best seasons and comes up with much more value for SB. You have to read through it but he's saying that WIlls netted 58 SB over this period resulting in 35 runs! Resulting in 7 wins! Much more than our models would tell us. He accounts for this in several ways: Wills stole more in close games (see table 5, he's about 20% more likely to run in a one run game), the lineup behind him, it was a low offensive era, and also wills produced a lot of balks, WP, PB, throwing errors (see table 8) not sure the previous literature addresses that. Anyhow here is the link:

https://sabr.org/journal/article/maury-wills-and-the-value-of-a-stolen-base/

Stealing 3B is lower value than 2B, maybe some of the discrepancy is 2B steals vs all steals.

Karl: I dont believe thats correct. At least with 0 or 1 out the chance of scoring one run increases about 25%. See the base/out tables I linked to above.

Or maybe I am missing something, what do you think?