Cobb's stolen-base percentage (and value)

[Wee Willie]

Regarding Cobb's stolen bases: Has anyone made an attempt to estimate what his SB% would've been, based on the caught-stealing data we DO have? Just using the seasons where CS data is available, his SB% is 64%.
I'm thinking that because league averages were so low, I think his career percentage wouldn't be much higher - high 60's at the most, but more likely around 66-67%.

Regardless, his percentage wouldn't be very high, historically. The major league average the last few years has been about 70%. Even in the 80's, when teams ran considerably more than today, league-wide SB percentages were in the high 60's. The only seasons we have league-wide CS data for Cobb's career is 1914-15, and 1920-28. In both 1914-15 and from 1920-28, the league SB average was about 56%. I imagine that throughout Cobb's career, the league average was also about 56-57%. I realize there were more hit-and-runs, and an out-on-base may not have been quite as costly as is would be today. But 56% seems to be a pretty low figure, regardless of the run-scoring environment. It's hard for me to estimate how much credit to give him for SB's when I make a reasonable estimate of his CS numbers.

If anyone else has ideas on this, I'd love to hear them.

[JRB]

Wee Willie; Nice to read one of your posts again.

As you mentioned they ran a lot more hit and run plays in that era. Wouldn't a fairly sizable percentage of the caught stealings in those days be the result of a busted hit and run play?

To my slow way of thinking, getting a caught stealing in those circumstances doesn't reallly detract from a players value as a baserunner. Remember, there were also a whole lot of extra bases gained from the hit and run plays that players don't really gain a benefit for in their baserunning stats, so why should they get docked for a CS if the batter screws up? Perhaps someone with far greater knowledge than I can come up with a reasonable estimate as to how often they attempted hit and run plays in that era as opposed to later eras and then add some portion of the percentage difference to the SB% totals of the players from Cobb's era, to come up with a comparative equivilance in value.

c JRB

[Wee Willie]

Wee Willie; Nice to read one of your posts again.

As you mentioned they ran a lot more hit and run plays in that era. Wouldn't a fairly sizable percentage of the caught stealings in those days be the result of a busted hit and run play?

To my slow way of thinking, getting a caught stealing in those circumstances doesn't reallly detract from a players value as a baserunner. Remember, there were also a whole lot of extra bases gained from the hit and run plays that players don't really gain a benefit for in their baserunning stats, so why should they get docked for a CS if the batter screws up? Perhaps someone with far greater knowledge than I can come up with a reasonable estimate as to how often they attempted hit and run plays in that era as opposed to later eras and then add some portion of the percentage difference to the SB% totals of the players from Cobb's era, to come up with a comparative equivilance in value.

c JRB

Hi, JRB. The way I see it, any caught stealing takes away at least a little value. If you're caught because of a busted hit-and-run, either the manager or the hitter has taken away a little of your value (rather than you taking away your own value). If Cobb is caught stealing, then his baserunning ability to go from 1st to 3rd on a single is wiped out if the guy at-bat hits a single after the CS. However, one probably shouldn't be penalized as much for a CS in those circumstances.

Without knowing a rough percentage of how many CS were busted hit-and-runs, perhaps the baseline to assess Cobb's value as a basestealer should be somewhere between the estimated league SB average (about 56% for Cobb's time) and the standard break-even SB percentage that's used in Runs Created formulas. Brett may have some light to shed on this.

[brett]

Wee Willie; Nice to read one of your posts again.

As you mentioned they ran a lot more hit and run plays in that era. Wouldn't a fairly sizable percentage of the caught stealings in those days be the result of a busted hit and run play?

To my slow way of thinking, getting a caught stealing in those circumstances doesn't reallly detract from a players value as a baserunner. Remember, there were also a whole lot of extra bases gained from the hit and run plays that players don't really gain a benefit for in their baserunning stats, so why should they get docked for a CS if the batter screws up? Perhaps someone with far greater knowledge than I can come up with a reasonable estimate as to how often they attempted hit and run plays in that era as opposed to later eras and then add some portion of the percentage difference to the SB% totals of the players from Cobb's era, to come up with a comparative equivilance in value.

c JRB

I think that what we have to say is that the stolen base stats are not the right place to look for Cobb's (or other deadball stealer's) value on the bases.

Currently we know that a top-top baserunner today will gain an extra 5 runs per season on the league average in taking bases that are not registered in the stat line. Such a player may even gain 8 runs in a spectacular season.
5 runs is about equivlent to 22 extra steals. 8 would be 35.

What I have to assume is that in Cobb's day, he was gaining proportionately MORE of these bases as indicated by the fact that he was probably caught in hit and run situations.

I suspect that Cobb may have doubled (or more) the additional bases of a top baserunner today, perhaps gaining an equivalent of 33 steals worth of additional bases a year.

I think most estimates are that he stole 892 and was caught about 350 times. If you give him another 660 "steals" over a 20 year career, that is about 1552 The success rate in these extra steals would be nearly 100% (based in part on the fact that current players, even bad runners are rarely caught trying to take an extra base-something like 20% of the time, and secondly based on the fact that when he WAS caught in those situations it probably got recorded as a CS).

Lets say a modern great steals 800 bases and is caught 200 times, and also picks up 22 extra bases a year over about 15 full seasons. That guy is gaining 1130 and getting caught about about 260 times on ALL attempts to take an extra base (not those recorded such as stretching a double into a triple).

Compare that to Cobb perhaps going 1552 and getting caught maybe a little more than 350 and he's great.

Also, I think that steals were relatively even more valuable in that period because they were usually done to take a player in a situation where he was unlikely to score (first base and 2 outs) and give a guy a chance to hit a single. A player would be taking a high situational value extra base, and if he's caught, consting a low situational value out. I think that steals probably occur in almost neutral overall strategic situations.

Also with more grounders potentially poked between first and second, the steal probably had a greater impact on improving the hitting of the player at the plate.

[Bill Burgess]

Just a kernel or two. In the Los Angeles Times, ran this little notice, on January 26, 1913, Section VII, page 7.

In 1912, the following runners led the AL in caught stealling.

1. Ty Cobb - 34
2. Harry Lord - 33
3. Clyde Milan - 31
4. Tris Speaker - 28
5. Eddie Collins - 22
-----------------------------------
Another item, is that whenever the Tigers were way ahead, Cobb ran wild, to discourage/demoralize the opposing team. So, when he was caught in those situations, he was not costing his team the same, standard of potential lost outs that lose games. So, his CS rate was less detrimental to his team than his rate would normally indicate.

We do know that his CS rate in cases of steals of home was 54 successful steals of home in 98 attempts.

Of further interest, 23 of his successful steals of home, came when he was the lead runner of a double steal, and 6 came when he was the lead runner of triple steals.

------------------------------------
Here are some CS percentages for Cobb, Speaker, Collins, Milan.
The official records of Total Baseball do not list CS for 1912, but a piece in the Los Angeles Times does list them. (LA Times, January 26, 1913, pp. VIII 7) Got it via Proquest.

--------Cobb------------Speaker--------Collins-----------Milan
Year---SB------%--------SB-------%-----SB-----%--------SB-----%
1912---61-34-64.2%-----52-28-65.0%----63-22-74.1%-----88-31-73.9%
1914---35-17-67.3%-----42-29-59.1%----58-30-65.9%-----38-21-64.4%
1915---96-38-71.6%-----29-25-53.7%----46-30-60.5%-----40-19-67.7%
1916---68-24-73.9%-----35-27-56.4%----40-21-65.5%-----34-21-61.8%

[Wee Willie]

Brett, I agree with a lot of your analysis, but I think the 350 CS estimate for Cobb's career is far too low. That would give him a .718 total percentage. Thanks to the inclusion of 1912 data that Bill was kind enough to furnish, we know that in seasons where we do have CS data, Cobb had 387 SB's and 212 CS's - good for a .646 percentage. A 350 career CS estimate would mean that Cobb's total SB% in years where we don't have CS data would be .785.

Do you think maybe the estimate actually was 450 and not 350? A 450 estimate would put his career SB% at .665, which I think is much more likely than .718.

[Bill Burgess]

Did anyone notice that Ty's CS rate was improving. I bet it improved until 1917-19.

[brett]

Brett, I agree with a lot of your analysis, but I think the 350 CS estimate for Cobb's career is far too low. That would give him a .718 total percentage. Thanks to the inclusion of 1912 data that Bill was kind enough to furnish, we know that in seasons where we do have CS data, Cobb had 387 SB's and 212 CS's - good for a .646 percentage. A 350 career CS estimate would mean that Cobb's total SB% in years where we don't have CS data would be .785.

Do you think maybe the estimate actually was 450 and not 350? A 450 estimate would put his career SB% at .665, which I think is much more likely than .718.


Somebody (I thought it was Bill) claimed to have record of Cobb's actual CS total for his career through some method, and it was in the mid to upper 300s.

[Wee Willie]

Somebody (I thought it was Bill) claimed to have record of Cobb's actual CS total for his career through some method, and it was in the mid to upper 300s.

I see. Based on the sampling of what we do have, I think that the method that was used underestimates it by quite a bit.

[Bill Burgess]

Somebody (I thought it was Bill) claimed to have record of Cobb's actual CS total for his career through some method, and it was in the mid to upper 300s.
Close, but not quite. What you're thinking of, is my work in estimating Ty's strike out rates.

Just as we don't have his CS data for many seasons, so we also do not have his strike outs, prior to 1913, due to the leagues not keeping documentation of that. But I was able to create a curve, based on his SO rates from 1913 on. Simply used his rates and worked backwards, adjusting slightly.

The NL started maintaining batter SOs in 1910, and the AL in 1913. The NL also kept batter SOs, 1871-1896. Why they stopped is anyone's guess.

Anyway, here was that long ago post.

http://baseball-fever.com/showpost.php?p=296551&postcount=27

Hope this clarifies the confusion. How'd you like my curve?

[brett]

Close, but not quite. What you're thinking of, is my work in estimating Ty's strike out rates.

Just as we don't have his CS data for many seasons, so we also do not have his strike outs, prior to 1913, due to the leagues not keeping documentation of that. But I was able to create a curve, based on his SO rates from 1913 on. Simply used his rates and worked backwards, adjusting slightly.

The NL started maintaining batter SOs in 1910, and the AL in 1913. The NL also kept batter SOs, 1871-1896. Why they stopped is anyone's guess.

Anyway, here was that long ago post.

http://baseball-fever.com/showpost.php?p=296551&postcount=27

Hope this clarifies the confusion. How'd you like my curve?

Bill, it looks like good work. The one thing I checked on was league striekout rates, and I noticed that they stayed pretty consistent during the period you are extrapolating to. If league rates had changed drastically, I might multiply your values by the relative rate for the league.

I also looked to see if his walk rates were paralleling his strikeout rates. I don't really see a pattern though.

I have always wondered about his 1915 walk total though. It may have been the best Tigers team during Cobb's career and he happened to have a singular spike in walks. Cobb was clearly the MVP in 1915 in my book, and while some people will focus on OPS+ I think that it ranks among his very best seasons. And even with the high CS total on paper, his steals make his 185 OPS+ into about 205. I always thought that '09, '15 and '16 get overlooked in his resume.

[Bill Burgess]

Cobb was clearly the MVP in 1915 in my book, and while some people will focus on OPS+ I think that it ranks among his very best seasons. And even with the high CS total on paper, his steals make his 185 OPS+ into about 205. I always thought that '09, '15 and '16 get overlooked in his resume.
As a Cobb researcher, I too have struggled to identify his best seasons, in order. It's not as simple as it appears. Here are some numbers I've thrown together to assist myself.
-----------------------------------------

Relative BA------------Relative SLG.-------OPS
1910---1.58%-----------1917--1.72%---------1917--209
1916---1.55%-----------1910--1.65%---------1910--206
1912---1.54%-----------1911--1.64%---------1912--200
1909---1.54%-----------1912--1.64%---------1911--196
1917---1.54%-----------1909--1.58%---------1909--194
1911---1.53%-----------1918--1.57%---------1913--194
1913---1.51%-----------1913--1.54%---------1918--193
1918---1.50%-----------1914--1.50%---------1914--190
1914---1.48%-----------1908--1.49%---------1915--185
1915---1.48%-----------1916--1.44%---------1916--179
1919---1.43%-----------1907--1.44%---------1922--170
1907---1.41%-----------1925--1.43%---------1925--170
1922---1.40%-----------1915--1.42%---------1908--169
1908---1.35%-----------1921--1.42%---------1907--169
1921---1.33%-----------1922--1.39%---------1921--166

[Wee Willie]

As a Cobb researcher, I too have struggled to identify his best seasons, in order. It's not as simple as it appears. Here is some numbers I've thrown together to assist myself.

I'd probably go with 1917 as Cobb's best. Best OPS+, best EqA, easily the best WARP, close to the best Win Shares.

[JRB]

Bill;

Thanks for that excellent list you posted. It's amazing how many great seasons Cobb had where he was able to separate himself so far above the league average.

Even though his OPS+ was higher in 1917, I would go with 1911 as Cobb's best all around year. Cobb posted career bests that year in batting average (.420), base hits (248), runs scored (147), runs batted in (127), doubles (47), triples (24) and total bases (367). Needless to say he lead the league in all those categories, and also in stolen bases, having his second best year in that category with a whopping total of 83. Cobb also lead the league in double plays and putouts. Cobb's 1911 season would be on my short list of greatest all around seasons ever, and may have been the best ever pre-Ruthian era season.

c JRB