Double steals - an underused strategy.

Discuss all things Cardinals Baseball
Post Reply
jim
Red Lobster for the seafood lover in you
Posts: 50608
Joined: May 1 06, 2:41 pm

Double steals - an underused strategy.

Post by jim »

I found this article in BP concerning double steals interesting. Based on the breakeven rates needed, and the actual success rates, double steals appear to be an underused strategy.
The Strategy

From a traditional Run Expectancy standpoint, calculating the breakeven percentage for a successful double steal attempt is straightforward. For example, for the period from 1999 through 2002 (the highest offensive environment represented in our data set) the breakeven percentages for a double-steal attempt when attempting to maximize runs with runners on first and second (which make up 79% of double-steal attempts since 1970) assuming the lead runner is put out are:

Outs BE%
0 .639
1 .558
2 .735

In leaner offensive times like those that persisted during much of the rest of the period since 1970, the breakeven percentages would be lower with less than two outs, since making outs on the bases would not have been as costly. For example, in 1980 the breakeven percentages fall to .600 and .530 with zero and one out, and raises slightly to .778 with two outs.

We can then compare this percentage to the actual results shown below for the entire period.

Outs Succ Att Pct
0 638 1127 0.566
1 1587 2258 0.703
2 769 776 0.991

Per the graph shown above, these percentages increase slightly in the period 1993 through 2006.

What is clear from these two tables is that success rates are higher than they technically need to be.

This is a indication that a double steal is probably perceived by players and managers as a more risky play than it actually is from the Run Expectancy standpoint. The primary reason for this is that the benefits from success are still only potential benefits and so remain partially hidden while the cost of failure is immediately obvious. Essentially, humans are reward-seeking and risk-averse.

What's more interesting, however, is that the table above highlights the disconnect between the actual and perceived value of moving a runner to third base with less than two outs. If it had been the case that managers internalized the much greater run expectancy with a runner on third and one out as opposed to that runner being on second, then we would expect them to take more risks with one out leading to a lower success rate.

As it is, the success rate is lowest with zero outs indicating that managers tend to take more risks with nobody out (there are certainly a percentage of the attempts with no outs that were actually broken hit and run plays, but the play by play data doesn't reliably allow us to exclude those). Even so, about 16% of the unsuccessful attempts in these scenarios actually results in the runner breaking towards second being thrown out rather than the runner advancing to third. When that is factored in, it drives the break even percentages down a few percentages points further.

Of the remaining 21% of double-steal attempts, 19% of those occur with runners on first and third. The success rates using the same definition used for calculating manager's success rates (where at least one runner is credited with a stolen base) are as follows:

Outs Succ Att Pct
0 20 70 0.286
1 146 478 0.305
2 349 457 0.764

Here the accounting of success and failure needs to take a different turn. In a delayed-steal situation, which the majority of these events likely fall into and which has been a hallmark of the game since before 1875, a success is actually dictated by whether or not the runner scores or no outs are recorded by the defense. Under that definition the success percentages for these types of double steals increases by 12%, raising the overall success rate to 63%.

Outs Succ Att Pct
0 31 70 0.443
1 195 478 0.408
2 409 457 0.895

What is perhaps the most interesting point in these two tables is the fact that the success rate is so much higher when there are two outs as opposed to zero or one out. At first glance it is not obvious why this should be the case although I'm sure our enlightened readers will provide some clues.
LaRussa was middle of the pack last year in attempting:

Code: Select all

2006 Double Steals Sorted by Attempts

Team    Manager              Succ     Att     Pct
LAA     Mike Scioscia          13      13   1.000
CIN     Jerry Narron           10      12   0.833
NYA     Joe Torre               8       9   0.889
NYN     Willie Randolph         7       7   1.000
WAS     Frank Robinson          4       7   0.571
CHN     Dusty Baker             5       6   0.833
DET     Jim Leyland             4       6   0.667
FLO     Joe Girardi             5       6   0.833
MIL     Ned Yost                5       6   0.833
MIN     Ron Gardenhire          3       6   0.500
ARI     Bob Melvin              4       5   0.800
BAL     Sam Perlozzo            5       5   1.000
SLN     Tony LaRussa            4       5   0.800
CHA     Ozzie Guillen           2       4   0.500
KCA     Buddy Bell              4       4   1.000
LAN     Grady Little            3       4   0.750
TBA     Joe Maddon              2       4   0.500
TOR     John Gibbons            2       4   0.500
COL     Clint Hurdle            3       3   1.000
SEA     Mike Hargrove           3       3   1.000
SFN     Felipe Alou             3       3   1.000
ATL     Bobby Cox               2       2   1.000
PIT     Jim Tracy               1       2   0.500
TEX     Buck Showalter          2       2   1.000
CLE     Eric Wedge              0       1   0.000
HOU     Phil Garner             0       1   0.000
OAK     Ken Macha               1       1   1.000
PHI     Charlie Manuel          1       1   1.000
SDN     Bruce Bochy             0       1   0.000
BOS     Terry Francona          0       0   0.000

Arthur Dent
Hall Of Famer
Posts: 12317
Joined: April 25 06, 6:43 pm
Location: Austin

Re: Double steals - an underused strategy.

Post by Arthur Dent »

jim wrote:

Code: Select all

Outs      Succ    Att     Pct
0          638   1127   0.566
1         1587   2258   0.703
2          769    776   0.991
:shock: :shock: :shock:


99.1% success rate for a double steal with two outs? Can that be right?

jim
Red Lobster for the seafood lover in you
Posts: 50608
Joined: May 1 06, 2:41 pm

Re: Double steals - an underused strategy.

Post by jim »

Arthur Dent wrote:
jim wrote:

Code: Select all

Outs      Succ    Att     Pct
0          638   1127   0.566
1         1587   2258   0.703
[b]2          769    776   0.991[/b]
:shock: :shock: :shock:
Can this be right?
I just cut and pasted it, and somehow I overlooked that. I'll shoot him an email, cuz that don't seem possible.

jim
Red Lobster for the seafood lover in you
Posts: 50608
Joined: May 1 06, 2:41 pm

Post by jim »

Turns out there is a bug in the way double steals are recorded with 2 outs. When the lead runner is thrown out, it usually isn't recorded that the runner on 1st is also moving unless the both runners are safe. So that explains the unusually high number of successful attempts with 2 outs, because most of the time they aren't counting the ones that don't work.

Post Reply