Last week, I began an examination of stolen base rates. The process is messy with too many variables and nuances to consider. I’m examining the information through several different lenses and seeing what applies. Today, I’m going to look at how success rate plays a role.
Team Level Analysis
As sabermetric principles are being utilized more and more by front offices, they quickly came around to the idea that for stolen bases to be helpful, the success rate needs to be high. In 2000, the success rate was 69% for the entire league and it has increased to 73% last season.
Knowing that each team is made of different players and their individual success rate are a factor, here are the three-year success rate along with total stolen base attempt percentage ((CS+SB)/(1B+HBP+BB)).
|Kansas City Royals||76.2%||17.9%|
|Boston Red Sox||76.2%||4.7%|
|New York Yankees||75.8%||12.4%|
|Toronto Blue Jays||74.1%||9.0%|
|San Diego Padres||73.5%||8.0%|
|Los Angeles Angels of Anaheim||70.0%||11.5%|
|New York Mets||69.7%||8.6%|
|San Francisco Giants||69.6%||7.5%|
|Los Angeles Dodgers||68.2%||5.6%|
|Tampa Bay Rays||67.0%||6.5%|
|Chicago White Sox||66.5%||7.2%|
|St. Louis Cardinals||66.1%||3.5%|
The difference between top and bottom is quite amazing with the Indians coming in at 79% success rate and the Tigers down at 64% and there are disparities throughout the list, but the differences are useless unless the rates are predictive from season to season. Without correcting for the talent level, the r-squared from Year-1 to Year-2 from 2015 to 2017 is .36 for all the matched pairs. Next, I removed the matched seasons when the team changed managers. The r-squared dropped to .27. While a decline, the values are similar.
The biggest takeaway seems to be front offices have more of an impact on the stolen base rate than managers.
One item which bugs me is the differences in team speed. To try to limit its effects, here are the success rate rankings but only for players with an average Speed Score (between 4.0 and 6.0) along with the overall success rate.
|Team||4 to 6 Speed Score SB%||Overall SB%|
|Boston Red Sox||78.7%||76.2%|
|Toronto Blue Jays||78.7%||74.1%|
|San Diego Padres||77.0%||73.5%|
|New York Yankees||76.2%||75.8%|
|San Francisco Giants||74.4%||69.6%|
|Kansas City Royals||73.7%||76.2%|
|Orange County Angels||72.0%||70.0%|
|Los Angeles Dodgers||71.3%||68.2%|
|New York Mets||69.7%||69.7%|
|Tampa Bay Rays||67.8%||67.0%|
|St. Louis Cardinals||67.7%||66.1%|
|Chicago White Sox||61.3%||66.5%|
Teams which accept low stolen base success rates from their average runners do so with all of them.
For fantasy owners, we should be leery of players going to teams which require a high success rate. An example is Stephen Piscotty (three for nine in last season or 33% SB%) going from the Cardinals (66% SB%) to the Athletics (71%). His Steamer projection has him at 4 SB next season. I could see the Athletics completely remove his stolen base opportunities (SBA%). He’s not a perfect example but take note of the teams on the extremes.
That’s it for team-level data, at least for now. Time to move onto the player data.
Player Level Analysis
The general idea I’m looking to investigate is how much impact success rate has on stolen base opportunities. My first test was to take hitters who had varying levels of success in season 1 and compare how their attempt rate change in season 2. I used hitters from 2006 to current who had 300 PA in each paired season (all values are in percentage point changes).
|Success rate||< 50%||50% to 60%||60% to 70%||70% to 80%||80% to 90%||> 90%|
A couple observations: The first is that all saw their attempt rates drop some. This not a surprise with most hitters reaching their peak speeds before they reach the majors. The second point may be more fantasy relevant. Those hitters who had high success rates in the previous season saw their SBA% drop more than average. Just because a hitter was successful in the previous season, it doesn’t mean he’ll steal a lot in the next one.
Next, here are the SBA% changes for runners who saw their success rate drop in a season’s first half (again all values are in percentage point changes).
|Change in SB%||1H to 2H SBA%||1H to 2H SB%||1H to 2H SBA%||1H to 2H SB%|
|>30% point Drop||3.4%||37.1%||4.1%||38.9%|
|20% to 30% Drop||1.5%||17.2%||2.7%||20.8%|
|10% to 20% Drop||-1.1%||5.2%||-0.4%||4.7%|
|0% to 10% Drop||0.7%||2.7%||0.6%||2.9%|
|0% to 10% Increase||-0.1%||-3.2%||0.1%||-1.9%|
|10% to 20% Increase||1.8%||-9.3%||2.3%||-7.1%|
|20% to 30% Increase||3.9%||-10.3%||3.0%||-13.3%|
|>30% point Increase||4.6%||-26.2%||5.7%||-23.7%|
Welcome to a table on human nature. Here’s how each pair column is important.
- Column one and three. If a player was highly successful or highly unsuccessful in the season’s first half, they will try to steal more in the season’s second half. I’m guessing those who were unsuccessful may have been injuried while those who were successful kept on stealing.
- Column two and four: The jump or drop in stolen bases comes back to regress to the mean.
Owners may be able to take advantage of high first half success rates and trade them off before their second half crash.
The following examples were the only factors I could find in which SB% is useful. I talked to someone who was working on a similar project and he gave me some much-needed insight on SB%. It’s not much of a predictive factor for anything and has probably been over utilized for years. I’m not going to give away any more of his work but generally ignore SB% except to find a team’s tolerance level and find those runners who will be over confident and keep running, but less successfully, into a season’s second half.
My quest for a better understanding of stolen bases numbers has been slow. Even while examining SB%, I went down several avenues with no luck. The next major change is using StatCast’s Sprint Speed values. I need to find and add the values to my database to easily link to players. I’m not sure when I will complete the task so any more stolen base talk is on hold for now. Until then, let me know if you have any ideas for possible study areas or improvements.
Jeff, one of the authors of the fantasy baseball guide,The Process, writes for RotoGraphs, The Hardball Times, Rotowire, Baseball America, and BaseballHQ. He has been nominated for two SABR Analytics Research Award for Contemporary Analysis and won it in 2013 in tandem with Bill Petti. He has won three FSWA Awards including on for his MASH series. In his first two seasons in Tout Wars, he's won the H2H league and mixed auction league. Follow him on Twitter @jeffwzimmerman.