Manager Influence on Stolen Bases by Jeff Zimmerman January 18, 2018 Earlier this month, I asked our readers for any aspects of the fantasy game which are missing. Okra stepped up and said: “I feel like we still do a poor job of predicting stole bases. I think we could better utilize the new Sprint Speed data and speed scores to predict SBs. Taking it one step further would be to try and quantify each managers propensity for SB attempts.” This statement is 100% true. We really don’t know which measurable factors fantasy owners should focus on when looking for stolen base breakouts. I’ve gone ahead and dived into the topic of just the manager influence with positive results. Truthfully, this topic is a mess. With players, managers, and front offices changing all the time, it’s tough to know where to begin. For now, I’m starting with an initial shallow breakdown of the managers. The biggest issue surrounding the entire topic are each team’s different speed talent levels. Dee Gordon stole 60 bases with the Marlins (RIP). Now, he’ll be stealing bases with the Mariners. Miami’s total stolen bases will drop and the Mariners will rise but how will Gordon’s stats change. I needed a baseline to compare each team. Here’s my first attempt. It’s heavy on the math and clunky. If anyone notices a way to improve it, please let me know. First, I needed a metric to show stolen base attempt percentage (SBA%) and I went with: (SB+CS)/(1B + HBP +BB) Now, I’ve seen but can’t find an article at Tom Tango’s new or old website which was a better representation of stolen base opportunities. My Google skills must be subpar at my advanced age. I’d probably use this unknown equation in future attempts but the above one will work for now. To show each player’s speed, I used the Speed Score available here at FanGraphs as a proxy. While not close to perfect, it provides some context for a hitter’s speed. I know Sprint Speed is available at Baseball Savant but I’m just looking for a basic model for now. Additionally, Sprint Speed is missing one component, acceleration. It may take one player three steps to get up to speed while another may peak at the same speed when reaching first base. To start with, here is a graph comparing Speed Score to SBA% for hitters with a minimum 200 PA from 2015 to 2017. I used these seasons because I want the option to examine Sprint Speed at some future point and teams are now becoming more attuned to the SBA% needed for positive results: r of .814 The best-fit equation will be used to find the league average SBA% with a known Speed Score. Now, the sample size and math get a little fuzzy. With the above equation, I found the median (some extremely high values messed up the average values) SBA% and Speed Score for each manager (Note: I will continue to state the manger’s influence but the influence could be from other sources like the front office or another coach.) over the past three seasons. With these values, I matched up the seasons when the manager had his job in back-to-back seasons. With these matched seasons, I adjusted SBA% based on the team’s median Speed Score. I ended up with an R of .66 (r-squared of .43). This finding is a BIG deal. With the R at .66, two-thirds of a manager’s adjusted stolen base rate can be predicted from the previous season’s value with the other one-third being the league average value. Manager stolen base tendencies are highly predictive but do their SBA% vary enough to matter. Starting with managers’ team Speed Scores, I found the normal SBA% and its ratio to the actual value. Here the 2017 results: Note: I had values leaning to the positive end and adjusted the values for a 1.0 average ratio. I’m guessing the logarithmic equation is causing the differences. I will try to correct later. It is quite the difference. Here is an example. Assume a player with a Speed Score of 6 and 200 opportunities (160 1B + 30 BB + 10 HBP). Plugging in his Speed Score for an average team, he ends up with a .073 SBA% or a projected 14.6 SB (.073*200). Now, using the two extreme values, his stolen bases would drop to 8.0 (SBA% .040) with Baltimore’s Buck Showalter or jump to 22.8 with the .114 SBA% from Orange County’s Mike Scioscia. A difference of 15 stolen bases is again a BIG deal. Note: One item I’ve noticed is that some players seem to have their own go tendencies independent of the team tendencies. Go figure. It’s another variable to consider. Remember, I just dove in and hope to understand more as I continue to research the topic. With the preceding information, here are the 2018 projected (very Beta version) manager adjustments with new managers getting a 1.00 value. Projected SBA% Manager Adjustments (BETA) Team Manager 2017 SBA% Multiplier 2018 adjustment w/ regression Los Angeles Angels Mike Scioscia 1.56 1.36 Milwaukee Brewers Craig Counsell 1.42 1.26 Seattle Mariners Scott Servais 1.28 1.18 Cincinnati Reds Bryan Price 1.21 1.13 Texas Rangers Jeff Banister 1.21 1.13 San Francisco Giants Bruce Bochy 1.18 1.11 Tampa Bay Rays Kevin Cash 1.15 1.09 San Diego Padres Andy Green 1.12 1.07 Oakland Athletics Bob Melvin 1.12 1.07 Houston Astros A.J. Hinch 1.08 1.04 Kansas City Royals Ned Yost 1.03 1.01 Cleveland Indians Terry Francona 1.02 1.00 Toronto Blue Jays John Gibbons 1.02 1.00 Washington Nationals Dave Martinez – 1.00 Boston Red Sox Alex Cora – 1.00 Detroit Tigers Ron Gardenhire – 1.00 New York Yankees Aaron Boone – 1.00 New York Mets Mickey Callaway – 1.00 Philadelphia Phillies Gabe Kapler – 1.00 Chicago White Sox Rick Renteria 0.99 0.98 Arizona Diamondbacks Torey Lovullo 0.97 0.97 Miami Marlins Don Mattingly 0.93 0.94 Pittsburgh Pirates Clint Hurdle 0.92 0.93 Minnesota Twins Paul Molitor 0.90 0.92 Los Angeles Dodgers Dave Roberts 0.90 0.92 St. Louis Cardinals Mike Matheny 0.88 0.91 Colorado Rockies Bud Black 0.81 0.87 Atlanta Braves Brian Snitker 0.81 0.87 Chicago Cubs Joe Maddon 0.75 0.82 Baltimore Orioles Buck Showalter 0.55 0.69 An interesting note is on Dee Gordon going to the Mariners. On average, Mattingly limited the number of stolen bases in Miami while Scott Servais lets his players run a little more. While projections have his stolen bases regressing from 60 to 43, the drop may not be as much as projected. In summary, here are the two main points so far. Managers have their own stolen base attempt tendencies with the previous season getting two-thirds the weight and the rest being the league average. Managers have extreme tendencies when it comes to allowing stolen base attempts. I found a little more useful information than I expected. While I expected there to be some differences with the managers, I didn’t expect the year-to-year numbers to be as sticky. It’s a start but the topic needs to be examined further. For my next step, I determine if the prior success rate is needed for a manager to keep sending a runner. I tried to include these values here but the process was too messy. Until then, let me know what you think or how the process can be improved.