Are baserunning and stolen base effectiveness random?

[stealofhome]

I've been trying to figure out what makes a good baserunner (by BsR) and base stealer (using linear weights for SB, basically +0.2 runs for SB and -0.4 runs for CS) in PT but have come up empty.

Even looking at cumulative career stats for players with a large number of SB attempts or PA, there is very little correlation (5-10%) between the SPE, STE, RUN ratings (I've looked at others but haven't seen a correlation) and a runner's effectiveness. Anyone else have better luck?

I imagine in PT, many people are using insanely defensive catchers which may reduce the ability for top runners, but I imagine there should still be some trend. I also realize that team and player strategy would have some role, but again I'm using cumulative stats for each player over multiple teams which would hopefully even some of that variation out.

[Thalion]

I didn't realize it until recently when I started drilling down on some pitching statistics, but I think a pitcher's hold runner rating may play a HUGE role in the success rate of base stealers, more than I thought previously.

[RchW]

Originally Posted by Thalion

I didn't realize it until recently when I started drilling down on some pitching statistics, but I think a pitcher's hold runner rating may play a HUGE role in the success rate of base stealers, more than I thought previously.

Yes it does.

[stealofhome]

Originally Posted by Thalion

I didn't realize it until recently when I started drilling down on some pitching statistics, but I think a pitcher's hold runner rating may play a HUGE role in the success rate of base stealers, more than I thought previously.

But this should even out over a large sample, right?

Originally Posted by RchW

Yes it does.

Care to elaborate any more details you may know?

[<Pion>]

Originally Posted by Thalion

I didn't realize it until recently when I started drilling down on some pitching statistics, but I think a pitcher's hold runner rating may play a HUGE role in the success rate of base stealers, more than I thought previously.

Not quite as large a difference as I thought it might be (depending on ones expectation of large). The league average SB rate is 65%. Those with a hold rating of 1 to 20 give up 70% and those with 80 to 100 give up 60%.

Edit - Thought I would check to see if the # of SB attempts went down considerably with hold rating. To get a workable number, I used a SBA9 (Stolen Base Attempts per 9 innings).

1-20: 1.40
81 to 100: 0.55
All: 0.87

SB9 (Stolen Bases per 9 innings)
1-20: 0.99
81 to 100: 0.33
All: 0.57

So yeah, when you combine the number being thrown out with the reduction in people running, there are three times more stolen bases for hold ratings of 1 to 20 than there are for 80 and over.

[<Pion>]

Originally Posted by stealofhome

But this should even out over a large sample, right?

FYI... The sample size I used was 959,000 innings

[stealofhome]

This is based on the pitcher's rating? Do you have numbers for how the rating of the runner impacts the effectiveness?

[mess]

I was trying to figure out why my Ichiro had negative WAR even though his stats are pretty standard across the board.

He's only stealing bases at a 50% clip. Didn't think that base running played into batting WAR (when obviously it does... duh)

[<Pion>]

Originally Posted by stealofhome

This is based on the pitcher's rating? Do you have numbers for how the rating of the runner impacts the effectiveness?

Wasn't sure how to rate them, whether it should be by stolen base rating, speed, baserunning ratings, a combo of one, two or three? But yeah, could do the same if we chose a methodology. Wasn't as straightforward as comparing the Hold rating, so I haven't done it yet.

[<Pion>]

Originally Posted by mess

I was trying to figure out why my Ichiro had negative WAR even though his stats are pretty standard across the board.

He's only stealing bases at a 50% clip. Didn't think that base running played into batting WAR (when obviously it does... duh)

I think people see Ichiro's batting average and he ends up getting rated very highly, but his RC27 is not worth all the accolades he gets. That being said, however, when adding in his defensive ability as well, he is good player.

[stealofhome]

Originally Posted by <Pion>

Wasn't sure how to rate them, whether it should be by stolen base rating, speed, baserunning ratings, a combo of one, two or three? But yeah, could do the same if we chose a methodology. Wasn't as straightforward as comparing the Hold rating, so I haven't done it yet.

I created a rating - rSB - by weighting SB as +0.2 runs and CS as -0.4 runs, which is about what is used for calculating wOBA. Then turned that into a rate stat (rSB/PA). You could estimate SB opportunities which I remember seeing a method for online but can't place right now. Then run a multiple regression using Speed, Stealing, and Baserunning to see how each variable impacts the likelihood of a positive number of stolen base runs per PA. But again, I did not see any correlation.

Edit: Found the page I was looking for: http://walksaber.blogspot.com/2005/1...peed-unit.html

So you could create Stolen Base Attempts = (SB+CS)/(1B+BB+HBP) and use rSB/SBA, or even go as far as creating a full Speed Score for each player (which I might do).

[freedavebrown]

Anyone else notice that runners get caught stealing way more in PT than in real life? The CS numbers seems way high. In both my Perfect and Bronze leagues right now the CS leaders would all be among the highest all time for a single season in MLB history when projected out to 162 games.

[<Pion>]

Originally Posted by stealofhome

I created a rating - rSB - by weighting SB as +0.2 runs and CS as -0.4 runs, which is about what is used for calculating wOBA. Then turned that into a rate stat (rSB/PA). You could estimate SB opportunities which I remember seeing a method for online but can't place right now. Then run a multiple regression using Speed, Stealing, and Baserunning to see how each variable impacts the likelihood of a positive number of stolen base runs per PA. But again, I did not see any correlation.

Edit: Found the page I was looking for: http://walksaber.blogspot.com/2005/1...peed-unit.html

So you could create Stolen Base Attempts = (SB+CS)/(1B+BB+HBP) and use rSB/SBA, or even go as far as creating a full Speed Score for each player (which I might do).

Nicely done. I was thinking you were asking something else.

In my previous posts, I was trying to find out the impact of the hold rating of a pitcher on stolen bases. The original OP, I think, was trying to find a correlation between the OOTP ratings (stolen base, speed, baserunning) and stolen base success. So what I was wondering was the best way to compare stolen base success back to the ratings of the player. With the hold number of pitchers, it was a singular rating to compare to. With runners, it could be one, two, or three ratings it would need to be compared to. That's why I didn't do an analysis similar to the hold rating analysis. Not sure if I should compare it to stolen base rating only, or a combo of stolen base and speed, etc. I would think, but I'm not sure, that one of those number is primarily used to determine frequency while the other may dictate success. If we knew how it was used, a comparison to results could be made.

[BPS]

Originally Posted by stealofhome

I've been trying to figure out what makes a good baserunner (by BsR) and base stealer (using linear weights for SB, basically +0.2 runs for SB and -0.4 runs for CS) in PT but have come up empty.

Even looking at cumulative career stats for players with a large number of SB attempts or PA, there is very little correlation (5-10%) between the SPE, STE, RUN ratings (I've looked at others but haven't seen a correlation) and a runner's effectiveness. Anyone else have better luck?

I imagine in PT, many people are using insanely defensive catchers which may reduce the ability for top runners, but I imagine there should still be some trend. I also realize that team and player strategy would have some role, but again I'm using cumulative stats for each player over multiple teams which would hopefully even some of that variation out.

If you haven't done so, you'll want interactive terms like (SPE x STE) in your regression.

[stealofhome]

So this is what I found:

Stolen Base Attempts per Opportunity = -0.05 + Speed*0.0035 (R Sq = 0.79, N = 202, minimum 100 SBO)

So it looks like speed alone (and strategy) defines how often a runner attempts to steal.

The effectiveness though is a little muddier.

Stolen Base Runs per Attempt = -0.2 + Steal*0.003 - Speed*0.0008 (R sq Adj = 0.43, N = 124, minimum 15 SBA, p value of intercept/Steal are basically zero, p value of speed is 0.000231.)

I did attempt to use the interaction of steal and speed but got basically the same thing (negative coefficient, R sq adj = 0.43). This result is actually much better than I've seen before but still basically a coin flip. It also makes no sense that speed would have a negative impact on how likely a runner is to successfully steal.

Edit: My only thought is that managers consistently over-set stealing strategy for fast players? For instance, Jose Altuve on the summary has a negative stolen base run value (-3.4 runs over 1,807 attempts). It may be that he has been set too aggressive and so the speed negatively impacts the success rate in that way.

[Weaseltail]

Originally Posted by stealofhome

My only thought is that managers consistently over-set stealing strategy for fast players? For instance, Jose Altuve on the summary has a negative stolen base run value (-3.4 runs over 1,807 attempts). It may be that he has been set too aggressive and so the speed negatively impacts the success rate in that way.

Really interesting stuff, thanks for posting. I know my success rate went through the roof when I reduced my global strategy slider to below the midpoint and changed all the player strategies to the midpoint or below. Fewer SBs, but 80% success rate.

I remember reading that a high slider value (global and player) results in more pitch outs and throws to first, so it would make sense that that being too aggressive would be counterproductive.

[BPS]

Originally Posted by stealofhome

So this is what I found:

Stolen Base Attempts per Opportunity = -0.05 + Speed*0.0035 (R Sq = 0.79, N = 202, minimum 100 SBO)

So it looks like speed alone (and strategy) defines how often a runner attempts to steal.

The effectiveness though is a little muddier.

Stolen Base Runs per Attempt = -0.2 + Steal*0.003 - Speed*0.0008 (R sq Adj = 0.43, N = 124, minimum 15 SBA, p value of intercept/Steal are basically zero, p value of speed is 0.000231.)

I did attempt to use the interaction of steal and speed but got basically the same thing (negative coefficient, R sq adj = 0.43). This result is actually much better than I've seen before but still basically a coin flip. It also makes no sense that speed would have a negative impact on how likely a runner is to successfully steal.

Edit: My only thought is that managers consistently over-set stealing strategy for fast players? For instance, Jose Altuve on the summary has a negative stolen base run value (-3.4 runs over 1,807 attempts). It may be that he has been set too aggressive and so the speed negatively impacts the success rate in that way.

Perhaps higher speed --> more risky stolen base attempts --> lower success rate of stealing bases.

[stealofhome]

This is what I came up with for a formula to determine expected stolen base runs over 500 PA:

500*
(0.2-0.0008*Gap+0.0014*Eye-0.0005*Power+0.0014*Contact)*
(-0.05+0.004*Speed)*
(-0.2+0.003*Steal-0.0008*Speed)

This accounts for Opportunities/PA, Attempts/Opportunity, and Success/Attempt.

Top 5 ranked that I have in my database: Ty Cobb 96, King Kelly 98, Billy Hamilton 61, Terrance Gore 44, and Alan Wiggins 62. They're all about 2-3 expected runs over 500 PA, although this method tends to reduce the extremes somewhat.

[bailey]

If the combination of pitcher's Hold Runners rating and the catcher's Arm rating is higher than the base stealers rating plus baserunning skills, he is very likely to get caught stealing. That is why Ty Cobb 96 is so successful. So, you could improve your chances for other players by micro-ing every game! Just my anecdotal observation, no math.

[freedavebrown]

Originally Posted by bailey

If the combination of pitcher's Hold Runners rating and the catcher's Arm rating is higher than the base stealers rating plus baserunning skills, he is very likely to get caught stealing. That is why Ty Cobb 96 is so successful. So, you could improve your chances for other players by micro-ing every game! Just my anecdotal observation, no math.

Interesting. That might explain the crazy high caught stealing numbers. It is probably the stat that I have noticed that is the most out of wack with real stats so far in PT. Way too many CS.