Existing User? Login here!

Back to Boards

Formula to effectively mimic stolen bas

Formula to effectively mimic stolen bas

Here is a formula suggestion that I think would more accurately assess the SP rating (whether or not players will steal a base given the chance).

Taking triples out of the equation (can stay in, but here is a simple formula).

Plate appearances * (OBP) = times on base.

Times on base / (divided by) steals = times on base for every 1 steal.

Anything below a 5.00 gets a 9. Anything below a 6.25 gets an 8.

5.00 would indicate 1 steal for every 5.25 times on base.

Plate appearances are easy enough to import, and could factor in for other offensive categories.

At bats count against players who do not take a walk and hence drives down their SB/AB. Thus the above formula would remove that heavy bias.

With this formula you can take Juan Samuel 1984:
737 plate appearances * .307 OBP = 226 times on base.

226 / 72 stolen bases = 1 stolen base per 3.14 times on base.

Compare this with Ricky Henderson 1980

722 plate appearances * .420 OBP = 303 times on base.

303 / 100 stolen bases = 1 stolen base per 3.03 times on base.

Almost exactly the same! When you factor in Samuel has 19 triples to 7 for Henderson AND Samuel had 34 triples or homers compared to 13 for Henderson. The triples or homers count towards times on base but are NOT stolen base opportunities.

Additionally, you can go one step further with the following formula:

Plate appearances * (OBP) = times on base.

Times on base - (minus) home runs - (minus) triples = Stolen base opportunities

Stolen base opportunities / (divided by) steals = stolen base opportunities for every 1 stolen base.

Here if you apply this formula to Juan Samuel and Rickey Henderson, you get 2.67 for Samuel and 2.90 for Henderson, rating Samuel as more likely to steal a base with an opportunity than Henderson.

Re: Formula to effectively mimic stolen

Anyone besides me think that this formula would be a more accurate portrayal of 'potential to steal in a given season?' And utilizing SB/AB which is not even a MLB statistic (for a good reason) should/could be changed if we are trying to portray one of the more important ratings in this game (SP)?

Alternatives to the formula above include the following from Fangraphs:
Base Running (BsR) is FanGraphs’ all encompassing base running statistic that turns stolen bases, caught stealings, and other base running plays (taking extra bases, being thrown out on the bases, etc) into runs above and below average. It is the combination of Weighted Stolen Base Runs (wSB), Weighted Grounded Into Double Play Runs (wGDP), and Ultimate Base Running (UBR) which are all available on the leaderboards and player pages.

BsR serves as the base running component of Wins Above Replacement (WAR) and provides a lot more information than simply looking at a player’s stolen base total if you are interested in judging a player’s base running performance.

The problem with this rating is that it is given based on the player's contribution that season over X number of games. So, one would need to have an 8+ rating for a full season of 700 PA - (For a 10 SP rating in PC) for instance -- or a 2+ for a 10 rating with less than 200 PA, etc.

Baseball reference has a more simplistic rating called 'rbaser'

Re: Re: Formula to effectively mimic st

Biggest part of the problem here is that you are not considering the catchers these guys are facing...they are not facing any random catchers from an average 30 teams season... they are in 12 teams league were most of the catchers are studs of their era playing in their best season..... this would have a lot more to do with sb efficiency... if you are facing Ivan Rodriguez, Johnny Bench, Gary Carter, Roy Campanella night in and night out, Sb efficiency will drop and so are SB attempts... A more refine catchers rating would be way more realistic than basing numbers on how often Rickey Henderson stole bases vs any given average Joe catching...

Re: Formula to effectively mimic st

The formula is not exactly sb/ab, but it's not an unfair point.

It's really challenging bringing in sabermetrics. dWAR took several months to import, so there has to be a really big bang for the buck.

I don't know that the existing formula works terribly, I still haven't seen this slew of examples where it's bad. But if there's a way to work within the existing stats, possibly... I say that because a major change to a stat everyone has gotten used to will result in a lot of complaining.

Re: Re: Formula to effectively mimic st

I dug into this a little more (it's been like over 10 years since I created the speed formula and haven't touched it since - so that says it must be working okay, but is obviously a little "dated"...)

Comparing Ricky Henderson 1980 to Juan Samuel 1984 isn't the best comparison because it looks like Henderson gets bumped to a 10 due to a multiplier effect on the total SBs... I think my intent here again was to reward the high volume/risk taking. I believe Henderson 1980 would be a 9 without that. I personally have no problem with it since Henderson was the best of the best. I like speed over a pure SB% success rate (same reason there's an Arm rating - SB and CS stats are too small, random and difficult to apply as straight percentages).

If the larger concern is that a "9" isn't succeeding enough, that's a different argument that could be looked at.

While I agree "ABs" being part of the formula probably doesn't make as much sense as times on base, it doesn't appear that it has resulted in a swath of unfair speed ratings, unless I'm missing something.

Re: Formula to effectively mimic st

Let's give another example who had similar stats to Samuel.

The key here is LOW WALKS. When someone has low walks, they have less times on base and less opportunities to steal.

Hence Samuel, Willie Wilson, and some of the newer players like Trea Turner are all penalized for not taking enough walks. So should we say, "Shame on you for not taking more chances?"

They are taking the same percentage of chances as Ricky Henderson! Sometimes more as proven with Willie Wilson.

But in the interest of numbers, here is Willie Wilson 1979 and 1980:

640 plate appearances * .351 OBP = 225 times on base.
225 / 82 stolen bases = 1 stolen base per 2.74 times on base.

745 plate appearances * .357 OBP = 266 times on base.
266 / 79 stolen bases = 1 stolen base per 3.37 times on base.

Juan Samuel 1984:
737 plate appearances * .307 OBP = 226 times on base.
226 / 72 stolen bases = 1 stolen base per 3.14 times on base.

Ricky Henderson 1980
722 plate appearances * .420 OBP = 303 times on base.
303 / 100 stolen bases = 1 stolen base per 3.03 times on base.

These are all elite numbers but Willie Wilson 1979 (9 rating) rates as the player most likely to steal given the chance (before removing 3B and HR from the 'times on base' equation). Willie Wilson '79 is the stolen base king over Henderson 1980!

-Guy would prefer to have a formula that uses existing metrics (does not use plate appearances). I have a proposal for that:
Hits + Walks + Hit By Pitch (HBP) = Times on base
Times on base / steals = Steal per times on base

The above formula gets a very close to accurate count of the times on base for that player that season.

Lastly, the formula could also look like this.

Times on base - (minus) HR - (minus) 3B = Stolen base opportunities

Then stolen base opportunities / stolen bases = Total opportunities for every steal.

Re: Speed Ratings

Now lets investigate 9 rating vs. 8 rating using the same metrics.

A great example is Dee Gordon 2015.
Gordon is/was a speedster that year and stole 58 bases but is rated an '8' by PC.

Stolen Bases are largely driven by not just how fast the player is, but if they have the 'green light' to steal.

Let's compare Dee Gordon to a couple of 9 ratings in PC.

Dee Gordon 2015 (8, should be 9):
653 plate appearances * .359 OBP = 234 times on base.
234 / 58 stolen bases = 1 stolen base per 4.04 times on base.

Willie Wilson 1980 (9, should be 10):
745 plate appearances * .357 OBP = 266 times on base.
266 / 79 stolen bases = 1 stolen base per 3.37 times on base.

There are actually very few 9's in the 2000's largely because less people are taking walks and less players receive the green light to steal. Those that do lead the league or are among the leaders are penalized in PC by their lack of walks, IE Gordon, Visquel, Ichiro.

Re: Re: Speed Ratings

This is all well and good, but it is not considering the Pitcher, is he a lefty? they are Harder to run against, does he have a great Pick off move?

Is the catcher very good at throwing out runners?

Some of the times a runner reaches base but doesn't attempt a steal cause its not in the game Flow? ( up say 7 runs) a man on 1st and 2nd and the speedster is on 1st caught behind the slow poke.

While the formula maybe could be improved, there is no set way to do it that includes all the variables

Re: Speed Ratings

SCM that's all true, I believe the OP is just suggesting an improvement to the current speed formula.

Re: Re: Speed Ratings

I thought it was about defining stolen Base capabilities and how it was driven by Speed which of course is a very important factor, but speed and Base stealing ability are two very different skills, a guy could be very fast and run the bases well, but horrible at reading pitchers and unable to get a good jump, so not a very big stolen base threat.

I was not trying to do anything but add a little more input, in my opinion Base stealing should be based strictly on a players actual stolen base abilities, PC really does not do a very good job with it, guys can have 10 steals in Real life and are incapable of doing anything more than being the back end of a double steal in PC. But it's a factor that we have to get used to even though it does devalue some guys - offense is more than Batting.

On that vein, does Slump Buster effect Base Stealing? it should, that's an offensive statistic, I don't think it does though, I mean, whats the point of Billy Hamilton walking if he isn't going to steal 2nd? I know that he normally will, but if he's slump busted he definitely should get you that extra base, or two...

Re: Formula to effectively mimic stolen

It’s the biggest issue with speed and base stealing yes. The problem is you simply can’t use stolen base percentage as a way to gauge base stealing because that only works for the guys who have a big enough sample size which is very few players. And as you mentioned there’s a lot more that goes into the success of a steal. Nowadays maybe there is an advanced metric that would work better. But all in all while it’s a little “arcadey” the speed rating has served surprisingly effective despite devaluing some players unfairly.

The OPs suggested tweak seems legit to me. But the silence on the thread I think speaks to the fact it’s not a roaring issue. We’re essentially talking about going from 9 to 10 or 8 to 9 for certain players. Granted that one point is fairly significant.

Re: Re: Formula to effectively mimic st

If you want accuracy in base running/stealing, play Strat-O-Matic online and pay 24.95 per season ;-)

- The Sheik

Re: Formula to effectively mimic st

In all honesty the base stealing isn’t a concern for me.
I have more issues with base running , way too many runners not scoring from 2nd base with 2 outs on bloop singles . And runners that can actually run. Very few players in MLB can’t score from 2nd with 2 outs on bloop or groundball singles.