My Problem with Stats (SLG and OPS)

[JayVee]

Hey Brian, you know you don't have to quote the whole post, right?

[John_Brian_K]

Hey Brian,

Yes I do, but it is quicker.

[will the thrill]

I think if you just put in terms of:

Hey player A has a .900 OPS, but not one stolen base and player B has an .830 OPS with 40 SB. In that case I will take player B all day.

So you take the guy the .355/.475 guy over the .400/.500 guy because of 40 steals? a stolen base every 4 days?

[John_Brian_K]

So you take the guy the .355/.475 guy over the .400/.500 guy because of 40 steals? a stolen base every 4 days?

I just threw .830 out there, but I am not sure every day I take that player. I would have to put some more thought into it. Just looking at it quick again.....I probably do take the 40 SB guy because those 40 SB are basically doubles right? Just what the OP was talking about. Add 40 2b to his SLG and I am sure he comes out ahead.

[will the thrill]

If we count a SB as getting a double, wouldn't we have to count a walk as a single? Same theory right. A 1B and a SB (or BB + SB) only equal a bases empty double, not a double with a runner on base that then gets him home for a run. 2B > 1B + SB how much greater is the bigger question.

[John_Brian_K]

If we count a SB as getting a double, wouldn't we have to count a walk as a single? Same theory right. A 1B and a SB (or BB + SB) only equal a bases empty double, not a double with a runner on base that then gets him home for a run. 2B > 1B + SB how much greater is the bigger question.

Yeah either way I guess.

How are you doing this morning Will?

[STLTiger69]

Thank you everyone for your comments and ideas, I very much do apprecate them.

For what it is worth I do not think that just by adding SB's to SLG (or how SLG is factored into OPS) suddenly makes it a "pefect" stat. I do think that in adding stolen bases to the SLG calculation makes it more meaningful to compare a base stealer to a home run hitter. And frankly some players who both hit home runs and steal bases (Matt Kemp 34 HR's/40SB's, Ryan Braun 31 HR's/31SB's, Curtis Granderson 40HR's/24 SB's for example) will really stand to gain a lot by this revision.

As John_Brian_K points out, a team or a fan is still able to say Player A is a better middle of the order hitter or player B is a better lead off hitter - and that makes sense. I would never say that Bourn is the ideal #5 hitter as Youkilis is a perfect hitter to bat #1 in a lineup. But I think it's more accurate to say that in terms of offensive production by total bases, they are at least comparable to each other in terms of production.

Thanks again for the comments and ideas everyone!

[JayVee]

Yes I do, but it is quicker.

My point is that it is good forum etiquette to edit out long quotes to prevent others from needing to scroll more than necessary.

[Biff Mayhem]

Will Brad Pitt be playing the part of Lee Panas?

Christopher Mintz-Plasse's agent is anxiously waiting for the call for his client to read for that part.

[Biff Mayhem]

For what it is worth I do not think that just by adding SB's to SLG (or how SLG is factored into OPS) suddenly makes it a "pefect" stat.

I would be surprised if there were anyone who said there was a perfect stat.

[will the thrill]

Yeah either way I guess.

How are you doing this morning Will?

I am excellent, Michigan Win, MSU and OSU loss makes for a good weekend.

[John_Brian_K]

My point is that it is good forum etiquette to edit out long quotes to prevent others from needing to scroll more than necessary.

ok

[Mr. Bigglesworth]

Somewhat off-topic, but I really wish they would break down the basic stats into percentages of plate appearances. For example:

2011 Miguel Cabrera:

In a given plate appearance, he:

Gets a hit 27.4% of the time - broken into:

1B - 16.7% of the time
2B - 6.7% of the time
3B - 0.0% of the time
HR - 4.0% of the time

He also gets a:

BB - 16.1% of the time and
HBP - 0.5% of the time

In total, he reaches base 44% of the time (27.4% of the time by hit + 16.1% by BB + 0.5% by HBP)

He makes an out 56% of the time, made up by:

K: 13.3% of the time
Fly Ball out: x.x%* of the time
Ground Ball out: x.x%* of the time
Line Drive out: x.x%* of the time

*I don't know where to locate this data, but I don't doubt it exists, and it isn't hard to convert it into a percentage.

I think that gives a much more accurate and easy to understand summary of what Miguel does at the plate.

[Mr. Bigglesworth]

For fun, I'll do Delmon Young since joining the Tigers:

He gets a hit 27.2% of the time per plate appearance (pretty similar to Miguel), of which he gets a:

1B - 19.9% of the time,
2B - 2.9% of the time,
3B - 0.7% of the time,
HR - 3.7% of the time,

He also gets a:

BB - 2.2% of the time, and a
HBP - 0.7% of the time

He gets on base 30.1% of the time (27.2% + 2.2% + 0.7%)

He makes an out 69.9% of the time, of which:

he strikes out 16.2% of the time (a higher rate than Cabrera, incidentally)
I do not know his ground out, fly out, or line-out rates.

So if you need a hit - any type of hit in a given AB, and a walk does not help you appreciably (e.g. runner at 3rd, two outs, tie game in the bottom half of the last inning), Young isn't much worse a bet than Cabrera. But pretty much all other situations he is, and it is very easy to see why with the rate data - less power, far fewer walks, etc. I contend it is easier to understand it when presented this way. JMO.

[john doe]

STLTiger69‎, This is by far one of the best observations of this subject that I've had the opportunity to read, great points and well thought out. Bravo! Now, the additional caveat, the day to day affects of the human condition on athletes, discussed widely among the sports medicine community, I would guess this would be next to impossible to quantify.

[Mr. Bigglesworth]

Speed also can't be factored in when you take an extra base on basehits or break up a double play or make a play on a groundball go to 1st instead of the force at second. It's the great matzah ball out there that stat people act like it doesn't occur. Great topic!!!!

The flip side is speed can lead to infield singles and doubles on balls cut off in the gap, which is good for the batter, but typically cannot score runners from second and first, respectively. So some of those 'speedy guy' hits are not as valuable with runners on base as a more orthodox single / double.

[Shelton]

Stolen bases are important, but so are caught stealing numbers, as many have already pointed out. There are so few base stealers that have a high enough success rate that ops is generally good enough for evaluating a hitter. I look at base stealing as an added bonus, much like defense. There is no perfect stat. You have to consider a variety of factors.

[tiger337]

For what it is worth I do not think that just by adding SB's to SLG (or how SLG is factored into OPS) suddenly makes it a "pefect" stat. I do think that in adding stolen bases to the SLG calculation makes it more meaningful to compare a base stealer to a home run hitter. And frankly some players who both hit home runs and steal bases (Matt Kemp 34 HR's/40SB's, Ryan Braun 31 HR's/31SB's, Curtis Granderson 40HR's/24 SB's for example) will really stand to gain a lot by this revision.

As John_Brian_K points out, a team or a fan is still able to say Player A is a better middle of the order hitter or player B is a better lead off hitter - and that makes sense. I would never say that Bourn is the ideal #5 hitter as Youkilis is a perfect hitter to bat #1 in a lineup. But I think it's more accurate to say that in terms of offensive production by total bases, they are at least comparable to each other in terms of production.

Thanks again for the comments and ideas everyone!

Adding stolen bases to the equation is a good idea. Not also penalizing a player for CS is a bad idea.

[DontGotSheed]

For fun, I'll do Delmon Young since joining the Tigers:

He gets a hit 27.2% of the time per plate appearance (pretty similar to Miguel), of which he gets a:

1B - 19.9% of the time,
2B - 2.9% of the time,
3B - 0.7% of the time,
HR - 3.7% of the time,

He also gets a:

BB - 2.2% of the time, and a
HBP - 0.7% of the time

He gets on base 30.1% of the time (27.2% + 2.2% + 0.7%)

He makes an out 69.9% of the time, of which:

he strikes out 16.2% of the time (a higher rate than Cabrera, incidentally)
I do not know his ground out, fly out, or line-out rates.

So if you need a hit - any type of hit in a given AB, and a walk does not help you appreciably (e.g. runner at 3rd, two outs, tie game in the bottom half of the last inning), Young isn't much worse a bet than Cabrera. But pretty much all other situations he is, and it is very easy to see why with the rate data - less power, far fewer walks, etc. I contend it is easier to understand it when presented this way. JMO.

ESPN's Gamecast used to have something like this. It was AVG instead of percentage (move the decimal back two spots and add a '%' and you have your percentage). I liked it because it gave situations like "Miguel Cabrera is .350 with a runner on second with less than 2 outs" (made up numbers). I always thought those statistics were interesting to me.

[T&P_Fan]

I just watch the games.

[tiger337]

STLTiger69‎, This is by far one of the best observations of this subject that I've had the opportunity to read, great points and well thought out. Bravo! Now, the additional caveat, the day to day affects of the human condition on athletes, discussed widely among the sports medicine community, I would guess this would be next to impossible to quantify.

True, but if we are trying to answer the question: "Who made a greater contribution to his team's offense this year: Youkilis or Bourn", we don't care too much about their human condition. That kind of thing is more important to a manager making day-to-day decisions.

[Yoda]

I just watch the games.

That's unfortunate.

[John_Brian_K]

That's unfortunate.

Smug much?

[T&P_Fan]

That's unfortunate.

Speaking of unfortunate, I have yet to see the season premier of always sunny.

I know you were posting in that thread, so it just reminded me that I need to watch it when I get home from work.

[Yoda]

What's smug abut it? :/ I like a good Seinfeld reference.

[Yoda]

Speaking of unfortunate, I have yet to see the season premier of always sunny.

I know you were posting in that thread, so it just reminded me that I need to watch it when I get home from work.

Man, it starts slow, but takes off in a hurry! Yes, go watch it.

[Mr. Bigglesworth]

Thanks for the Sunny reminder.

[T&P_Fan]

What's smug abut it? :/ I like a good Seinfeld reference.

I like all Seinfeld references. Nothing worse than when you drop a good reference and the receiving party is picking up what you're laying down.

[tiger337]

Smug much?

"I just watch the games" could be just as smug as "That's unfortunate" but I don't think either one was in this case.

[sabretooth]

Somewhat off-topic, but I really wish they would break down the basic stats into percentages of plate appearances. For example:

2011 Miguel Cabrera:

In a given plate appearance, he:

Gets a hit 27.4% of the time - broken into:

1B - 16.7% of the time
2B - 6.7% of the time
3B - 0.0% of the time
HR - 4.0% of the time

He also gets a:

BB - 16.1% of the time and
HBP - 0.5% of the time

In total, he reaches base 44% of the time (27.4% of the time by hit + 16.1% by BB + 0.5% by HBP)

He makes an out 56% of the time, made up by:

K: 13.3% of the time
Fly Ball out: x.x%* of the time
Ground Ball out: x.x%* of the time
Line Drive out: x.x%* of the time

*I don't know where to locate this data, but I don't doubt it exists, and it isn't hard to convert it into a percentage.

I think that gives a much more accurate and easy to understand summary of what Miguel does at the plate.

That's a great point -- I for one get very frustrated with BB% and SO%, as well as GB/FB/Popup and HR/FB percentages, since they have different denominators. It makes for difficult head-calculations and problematic explainations.

[Nastradamus]

Stolen bases are important, but so are caught stealing numbers, as many have already pointed out. There are so few base stealers that have a high enough success rate that ops is generally good enough for evaluating a hitter. I look at base stealing as an added bonus, much like defense. There is no perfect stat. You have to consider a variety of factors.

Why does CS matter enough to devalue the stat? +1 total base for a SB, -1 for a CS and you have a pretty good idea. For example a guy like Ellsbury would gain 23 TBs. That matters.

[tiger337]

Why does CS matter enough to devalue the stat? +1 total base for a SB, -1 for a CS and you have a pretty good idea. For example a guy like Ellsbury would gain 23 TBs. That matters.

A caught stealing does more damage than a SB does good. For example, suppose there is a man on first and nobody out. If he steals second, he increases the team's expected runs in the inning by .24. If he is caught stealing, he decreases his team's expected runs in the inning by .37.

On average, a steal is worth 0.2 runs and a caught stealing costs 0.4 runs. So if a player steals 10 bases, he'll add an estimated two runs. If he is caught stealing 10 times, he'll cost his team about four runs.

Also, the average single is worth .47 runs, so a single is worth quite a bit more than a steal.

[EchO]

A caught stealing does more damage than a SB does good. For example, suppose there is a man on first and nobody out. If he steals second, he increases the team's expected runs in the inning by .24. If he is caught stealing, he decreases his team's expected runs in the inning by .37.

On average, a steal is worth 0.2 runs and a caught stealing costs 0.4 runs. So if a player steals 10 bases, he'll add an estimated two runs. If he is caught stealing 10 times, he'll cost his team about four runs.

Also, the average single is worth .47 runs, so a single is worth quite a bit more than a steal.

(For the record, I'm not being an ***)

Where did you get those %'s and if you could possibly provide any literature on how those %'s where figured, I would love to read it.

Snarky Aside/Edit:

We need to get Billy Hamilton from the Reds - 103 SB this season in A ball...crazy.

[tiger337]

(For the record, I'm not being an ***)

Where did you get those %'s and if you could possibly provide any literature on how those %'s where figured, I would love to read it.

Snarky Aside/Edit:

We need to get Billy Hamilton from the Reds - 103 SB this season in A ball...crazy.

It's based on the results of situations taken from thousands of games. It's pretty complicated and it takes several pages of my book. I can't find a free site which it explains the whole process very well, but I'll try to explain it briefly. You can find the probabilities for all the base out states here:

Tango’s Run Expectancy Matrix (1950-2010) « White Sox news, Minor Leagues updates and more

You can see (on the 1993-2010 chart) that when there is nobody out and bases empty, the average team can be expected to score .544 runs in the inning. If the next batter gets a single, there will be a runner at first and nobody out. Looking at the chart, the average team is expected to score .941 runs in that scenario. .941-.544 = .397. So, the average single with the bases empty and nobody out is worth .397 runs.

Not all singles are worth the same though. Suppose, there is a runner on first with one out. The next batter singles to put runners on first and third. Looking at the chart, that kind of single is worth 1.211-.562 = .649 on average. If you take all the averages of all the different kinds of singles, it comes out to about .47. That's called the linear weight for singles.

You can do the same kind of thing with doubles, stolen bases or any kind of event. This process is the basis of many of the saber stats you see - wOBA, WPA, etc.

[EchO]

It's based on the results of situations taken from thousands of games. It's pretty complicated and it takes several pages of my book. I can't find a free site which it explains the whole process very well, but I'll try to explain it briefly. You can find the probabilities for all the base out states here:

Tango’s Run Expectancy Matrix (1950-2010) « White Sox news, Minor Leagues updates and more

You can see (on the 1993-2010 chart) that when there is nobody out and bases empty, the average team can be expected to score .544 runs in the inning. If the next batter gets a single, there will be a runner at first and nobody out. Looking at the chart, the average team is expected to score .941 runs in that scenario. .941-.544 = .397. So, the average single with the bases empty and nobody out is worth .397 runs.

Not all singles are worth the same though. Suppose, there is a runner on first with one out. The next batter singles to put runners on first and third. Looking at the chart, that kind of single is worth 1.211-.562 = .649 on average. If you take all the averages of all the different kinds of singles, it comes out to about .47. That's called the linear weight for singles.

You can do the same kind of thing with doubles, stolen bases or any kind of event. This process is the basis of many of the saber stats you see - wOBA, WPA, etc.

That makes a lot of sense, you go over this in your book? I might have to buy the PDF version. If/when I do, is there a site that you would prefer me to buy from?

[STLTiger69]

It's based on the results of situations taken from thousands of games. It's pretty complicated and it takes several pages of my book. I can't find a free site which it explains the whole process very well, but I'll try to explain it briefly. You can find the probabilities for all the base out states here:

Tango’s Run Expectancy Matrix (1950-2010) « White Sox news, Minor Leagues updates and more

You can see (on the 1993-2010 chart) that when there is nobody out and bases empty, the average team can be expected to score .544 runs in the inning. If the next batter gets a single, there will be a runner at first and nobody out. Looking at the chart, the average team is expected to score .941 runs in that scenario. .941-.544 = .397. So, the average single with the bases empty and nobody out is worth .397 runs.

With all do respect (and I mean this very sincerely), the chart you mention doesn't have anything to do with stolen bases. The chart you quoted simply states "the average number of runs that scored, from that base/out state, to the end of that inning". Which means that if you have a runner at 1B with no outs the average runs you'll score per inning is .941 runs and if you have a runner at 2B and nobody out your average rises to 1.170 (if ONE runner scores from any base in any situation it still only counts for 1 run). This is the "average" numbers of runs that score accounting for an "average" of base runners in the major leagues.

I'm raising this issue because mixing base stealers into the entire MLB rosters of any player who's ever had an at bat from 1993-2010 included in this number simply devalues base stealing contributions as a whole. For 2011, there are only 43 players that have 20 or more stolen bases for the year. As a matter of fact, there are only 163 players that have currently 6 SB's or more for the ENTIRE 2011 season. These "base stealers" will have there base running skills averaged together with 1,100 or so other MLB players for the 2011 season to provide "the average number of runs that scored, from that base/out state, to the end of that inning". I also have a strong hunch that pitchers in the NL are also included in these numbers as well for their offensive contribution to "average runs that scored to an inning."

Just putting the numbers aside for a minute, who is more likely to score (even without a stolen base) from 1B with nobody out, Adam Dunn or Michael Bourn? Who will avoid being caught in a double play? Who is more likely to advance on a ground ball, tag up on a fly ball, advance a base on a ball in the dirt? Go from 1B to 3B on a single? Now put them both at 2B and nobody out, tell me which one is more likely to score? What about 3B and nobody out? -You get my drift.

[Mr. Bigglesworth]

With all do respect (and I mean this very sincerely), the chart you mention doesn't have anything to do with stolen bases. The chart you quoted simply states "the average number of runs that scored, from that base/out state, to the end of that inning". Which means that if you have a runner at 1B with no outs the average runs you'll score per inning is .941 runs and if you have a runner at 2B and nobody out your average rises to 1.170 (if ONE runner scores from any base in any situation it still only counts for 1 run). This is the "average" numbers of runs that score accounting for an "average" of base runners in the major leagues.

I'm raising this issue because mixing base stealers into the entire MLB rosters of any player who's ever had an at bat from 1993-2010 included in this number simply devalues base stealing contributions as a whole. For 2011, there are only 43 players that have 20 or more stolen bases for the year. As a matter of fact, there are only 163 players that have currently 6 SB's or more for the ENTIRE 2011 season. These "base stealers" will have there base running skills averaged together with 1,100 or so other MLB players for the 2011 season to provide "the average number of runs that scored, from that base/out state, to the end of that inning". I also have a strong hunch that pitchers in the NL are also included in these numbers as well for their offensive contribution to "average runs that scored to an inning."

Just putting the numbers aside for a minute, who is more likely to score (even without a stolen base) from 1B with nobody out, Adam Dunn or Michael Bourn? Who will avoid being caught in a double play? Who is more likely to advance on a ground ball, tag up on a fly ball, advance a base on a ball in the dirt? Go from 1B to 3B on a single? Now put them both at 2B and nobody out, tell me which one is more likely to score? What about 3B and nobody out? -You get my drift.

I get your drift, but I really don't think it is nearly the issue you suggest it is. Over the course of the season a good base runner will net a team some extra runs and as best we can, we should note this. But it should also be remembered that guys who are stealing bases are probably making more outs on the bases either trying to steal or take an extra base on a hit. That should count against them. On balance, my sense of it is the impact isn't that great. And there are metrics for evaluating the value of baserunning anyway. It isn't as if one can't inform their OPS evaluation while looking at baserunning data.

Similarly, you speak of concerns about run expectancy with a pitcher hitting versus a DH. Yes, the DH nets a team some more offense - no argument. But as a practical matter, the value of a single simply is not impacted all that much whether you are in the AL or NL.

[tiger337]

That makes a lot of sense, you go over this in your book? I might have to buy the PDF version. If/when I do, is there a site that you would prefer me to buy from?

you can get it here:
tiger337's Books and Publications Spotlight

[tiger337]

With all do respect (and I mean this very sincerely), the chart you mention doesn't have anything to do with stolen bases. The chart you quoted simply states "the average number of runs that scored, from that base/out state, to the end of that inning". Which means that if you have a runner at 1B with no outs the average runs you'll score per inning is .941 runs and if you have a runner at 2B and nobody out your average rises to 1.170 (if ONE runner scores from any base in any situation it still only counts for 1 run). This is the "average" numbers of runs that score accounting for an "average" of base runners in the major leagues.

I'm raising this issue because mixing base stealers into the entire MLB rosters of any player who's ever had an at bat from 1993-2010 included in this number simply devalues base stealing contributions as a whole. For 2011, there are only 43 players that have 20 or more stolen bases for the year. As a matter of fact, there are only 163 players that have currently 6 SB's or more for the ENTIRE 2011 season. These "base stealers" will have there base running skills averaged together with 1,100 or so other MLB players for the 2011 season to provide "the average number of runs that scored, from that base/out state, to the end of that inning". I also have a strong hunch that pitchers in the NL are also included in these numbers as well for their offensive contribution to "average runs that scored to an inning."

Just putting the numbers aside for a minute, who is more likely to score (even without a stolen base) from 1B with nobody out, Adam Dunn or Michael Bourn? Who will avoid being caught in a double play? Who is more likely to advance on a ground ball, tag up on a fly ball, advance a base on a ball in the dirt? Go from 1B to 3B on a single? Now put them both at 2B and nobody out, tell me which one is more likely to score? What about 3B and nobody out? -You get my drift.

Without numbers, the problem with giving a player the same credit for both a SB and a 1B is this: A stolen base always adds just one base to a team's attempt to score a run. A single contributes at least one base, but often contributes more. If there are runners on base, the single moves them up a base or two. So, a single has to get more weight than a SB. Otherwise, your stat becomes biased in favor of base stealers.

A CS is bad because it does two damaging things: It removes a base runner and also adds another out to the inning. Thus, it is doubly bad as a stolen base is good.

As for the linear weights, it is true that not every situation is an average situation. For example, if a player on a good hitting team gets a hit, he's more likely to score than if he's on a bad hitting team. Still, using the average is going to get you closer to true run contribution than arbitrarily assigning a one to SB, CS and 1B. Regardless of how fast a runner is, caught stealings still cost a lot more than stolen bases gain. Any statistics which include them should reflect that fact.

As for your other concern about a fast runner being more likely to score, they have separate statistics to measure base running contribution. I would prefer to use those. You are the one that proposed mixing base running with batting. I was offering a better alternative if you did want to do that. Your proposal was to give an arbitrary value of one to 1B, SB and CS which doesn't work well and also doesn't address your criticism of the concept of average.

[Nastradamus]

A caught stealing does more damage than a SB does good. For example, suppose there is a man on first and nobody out. If he steals second, he increases the team's expected runs in the inning by .24. If he is caught stealing, he decreases his team's expected runs in the inning by .37.

On average, a steal is worth 0.2 runs and a caught stealing costs 0.4 runs. So if a player steals 10 bases, he'll add an estimated two runs. If he is caught stealing 10 times, he'll cost his team about four runs.

Also, the average single is worth .47 runs, so a single is worth quite a bit more than a steal.

I suppose the question isnt whether a steal is worth more than a single though, its whether a single and steal is worth more or less than a double.

Its hard for me to put in perspective the .11 change in expected runs scored admittedly. Still, even if it wasn't involved in slugging percentage, I'd like to see a stat that factors these things in. No matter how you look at it, a steal is not worth zero to your team.