Originally posted by bluesky5
View Post
Announcement
Collapse
No announcement yet.
Progressing Toward Better Stats Thread
Collapse
X

"It ain't braggin' if you can do it." Dizzy Dean

Originally posted by Sultan_18951948 View PostSo add them both and subtract 1? I need to keep working on the career chart, but the top fives are more fun!
Why the minus 1?
That all rhymed.
The thing is that it is close to directly proportional with a players batting value. Not perfect, but if we know that a player had a 155 OPS+, but his road numbers suggest he should have been 170, then it means he could have been 70% above average rather than 55%, or at least 70/55 as good. (again OPS+ is not perfect, but the ratio of road predicted compared to overall for a single given player is going to be very close to the ratio of expected offense by that player based on his road numbers).
What I would really like to see, as an eventual outcome, is a single number for each player that is based on road numbers, and relative to his league that tells us what his "road predicted" batting production really was. So we can say, based on road relative numbers, Boggs is really only 116% of average instead of 130%, while Mathews is really 155% instead of 143%.
If we get that road relative OPS+ we can just directly compare it to the players total OPS+ to see how much he might have risen or fallen in a neutral park (not an offense neutral park, but one that didn't either hurt that player or help him in particular.
We could either use the road expected OPS+, or we could use it with our total batting value stat (let me look it up it was the Rbat one) to adjust out his particular home benefit or loss.
Comment

Oh no! I forgot something. League slugging, well, you are taking the league totals right? I forgot that pitchers are usually taken out when we do the relative numbers. This may explain a little of why newer players look a little worse in the AL. You don't actually HAVE to take pitchers out, I mean what we are looking at is real, and well WARs Rbat is ultimately based on league offense, not just position players' offense.
But to take it out, the best way to get real close is to take the league OPS+ for the player, (the .94 something that you got in earlier work), but instead of using that, you use 1 + OPS+ all divided by 2. So if a player's league was .938 you would take (1+.9380/2) which would be 0.969. If you divide the actual league slugging by that number you would get it without pitchers to within a hair. Same for OB%. It wouldn't affect guys in leagues where pitchers didn't hit, but it would drop guys in leagues where they did hit by about 3%. I am sorry, I lost track that that step was happening.
The other option though, what I would probably do is leave it "as is". It is a real ratio of the players slugging to his league, but if we get to the point where we create an OPS+ out of it, we would need to just multiply the player's predicted OPS+ by the league OPS+ to take out pitchers from there.Last edited by brett; 05252014, 05:24 PM.
Comment

Originally posted by brett View PostOh no! I forgot something. League slugging, well, you are taking the league totals right? I forgot that pitchers are usually taken out when we do the relative numbers. This may explain a little of why newer players look a little worse in the AL. You don't actually HAVE to take pitchers out, I mean what we are looking at is real, and well WARs Rbat is ultimately based on league offense, not just position players' offense.
But to take it out, the best way to get real close is to take the league OPS+ for the player, (the .94 something that you got in earlier work), but instead of using that, you use 1 + OPS+ all divided by 2. So if a player's league was .938 you would take (1+.9380/2) which would be 0.969. If you divide the actual league slugging by that number you would get it without pitchers to within a hair. Same for OB%. It wouldn't affect guys in leagues where pitchers didn't hit, but it would drop guys in leagues where they did hit by about 3%. I am sorry, I lost track that that step was happening.
The other option though, what I would probably do is leave it "as is". It is a real ratio of the players slugging to his league, but if we get to the point where we create an OPS+ out of it, we would need to just multiply the player's predicted OPS+ by the league OPS+ to take out pitchers from there.
DiMaggio would be 59% (SLG) + 19% (OBP) = 78% Adjusted OPS advantage over league
So we ignore park effects with this stat, and use road relative OPS+ to adjust for park effects?
Comment

Originally posted by layson27 View PostI think "As is" makes sense, but it's Sultan's stat of course. It's what % a player is above league avg. OPS, correct?
DiMaggio would be 59% (SLG) + 19% (OBP) = 78% Adjusted OPS advantage over league
So we ignore park effects with this stat, and use road relative OPS+ to adjust for park effects?
I would not adjust relative slugging or relative on base percentage, in part because the accuracy in the adjustment is not perfect. But Dimaggio's real OPS+ was 155 and that had pitchers factored out, so to get a fair comparison they should be taken out of the OPS+.
Comment

Originally posted by brett View PostYes, but the final OPS+ of 178 should be multiplied by the league average OPS+ which I think was about .94. So he's more like 167. It still moves him into the elite high 160s range.
I would not adjust relative slugging or relative on base percentage, in part because the accuracy in the adjustment is not perfect. But Dimaggio's real OPS+ was 155 and that had pitchers factored out, so to get a fair comparison they should be taken out of the OPS+.Last edited by layson27; 05252014, 06:41 PM.
Comment

Originally posted by layson27 View PostKeeping in mind Joe D's actual road OPS+ is 165, I'd say that's a very close approximation.
We found a player's (road) slugging and on base relative to the actual league numbers and then adjusted pitchers out and
We took road sOPS+ (with pitchers) season by season, weighted by plate appearances and then factored pitchers out.
And I think the league OPS+ was more like .935 which would give us 166.4
There is a tiny percent we didn't account for and that is that he should have hit in YS on the road a fraction of the time, and there's your other point and a fraction.
Comment

Originally posted by brett View PostSee it works! We just calculated OPS+ two different ways and got the same answer.
We found a player's (road) slugging and on base relative to the actual league numbers and then adjusted pitchers out and
We took road sOPS+ (with pitchers) season by season, weighted by plate appearances and then factored pitchers out.
And I think the league OPS+ was more like .935 which would give us 166.4
There is a tiny percent we didn't account for and that is that he should have hit in YS on the road a fraction of the time, and there's your other point and a fraction.
And as I recall Williams' & Gehrig's road OPS+ were almost identical as well.
Comment

I'm confused.
Multiple directions we're going in now, but I wanna make sure, this one direction will lead to something.
Bret...if I complete the right side of the chart I just posted, the one having to do with OBP, we will have ADJUSTED RELATIVE, or I guess just relative, SA and OBP. What can you do with those two final numbers?
Also...for SLGeff.....the final number doesn't look clean or easily understood. Is there a way to put it into BA form or something?
Comment

Originally posted by Sultan_18951948 View PostI'm confused.
Multiple directions we're going in now, but I wanna make sure, this one direction will lead to something.
Bret...if I complete the right side of the chart I just posted, the one having to do with OBP, we will have ADJUSTED RELATIVE, or I guess just relative, SA and OBP. What can you do with those two final numbers?
Also...for SLGeff.....the final number doesn't look clean or easily understood. Is there a way to put it into BA form or something?
It would be "road predicted OPS+" or just predicted OPS+
So a 1.63 relative slugging 1.16 relative OB%, you get 2.78. Subtract 1 to get 1.78. Multiply by the league OPS+ say .941 to get 1.67498. Then multiply 100 to get 167 (here you should round off to three digits.
For SLGeff let me get back to you in 5 minutes.
Comment

For slugging efficiency, on the one hand, I think people who have looked at stats of players know that if you can put up more than 1 total base per out, you are having a great season. For example in '98 McGwire had 383 total bases and 357 outs.
In the past I have recommened not using rates per 27 outs because a player doesn't get 27 outs. If we though assume that a team has 459 outs per spot in the lineup in a 162 game season (162 x 25.5/9) we could multiply slugging efficiency by 459. Then it would look like a total base total for a full season. Total bases per 459 outs. 400 is awesome. Or if you double it it would be similar to an OPS+ score and probably mean about the same thing.
Comment

Originally posted by brett View PostFor slugging efficiency, on the one hand, I think people who have looked at stats of players know that if you can put up more than 1 total base per out, you are having a great season. For example in '98 McGwire had 383 total bases and 357 outs.
In the past I have recommened not using rates per 27 outs because a player doesn't get 27 outs. If we though assume that a team has 459 outs per spot in the lineup in a 162 game season (162 x 25.5/9) we could multiply slugging efficiency by 459. Then it would look like a total base total for a full season. Total bases per 459 outs. 400 is awesome. Or if you double it it would be similar to an OPS+ score and probably mean about the same thing.
I like how you're thinking outside the box.
The first way makes sense, except the subtracting 1 and all that, but it's probably a stat thing over my head.
For the SLGeff....I would really like to keep it to what the player and league did. Not sure about extrapolating that out to a whole season worth of outs for the one player. I do like the idea of getting a TB type figure but wouldn't we divide that by their actual AB to get something else? Or are we then back and square one.
Man, how does this stuff not give you a headache.
I'm through 1940 for AL and NL on the new SLGeff chart, since I have to redo it. Only Ruth done so far...his Adjusted relative is 2.04. Very close to his actual OPS+
Comment

Originally posted by Sultan_18951948 View PostAwesome Brett.
I like how you're thinking outside the box.
The first way makes sense, except the subtracting 1 and all that, but it's probably a stat thing over my head.
For the SLGeff....I would really like to keep it to what the player and league did. Not sure about extrapolating that out to a whole season worth of outs for the one player. I do like the idea of getting a TB type figure but wouldn't we divide that by their actual AB to get something else? Or are we then back and square one.
Man, how does this stuff not give you a headache.
I'm through 1940 for AL and NL on the new SLGeff chart, since I have to redo it. Only Ruth done so far...his Adjusted relative is 2.04. Very close to his actual OPS+
Comment

Originally posted by bluesky5 View PostThe numbers. I certainly noticed the tinted pictures though. They make the cluttered effect you get with all the numbers on a spreadsheet dissipate completely. A very nice touch. :applaud:
Not sure if you were joking or not but I was shocked he beat him so handily too.
Interesting that in Musial's career he hit more HR: +29 and doubles: +61 at home than on the road but in his five best seasons he was way better on the road [except '51]. Can someone help me out here: did they change the dimensions at Sportsman's Park during Musial's tenure?
Back to the main point. This is some very interesting data. keep up the good work.
Comment
Ad Widget
Collapse
Comment