One thing I am finding is that Win Shares is putting the value of my closer through the roof. I have looked at three teams now.
Using a modification of this method to fit the numbers I have
http://walksaber.blogspot.com/2005/12/win-shares-walkthrough-pt-5.html
The only major limitation the available stats present with this method is the lack of holds. So it is bound to overstate closers over other relievers.
It may be a coincidence of how I build teams, or the teams I have chosen to look at but here's something that is emerging:
First team: No closer Sambito leads team in saves with 10, Babe Adams leads team with 26.5 WS (kicked in 10 relief appearances including 8-1 saves)
Second team: 100M theme league (two pitchers over 7M allowed, 2 batters over 8M, 4 stars must consume specific proportion of 100M cap) Radatz takes home 27.32 WS by going 9-12 with 34 saves, 15 BS, 130 IP, team leading 3.60 ERA and team leading WHIP of 1.29. Schilling 2001 goes 16-16, throws 282.1 IP, first among starters in IP, ERA, WHIP, and gets 17.4 WS. Cliff Lee goes 20-9, second in innings, third in ERA and WHIP and gets 13.8.
Third Team: 140M theme league:
Ted Lyons takes the most (22-15, 325.1 IP, 4.76 ERA, 1.38 WHIP) with 10.4 WS. Miljus takes second place with 13-7 17 saves, 9 blows, 164 IP, 4.72 ERA, 1.48 WHIP. He is second best pitcher on team.
Fourth team: 100M random parks league
Sutter is godlike (15-11, 47 sv, 15 bs 190.1 IP 90 games, 3.12 ERA 1.03 WHIP) for 21.9 WS
Mussina is second (19-11, 267.2 IP, 4.34 ERA, 1.42 WHIP) for 9.11.
The lack of HOLDS as a stat skews the bullpen towards a big closer. In this case, Randy Myers, who threw 161.2 IP out of the pen is left with 4 WS despite appearing in 91 games and earning who knows how many holds.
What I'm getting at here is that where I have a closer who is even reasonably shut down, he takes an inordinate number of Win Shares. I understand the concept of leverage, but this seems a touch broken.
(I have also looked at adding a ding for a blown save, which mitigates it somewhat and redistributes WS from the closer to the rotation, but Sutter is still number one on this team by a wide margin... of course, if I eliminate the the third claim factor , he STILL has the top rating in the other two claim factors. Lot closer though... something to mull here.
Evaluating Your Team - Post Facto
20 posts
• Page 2 of 2 • 1, 2
by PotKettleBlack » Tue Jan 10, 2012 5:14 pm
- PotKettleBlack
- Posts: 55
- Joined: Tue Jul 03, 2012 2:34 pm
by PotKettleBlack » Wed Jan 11, 2012 1:20 pm
Have been going back to DeanTSC's Offense Vs. Defense Article, the first section. 108,216,324,648,5832 - Oh My!
In the article, he uses 648 PA to value a batter, which is good for projection (touch low for ATG at 100M+), but unnecessary for reflection.
In each PA: 50% chance the hitter card determines the outcome, 36.1% the pitcher controls it, and 13.9% chance of an X-roll leading to a fielder taking responsibility. Good for offense. Not great for defense, as a full time fielder will see ~9 times as many plays* in the field as at the plate.
In projection, Dean suggests that 5832 plays in the field is a good number (648*9). Since I'm looking at reflection, I can actually work this from something like a real number ((IP*3)+H+BB+HBP-CS-DP). Working with the OvD framework, however: SS sees ~189 X-rolls, 2B sees 162, 3B & CF see 81 a piece, LF/RF/1B/P see 54. You could break out expected IP for the pitchers to predict how many each will get. In ATG, it's actually a little more than that, as the team number can be higher (offensive quality increases PA).
What to make of this.
1- Vizquel. Vizquel Vizquel Vizquel. Figure in ATG 100, he sees ~200 X-rolls at 1e4, and you can put a run number on that towards wins that really makes up for his bat. Especially if you bat him 8th or 9th, where his lack of hitting will hurt a lot less.
2- ARod. Doesn't he have a 1e8 SS that everyone hates because it doesn't hit like a 10M card? Again, that's ~200 plays with 8 errors and a minimum of hits allowed. Oh, and you don't have to bat him 9th.
3- I think it's interesting that people will throw just about anyone out there at 3rd, but will limit CF to only the select few. I do that. Granted, OF errors/range problems are more costly. But something worth considering... 3B is an important D position.
4- I think I'd like to expand Deans work to catchers, at least for range/error ratings. Obviously, there is a lot more complexity (throwing errors, passed balls, etc).
*A leadoff hitter will have more PA than the light hitting middle infielder who bats last. In fact, (assuming similar injury patters) a leadoff hitter will have more PA than the #2 hitter, the #2 will have more than the #3, and so on. Something to be aware of, as that slick fielding, no hit middle infielder who bats 9th, may have the value tilted further in favor of the glove if you bat him 9th and field him daily. More later.
In the article, he uses 648 PA to value a batter, which is good for projection (touch low for ATG at 100M+), but unnecessary for reflection.
In each PA: 50% chance the hitter card determines the outcome, 36.1% the pitcher controls it, and 13.9% chance of an X-roll leading to a fielder taking responsibility. Good for offense. Not great for defense, as a full time fielder will see ~9 times as many plays* in the field as at the plate.
In projection, Dean suggests that 5832 plays in the field is a good number (648*9). Since I'm looking at reflection, I can actually work this from something like a real number ((IP*3)+H+BB+HBP-CS-DP). Working with the OvD framework, however: SS sees ~189 X-rolls, 2B sees 162, 3B & CF see 81 a piece, LF/RF/1B/P see 54. You could break out expected IP for the pitchers to predict how many each will get. In ATG, it's actually a little more than that, as the team number can be higher (offensive quality increases PA).
What to make of this.
1- Vizquel. Vizquel Vizquel Vizquel. Figure in ATG 100, he sees ~200 X-rolls at 1e4, and you can put a run number on that towards wins that really makes up for his bat. Especially if you bat him 8th or 9th, where his lack of hitting will hurt a lot less.
2- ARod. Doesn't he have a 1e8 SS that everyone hates because it doesn't hit like a 10M card? Again, that's ~200 plays with 8 errors and a minimum of hits allowed. Oh, and you don't have to bat him 9th.
3- I think it's interesting that people will throw just about anyone out there at 3rd, but will limit CF to only the select few. I do that. Granted, OF errors/range problems are more costly. But something worth considering... 3B is an important D position.
4- I think I'd like to expand Deans work to catchers, at least for range/error ratings. Obviously, there is a lot more complexity (throwing errors, passed balls, etc).
*A leadoff hitter will have more PA than the light hitting middle infielder who bats last. In fact, (assuming similar injury patters) a leadoff hitter will have more PA than the #2 hitter, the #2 will have more than the #3, and so on. Something to be aware of, as that slick fielding, no hit middle infielder who bats 9th, may have the value tilted further in favor of the glove if you bat him 9th and field him daily. More later.
- PotKettleBlack
- Posts: 55
- Joined: Tue Jul 03, 2012 2:34 pm
by Valen » Thu Jan 12, 2012 9:27 am
Some interesting stuff here. Some of it sounds a bit time consuming.
That makes me wonder how much time those who run 10-20 teams at a time bother spending analyzing finished teams.
It might also seem some of this would be better done after about 30 games when there is time to make changes. If you have followed your team throughout the season how much are you really likely to gain from a postseason analysis that you should not already know?
Back when I started in the 2001 game I did a lot of postseason analysis. But back then I did not play as many teams at once. probably one reason I was more successful back then. Each team got more attention. I often spent as much time analyzing other teams as I did my own. Did they use and get good results from players I would not have considered using? What platoons were they using?
I suppose you could boil it down to seeing if they had found any value cards I had missed. That was more important in the 200x games where the cards changed every year even though the names on cards might be the same. less need of that in ATG as the player pool does not change and any given league is not likely to reveal any new values. Focus shifts from who the values are to where they need to be on your draft card to maximize chances of getting.
That makes me wonder how much time those who run 10-20 teams at a time bother spending analyzing finished teams.
It might also seem some of this would be better done after about 30 games when there is time to make changes. If you have followed your team throughout the season how much are you really likely to gain from a postseason analysis that you should not already know?
Back when I started in the 2001 game I did a lot of postseason analysis. But back then I did not play as many teams at once. probably one reason I was more successful back then. Each team got more attention. I often spent as much time analyzing other teams as I did my own. Did they use and get good results from players I would not have considered using? What platoons were they using?
I suppose you could boil it down to seeing if they had found any value cards I had missed. That was more important in the 200x games where the cards changed every year even though the names on cards might be the same. less need of that in ATG as the player pool does not change and any given league is not likely to reveal any new values. Focus shifts from who the values are to where they need to be on your draft card to maximize chances of getting.
- Valen
- Posts: 54
- Joined: Tue Jul 03, 2012 2:34 pm
by PotKettleBlack » Thu Jan 12, 2012 12:28 pm
Okay, here's an interesting (somewhat time consuming) idea on catcher value (Stolen from Tom Tango). With or Without You (or WOWY).
What you need is:
1- a catcher and a backup who have varying levels of defensive skill
2- a starter who missed a decent amount of time.
3- a list of the exact games he was injured.
I just ran this for one of my teams.
http://fantasygames.sportingnews.com/stratomatic/team/team_other.html?user_id=331178
I tossed out the games where O'Farrell was injured (as I didn't want to overly complicate things) and compared games where O'Farrell played the whole game (or was replaced defensively) to ones Ken Rudolph played the whole game (or was replaced stupidly by HAL with a non catcher).
With O'Farrell behind the dish, 5.57 runs/game.
With Rudolph, 4.04 runs/game.
This is overstated as Rudolph happened to catch a lot of his games against the weakest offensive teams, but where I can make head-to-head comparisons on smaller sample sizes on the same teams, the effect is still that Rudolph is somewhere between a little and a lot better. Against a small ball team (worst offensive team in the league) in 3 games, Team w/ Rudolph 3.33 RA, With O'F: 9 games, 6.11 RA. (this opponent stole 300+ bases on the season, so probably robbed O'Farrell blind. Conversely, this team stole at a .656 percent, so probably against Rudolph... As with everything, context matters.
Some deeper digging:
O'Farrell saw 142 stolen bases, caught 45, committed 13 errors, and passed the lion's share (if not all) of 12 passed balls.
Rudolph allowed 10 steals, caught 8, with no errors and likely no passed balls.
That said, since I had all the numbers together, I checked how we scored (since Rudolph's bat is nigh nonexistent)... O'Farrell 6.62/g. Rudolph 5.2/g. (above caveat about sample size and sample skew for Rudolph's opponents).
Would have to look at a more injury prone catcher or fielder to do more WoWY analysis. And it doesn't bring me closer to a valuation.
What you need is:
1- a catcher and a backup who have varying levels of defensive skill
2- a starter who missed a decent amount of time.
3- a list of the exact games he was injured.
I just ran this for one of my teams.
http://fantasygames.sportingnews.com/stratomatic/team/team_other.html?user_id=331178
I tossed out the games where O'Farrell was injured (as I didn't want to overly complicate things) and compared games where O'Farrell played the whole game (or was replaced defensively) to ones Ken Rudolph played the whole game (or was replaced stupidly by HAL with a non catcher).
With O'Farrell behind the dish, 5.57 runs/game.
With Rudolph, 4.04 runs/game.
This is overstated as Rudolph happened to catch a lot of his games against the weakest offensive teams, but where I can make head-to-head comparisons on smaller sample sizes on the same teams, the effect is still that Rudolph is somewhere between a little and a lot better. Against a small ball team (worst offensive team in the league) in 3 games, Team w/ Rudolph 3.33 RA, With O'F: 9 games, 6.11 RA. (this opponent stole 300+ bases on the season, so probably robbed O'Farrell blind. Conversely, this team stole at a .656 percent, so probably against Rudolph... As with everything, context matters.
Some deeper digging:
O'Farrell saw 142 stolen bases, caught 45, committed 13 errors, and passed the lion's share (if not all) of 12 passed balls.
Rudolph allowed 10 steals, caught 8, with no errors and likely no passed balls.
That said, since I had all the numbers together, I checked how we scored (since Rudolph's bat is nigh nonexistent)... O'Farrell 6.62/g. Rudolph 5.2/g. (above caveat about sample size and sample skew for Rudolph's opponents).
Would have to look at a more injury prone catcher or fielder to do more WoWY analysis. And it doesn't bring me closer to a valuation.
- PotKettleBlack
- Posts: 55
- Joined: Tue Jul 03, 2012 2:34 pm
by BDWard » Thu Jan 12, 2012 1:01 pm
PKB: I've been watching this thread with great interest. Your analysis and discussion has been very enlightening. Thank you so much.
Anyway, although I strongly suspect that you have the following user script, I've attached a link below for a user script that will allow you to glean more information more quickly from box scores (as if you don't already have enough to do!).
Although I'm no computer whiz, the script can probably be easily modified to search for wild pitches, passed balls and any other occurrence of interest that's not usually tracked.
I hope this helps in your analysis.
http://userscripts.org/scripts/show/112867
Anyway, although I strongly suspect that you have the following user script, I've attached a link below for a user script that will allow you to glean more information more quickly from box scores (as if you don't already have enough to do!).
Although I'm no computer whiz, the script can probably be easily modified to search for wild pitches, passed balls and any other occurrence of interest that's not usually tracked.
I hope this helps in your analysis.
http://userscripts.org/scripts/show/112867
- BDWard
- Posts: 55
- Joined: Tue Jul 03, 2012 2:34 pm
by Valen » Thu Jan 12, 2012 8:20 pm
That is some good analysis on those catchers, very impressive.
I expect if I did more analysis like that on my teams my winning percentage would go up. But then that also sounds like a lot of work if you broke every position on every team down like that. Not sure I even work that hard on the job I get paid for. :lol:
I expect if I did more analysis like that on my teams my winning percentage would go up. But then that also sounds like a lot of work if you broke every position on every team down like that. Not sure I even work that hard on the job I get paid for. :lol:
- Valen
- Posts: 54
- Joined: Tue Jul 03, 2012 2:34 pm
by PotKettleBlack » Thu Jan 12, 2012 9:22 pm
[quote:45425dd38f="Valen"]That is some good analysis on those catchers, very impressive.
I expect if I did more analysis like that on my teams my winning percentage would go up. But then that also sounds like a lot of work if you broke every position on every team down like that. Not sure I even work that hard on the job I get paid for. :lol:[/quote:45425dd38f]
You need a starter who missed some time. In the above example, O'Farrell was injured a lot, but only missed 25 whole games. I'd feel better with every game above 30. An interesting idea would be to take two cheapie catchers, one with high injury and a glove like O'Farrell's (2esomething +1 arm, T6, PB5 or something like that) and Rudolph and do WOWY. Alternatively, platoon runs against... I use Boone/Walling on a bunch of teams that might produce something... I know a lot of people use Galaraga who is outstanding defensively at a spot that is less than super relevant, and suspect they balance him with someone who is a lesser defender... of course, DeanTSC captured the D value at various positions pretty well in the OvDv2 article, just no catcher value, so that's an area of focus.
Need to develop some kind of linear weight for regular errors (as opposed to throwing errors) and passed balls, and maybe a value for range at C, not including play at the plate value (seems based on the odds it wouldn't come up much). Then, arm as deterrent. How many runners does Pudge dissaude from even attempting to run? There is a TON of stuff to look at with catchers. I could drive myself insane.
I expect if I did more analysis like that on my teams my winning percentage would go up. But then that also sounds like a lot of work if you broke every position on every team down like that. Not sure I even work that hard on the job I get paid for. :lol:[/quote:45425dd38f]
You need a starter who missed some time. In the above example, O'Farrell was injured a lot, but only missed 25 whole games. I'd feel better with every game above 30. An interesting idea would be to take two cheapie catchers, one with high injury and a glove like O'Farrell's (2esomething +1 arm, T6, PB5 or something like that) and Rudolph and do WOWY. Alternatively, platoon runs against... I use Boone/Walling on a bunch of teams that might produce something... I know a lot of people use Galaraga who is outstanding defensively at a spot that is less than super relevant, and suspect they balance him with someone who is a lesser defender... of course, DeanTSC captured the D value at various positions pretty well in the OvDv2 article, just no catcher value, so that's an area of focus.
Need to develop some kind of linear weight for regular errors (as opposed to throwing errors) and passed balls, and maybe a value for range at C, not including play at the plate value (seems based on the odds it wouldn't come up much). Then, arm as deterrent. How many runners does Pudge dissaude from even attempting to run? There is a TON of stuff to look at with catchers. I could drive myself insane.
- PotKettleBlack
- Posts: 55
- Joined: Tue Jul 03, 2012 2:34 pm
20 posts
• Page 2 of 2 • 1, 2
Return to Strat-O-Matic Baseball: All-Time Greats
Who is online
Users browsing this forum: No registered users and 14 guests