Best Starters

[DonkeyKongSr]

Originally Posted by max venerabel

I just paid 12.5K for him. Definitely an impulse purchase.

In my Perfect League, my Keuchel has a 3.21 ERA, is 5th in FIP, and has given up 0 HR vs. 426 batters.

The problem is he somehow stops both teams from scoring, getting 2.5 Run Support for a 4-7 record. My team average 4.5 runs per game. He just lost his last game 1-0. Unreal.

[SayHey1]

Originally Posted by Orcin

He's been terrible for me since I bought him.

He's 8-4 for me with a 3.41 ERA and a WHIP of 1.36 in Diamond League.


My 2nd best starter this year.


My best? 89 Corbin with a 9-0 3.13 ERA and WHIP of 1.16


Go figure.

[ASURay]

Originally Posted by DonkeyKongSr

Blake lost 8 games last season, but otherwise I guess he wasn't that bad when I pulled him as closer. He ended at a 117 ERA+ with a 3.18 FIP.

Anyway, I started trying out 2004 All-Star Mariano Rivera there. Even though he's an 86, I have them both in the Top 10 relievers. It's the best Rivera card, even better than the 93-rated card. But, that hasn't worked out so well.

This may seem crazy, but now I'm going to try out Chris Sale as Closer. He's just very average as a starter in Perfect and I've grabbed a few bargains from my rankings. My currrent starting rotation is:

Jacob deGrom (100)
Randy Johnson (97)
Dallas Keuchel (86)
Dizzy Trout (91) - got him late last night
Dizzy Dean (88)

Quite a dizzying rotation...hehehe

If one of the Dizzys don't work out, I'll probably drop Sale back in and go with Treinan at closer again.

The 2004 Rivera card is, IMO the most underrated RP/CL card in the game. It still goes for 6K+, but it's ratings make it prb a Top 5 RP/CL card, in theory. The one I have is up and down but generally quite good.

[ASURay]

Originally Posted by pappyzan

Are individual pitches or pitch ratings taken into consideration with any of these models/equations?

Not currently. The individual pitches are mostly important for determining the STUFF, STUFF vs LH, and STUFF vs RH ratings. I've never gotten any confirmation as to whether or not they do anything beyond that. For example, I don't think the in-game probability of a hitter putting the ball on the ground is any higher for a low sinker than it is for a high fastball or that the probability of a swing-and-miss is any higher if the pitcher throws their 95 FB as opposed to their 35 CB. I have always assumed that the pitch selection and location shown when you're playing out games is just random and doesn't have an effect. I think all of the outcomes are determined by the Stuff/Mov/Con ratings. I think the pitches are there for window dressing. I'm guessing the GB/FB tendency would have an impact, but I haven't modeled it yet. It would add a lot of variables to the model. More parameters to estimate requires a larger sample size. The biggest factor that my model doesn't account for is platoon splits. I'm just predicting overall FIP. We're seeing that LHP, for example tend to suffer in Perfect leagues quite a bit. An easy next step would be to add STF/MOV/CON vs LH/RH to my database and then just apply the same formula to get a rough idea of predicted FIP vs LH/RH. Chris Sale is a great example here. He gets torched by RH in Perfect leagues but does well vs LH. I've moved him to the bullpen and gone with an all-RHP rotation.

[Orcin]

Originally Posted by SayHey1

He's 8-4 for me with a 3.41 ERA and a WHIP of 1.36 in Diamond League.


My 2nd best starter this year.


My best? 89 Corbin with a 9-0 3.13 ERA and WHIP of 1.16


Go figure.

Yeah, and this year Blanton has suddenly become my #2 starter.

[Kushiel]

Originally Posted by ASURay

Not currently. The individual pitches are mostly important for determining the STUFF, STUFF vs LH, and STUFF vs RH ratings. I've never gotten any confirmation as to whether or not they do anything beyond that. For example, I don't think the in-game probability of a hitter putting the ball on the ground is any higher for a low sinker than it is for a high fastball or that the probability of a swing-and-miss is any higher if the pitcher throws their 95 FB as opposed to their 35 CB. I have always assumed that the pitch selection and location shown when you're playing out games is just random and doesn't have an effect. I think all of the outcomes are determined by the Stuff/Mov/Con ratings. I think the pitches are there for window dressing. I'm guessing the GB/FB tendency would have an impact, but I haven't modeled it yet. It would add a lot of variables to the model. More parameters to estimate requires a larger sample size. The biggest factor that my model doesn't account for is platoon splits. I'm just predicting overall FIP. We're seeing that LHP, for example tend to suffer in Perfect leagues quite a bit. An easy next step would be to add STF/MOV/CON vs LH/RH to my database and then just apply the same formula to get a rough idea of predicted FIP vs LH/RH. Chris Sale is a great example here. He gets torched by RH in Perfect leagues but does well vs LH. I've moved him to the bullpen and gone with an all-RHP rotation.

The OVR 96 Kershaw is the only LH I have found that starts well for me in perfect. This is my second season with him and his ERA has been staying in the 3's. I do not know why he translates but he seems to so far.
He is allowing me to keep "The Big Train" as my closer.

[DonkeyKongSr]

Originally Posted by ASURay

Chris Sale is a great example here. He gets torched by RH in Perfect leagues but does well vs LH. I've moved him to the bullpen and gone with an all-RHP rotation.

He's now given up 1 run in my 2 months with him as my Perfect League closer. This from a guy with 5+ ERA as a starter in the first couple PL seasons.

My 97 Randy Johnson through, he can pitch as a starter. Haven't seen many other LH starter successes than him and his perfect version.

[chazzycat]

Originally Posted by ASURay

For example, I don't think the in-game probability of a hitter putting the ball on the ground is any higher for a low sinker than it is for a high fastball or that the probability of a swing-and-miss is any higher if the pitcher throws their 95 FB as opposed to their 35 CB

I don't think this is true. Try going into commish mode and add a good sinker to any pitcher. you should see the MOV rating go up.

[DonkeyKongSr]

Originally Posted by chazzycat

I don't think this is true. Try going into commish mode and add a good sinker to any pitcher. you should see the MOV rating go up.

But that just means it already factors into Movement so the individual pitch itself isn't likely relevant in your choice of pitcher.

[chazzycat]

Yes that is true. But that is different from "the pitches don't matter and are just window dressing".

[chazzycat]

Nevermind...ignore me. I could have sworn I saw this effect before, but I just tried again and was not able to recreate it. Sorry for the confusion

[max venerabel]

Originally Posted by DonkeyKongSr

In my Perfect League, my Keuchel has a 3.21 ERA, is 5th in FIP, and has given up 0 HR vs. 426 batters.

The problem is he somehow stops both teams from scoring, getting 2.5 Run Support for a 4-7 record. My team average 4.5 runs per game. He just lost his last game 1-0. Unreal.

Yeah I have high hopes for him. Now that he's in the rotation, I'm playing much more defensive hitters at up the middle positions. The one's I had picked up for their offense aren't elite enough to be doing anything anyway, so... Welcome to the lineup Devon White and Nellie Fox. Lets see how this goes.

[ASURay]

Originally Posted by chazzycat

Nevermind...ignore me. I could have sworn I saw this effect before, but I just tried again and was not able to recreate it. Sorry for the confusion

I don't think anything impacts MOV/CON. The pitches just impact STUFF. Either way, the in-game action is just impacted by STUFF/MOV/CON, as far as I know. I don't think it matters which actual pitches your pitcher throws in-game. For simmed games, I don't think pitchers even throw actual pitches, which is one of the reasons OOTP doesn't track data at the level of the individual pitch.

[DonkeyKongSr]

So, update on the Chris Sale to Closer, Treinen to reduced role experiment in my Perfect League.

In 42 games as a reliever, Sale is 4-3 with 19 of 20 saves converted with a 3.08 ERA & 2.75 FIP. Great stuff, with all 3 losses and the blown save coming in games where he relieved a LH Starter. 2 of the losses were against our league's super 0.800 record team, so I don't blame him, but I'm going to have to beat them in the LCS if I want to win the title.

On the season, RH are hitting 0.232 and LH are hitting 0.234, with 6 HR given up all to RH. His OBP vs. R is almost 50 points higher because he walks them a ton.

Meanwhile, by my estimates since Treinen lost the closer job, he has had about 30 appearances with a sub-2 ERA and a low 2 FIP. On the season, RH hitting 0.172, LH hitting 0.207 with 0 HR given up. Amazing stuff.

So, Chris Sale to the pen was a success, but Treinen went nuts. I don't want to shake anything up in case these are the better roles for each guy. I do wish there was a way for me to prevent Sale from playing on days I have a LH starter (have Randy and Keichel). I think in the playoffs, I going to have to manually change his role each game to keep him out of those games.

[DonkeyKongSr]

And speaking of the LH Starter problems in Perfect, I did some analysis of the hitter make-up in my Perfect League and my 2 Gold Leagues. What I found seemed to support all theories of better platooning at the higher leagues...

LH SP face in each league (Gold 1 - Gold 2 - Perfect):
LHB: 30% - 30% - 16%
RHB: 70% - 70% - 84%

RH SP face in each league (Gold 1 - Gold 2 - Perfect):
LHB: 42% - 45% - 32%
RHB: 58% - 55% - 68%

I analyzed all starters with at least 15 starts. Some individual samples show these numbers are within +/-2 percentage points for every pitcher I examined.

So, LH Perfect pitchers are facing about half the LHB they faced in Gold and half what RHP face in Perfect. Everyone's ERA is blowing up with the better competition, but LHP are getting hit much harder with so many fewer favorable matchups. The super team in my league runs a 100% RH lineup vs. LHP and has 3 LH and a SH for RHP.

[zrog2000]

Originally Posted by Orcin

I believe there is a general link between ratings and performance, but there is also an extreme amount of variability that makes the performance over small sample sizes (a season) unpredictable. Therefore, the best approach is to collect a stable of starters that can be rotated in and out during first few months until you settle on the best for that particular season. Eight or ten should be enough if they are all good.

I have a difficult time grasping the concept of hot and cold in OOTP, especially considering they are cards, not actually people. Doesn't it make more sense that a card with certain ratings is going to perform closer to his ratings with a large enough sample size? That's why I usually ride out cold streaks when I know that card should be performing better at that level because of the dozens of other identical cards that I've seen doing it. Because of this, I don't rotate much at all. I'm not going to take Mike Trout out of the lineup when he's not even OPS'ing .700 by June because it just doesn't make sense that he would continue hitting like that.

Does a poorly performing card keep performing poorly because of past results? It doesn't make sense to me that it does.

[SF Giants]

Originally Posted by ASURay

Here's the best model I have right now:

FIP_p = 8.661141 - (.01747*STUFF) - (.03291*MOV) - (.01737*CON)

This thread motivated me to update some of my numbers today. This model uses data from all pitchers in my league with at least 500 career IP across any league level. It actually correlates better to Perfect League FIP from last week than the model that uses only PL data. I'm guessing that will change as the amount of PL-only data increases.

I still cannot wrap my head around this. In classic OOTP there is a "calculator" of FIP in commisoner mode which trades stuff/control for movement in positive way but in this formula it is vice versa. I also ran few simulations and for classic OOTP I got better results when I opted for stuff and control. Is PT world waaaaay different from classic?

[Orcin]

Originally Posted by SF Giants

Is PT world waaaaay different from classic?

It's the same engine, but the distribution of batters is much different. It's more like playing with the "all-time greats" quickstart using the DH. There's a lot more power hitters than an ordinary MLB quickstart or even a historical year in the steroid era.

[Blazinghalo]

Deleted

[zrog2000]

Originally Posted by ASURay

Here's the best model I have right now:

FIP_p = 8.661141 - (.01747*STUFF) - (.03291*MOV) - (.01737*CON)

This thread motivated me to update some of my numbers today. This model uses data from all pitchers in my league with at least 500 career IP across any league level. It actually correlates better to Perfect League FIP from last week than the model that uses only PL data. I'm guessing that will change as the amount of PL-only data increases.

Is this still the same formula that you're using?

One thing that I noticed is that a few of the future legend cards have absolutely massive splits. (Medina and Whitley anyway) They're both so much better vs. RHB. That should affect their xFIP since they would be facing more RHB in general. I typically like to use them as RH Specialists so they're even better. I kind of think these cards have errors because the splits are so large.