Ootp 26 fip

[jdw]

So... just noticed this:

Mike Torrez
4.07 Fangraphs
4.07 BR
4.08 OOTP19 (Real 1+2+1 import)
4.53 OOTP26 (Neutralized 1+2+1 import)
4.53 OOTP26 (Real 1+2+1 import)

Paul Splittorff
3.73 Fangraphs
3.73 BR
3.77 OOTP19 (Real 1+2+1 import)
4.26 OOTP26 (Neutralized 1+2+1 import)
4.26 OOTP26 (Real 1+2+1 import)

Steve Carlton
3.15 Fangraphs
3.15 BR
3.16 OOTP19 (Real 1+2+1 import)
3.58 OOTP26 (Neutralized 1+2+1 import)
3.58 OOTP26 (Real 1+2+1 import)

Any example of a seasonal change:

1972 Carlton
1.97 Fangraphs
2.01 BR
1.98 OOTP19
2.74 OOTP26

These are pretty radical changes.

Perhaps something to try to draw FIP more in line with Runs Allowed rather than ERA?

Has the League FIP been adjusted as well, to make FIP- work? It was always useful to look at whether a pitcher was sub-100 or above 100 (i.e. League Average & park adjusted), and the degree to which.

Is there perhaps a useful FAQ that has this stuff? I didn't see FIP specifically in the announcement on the historical changes in OOTP25.

[jdw]

This seems like a simple question that someone must know the answer too, or it was discussed before in a thread and I could be pointed to.

Perhaps a visual representation. This is a New 1970 Import using the Wizard with 1 year Recalc with Real Stats. Any adjustments in the Wiz were minor, like using the 1-100 Scale rather than the 20-80 scale (old habit that's impossible to break).

First image is Steve Carlton in the game. I scrolled down on my screen so that Career could be shown, which worked nicely as the famed 1972 is the first Season shown and easy to see.

Second is his Fangraphs page.

Fangraphs - Steve Carlton

Can someone jump in? Both FIP and FIP- are off. I actually find value in FIP- as a useful shorthand when looking at non-obvious players rather than the Carlton, Ryan, Blyleven, Seavers of the world. Hence the examples of Torrez & Splittorff in the first post.

2.01 Fangraphs in 1972 --> 2.74 in OOTP 1972 is a pretty massive variance for what is a pretty simple math equation.

[jdw]

So... while waiting for some clarity, MAtt Arnold posted in this thread here back in 2024:


quibbles and questions about OOTP WAR implementation

Originally Posted by Matt Arnold

Each league year, depending on the totals, calculates different constants. So internally we more or less create a table like the Fangraphs guts table (https://www.fangraphs.com/guts.aspx?type=cn) for calculating wOBA/WAR/FIP/etc..

For reference, Fangraph's Guts Tool Table is here:

Fangraphs Guts!

Their Library page on FIP is here:

Sambermetric Librabry: FIP

The FIP "Constant" for 1972 is 2.399.

This ties in when creating a spreadsheet to determine Carlton's 1972 Fangraphs FIP from his Stats & the Constant:

17 HR
87 BB
1 HBP
310 K
346.333IP

FIP = ((13*HR)+(3*(BB+HBP))-(2*K))/IP + Constant

Putting in 2.399 in the Constant, it rounds up to 2.01 which is what's reflected on Fangraphs' Carlton page.

If all of the Other Stats & Multipliers remain the same in OOTP's FIP calculation, the 1972 OOTP Constant would need to be roughly 3.134.

Let's see if that 3.134 accounts for the calculation "error" in Mike Torrez & Paul Paul Splittorff in 1972 OOTP:


4.36 Torrez - matching his OOTP FIP of 4.36 rather than his 3.62 Fangraphs FIP
3.49 Splittorff - very close to his OOTP FIP of 3.52 rather than his 3.62 Fangraphs FIP

It's not a perfect match, but it's extremely close.

The question of how OOTP comes up with a ~3.134 oConstant for 1972 if that's the case... no idea. Same for how OOTP calculates -FIP

Here's an example:

Torrez -FIP 1972-81:
106-117-101-107-099-106-098-099-098-106 Fangraphs
105-120-103-114-103-118-104-111-104-109 OOTP26

Focusing in on these:

Torrez -FIP 1974-80:
101-107-099-106-098-099-098 Fangraphs
103-114-103-118-104-111-104 OOTP26

Fangraphs has Torrez as a 101 -FIP pitcher on average from 1974-80, across 7 seasons.

OOTP has him as roughly a 108 -FIP.

7 relative to League doesn't seem to look like a lot when using it as a reference, but...

1974-80 -FIP --> -oFIP
87 --> 92 Carlton
93 --> 98 Sutton
101 --> 108 Torrez

6-8 -FIP is quite a bit. Especially when one doesn't know "why" to mentally account for the changes.

Anyway, it would be nice for someone who Knows what is causing this to check in.

[Matt Arnold]

If you start a historical game, we only load league totals for before the season in question. So for any season which is on or after the historical start, we are forced to basically use the "default" values for stuff like FIP constants.

So if you start a game on or before 1972, for Carlton, when we calculate his FIP for 1972, we don't have the historical league stats, so we just use the default for that, which we use a 3.2 for the FIP constant. You can see even if you start a game in 1972 but sim out a few months, at a certain point, we get enough data to recalculate the fip constant internally, and the real stats line FIP will use the new constant for its calculations.

[jdw]

Thanks!