Adjusting stats across eras? Tools? Formulas?

[OhioBobcats]

What have roster makers out there done to adjust stats across eras? Looking for any tools I could possibly mess with. Thanks.

[Culture Rot]

No point. The show uses a stat ceiling based around "modern" stats. Statistically Anomalies are not able to be recreated no matter how you set ratings. The game artificially increases and decreases stats to level things out during a season

[Knight165]

Quote: Originally Posted by xlatinoheatx

No point. The show uses a stat ceiling based around "modern" stats. Statistically Anomalies are not able to be recreated no matter how you set ratings. The game artificially increases and decreases stats to level things out during a season

Not true...at all.
I've had players hit .365-.375 easily in the game.
60+ homers
K #s can get quite high.

The hardest is probably wins.

M.K.
Knight165

[Culture Rot]

Quote: Originally Posted by Knight165

Not true...at all. I've had players hit .365-.375 easily in the game. 60+ homers K #s can get quite high. The hardest is probably wins. M.K. Knight165

All due respect I disagree. At all? Are you suggesting not only that I'm wrong but that there are no stat ceilings at all?

If your getting those numbers, it's gotta be through playing not pure simulation stats.

I tested this experiment many times. Take a 99 rated pitcher. Throw 3 shutouts with him user control. Sim 5 starts and take note of what happens. His era will sky rocket to "level out". This also happens with hitters. It's why maxed out hitters only max out at current era levels. It's why in your own rosters, Tony Gwynn can barely crack .330 in years he hit 360 plus. I'd love to be wrong. If this no longer exists, it must have changed on next gen. I tested this from mlb the show 10 to 15 on past gen. I strongly suspect nothing has changed but have to respect your two cents on this Knight. Can you prove or confirm otherwise? Truly?

[JTommy67]

Quote: Originally Posted by Knight165

Not true...at all. I've had players hit .365-.375 easily in the game. 60+ homers K #s can get quite high. The hardest is probably wins. M.K. Knight165

Definitely agree. I have a 1941 test roster ('15 version) in the vault which uses scaled ratings and Ted Williams can hit in the .390s in a good sim year.

To the original poster, the pitcher and hitter ratings work together. To produce higher batting averages, you needn't necessarily raise contact ratings; lowering H/9 ratings will suffice. The same holds true for walks, strikeouts, and home runs. It's better to manipulate pitcher ratings downward for more offense; you have more room to work with.

[JTommy67]

Quote: Originally Posted by xlatinoheatx

All due respect I disagree. At all? Are you suggesting not only that I'm wrong but that there are no stat ceilings at all? If your getting those numbers, it's gotta be through playing not pure simulation stats. I tested this experiment many times. Take a 99 rated pitcher. Throw 3 shutouts with him user control. Sim 5 starts and take note of what happens. His era will sky rocket to "level out". This also happens with hitters. It's why maxed out hitters only max out at current era levels. It's why in your own rosters, Tony Gwynn can barely crack .330 in years he hit 360 plus. I'd love to be wrong. If this no longer exists, it must have changed on next gen. I tested this from mlb the show 10 to 15 on past gen. I strongly suspect nothing has changed but have to respect your two cents on this Knight. Can you prove or confirm otherwise? Truly?

For the 2013 version of the game I created a test roster where every attribute for every player (pitchers and position players) was set to zero. I then added a player to every roster who was 99 for every rating. These guys were batting .500 with 80-100 homers typically.

I did this to see if ratings could be scaled to reproduce more historically accurate sim engine results.

So, yes, the range of possibilities in the sim engine is greater than it might appear.

[Culture Rot]

Quote: Originally Posted by JTommy67

For the 2013 version of the game I created a test roster where every attribute for every player (pitchers and position players) was set to zero. I then added a player to every roster who was 99 for every rating. These guys were batting .500 with 80-100 homers typically. I did this to see if ratings could be scaled to reproduce more historically accurate sim engine results. So, yes, the range of possibilities in the sim engine is greater than it might appear.

Thanks for your findings on this. I was so disappointed by this, I swore off the series for a long while outside of online play. It doesn't explain my findings though. Obviously one cant play with a 0/99 roster just to get a few statistically noteworthy seasons out of hall of fame players. I wonder what the sweet spot here is.

If there is an actual retro roster that has been able to solve this "historical stats" issue, I've yet to play it :/

[JTommy67]

No problem. There's a long history to my conclusions.

After I confirmed that it was possible to scale ratings I created a calculator and used it to rate players for all years in MLB history, and included these ratings in my roster utility (see link below).

I used what statistics I believe the developers are using to calculate ratings and constructed my scales and formulas to match theirs as closely as I could.

I then applied some mathematical scaling to historical statistics to ensure a better balance between pitchers and hitters which would yield more historically accurate statistics. Without getting into too much detail, hitting ratings are rescaled (based on league averages for the year in question) to maintain a standard distribution while pitching ratings are allowed to freely fluctuate based on the current scale observed in the game.

Some members have used my utility and reported positive results. And, as I mentioned previously, check out my 1941 test roster in the 2015 roster vault. There are some things in there which have since been fixed, but you'll notice DiMaggio and Williams hitting in the .350 - .390 range if you sim many seasons and league-wide averages more or less on the mark for what happened that year.

The problem is that the roster-making process is a slow one and while there are some rosters out there made by people who've used my utility, I still don't think there is a full-on test where my numbers have been used exclusively, which makes it difficult to evaluate results.

Also, keep in mind that for historical years you have empty team slots (due to expansion) which must be filled with players. How this is done can significantly alter what you see in the sim engine.

Anyway, my conclusion is that accurate scaling is certainly possible, with the exception being stolen base numbers.

Here is a link to my utility:

https://www.mediafire.com/folder/pp9...shf/RB_Utility