Deadball Era Pitching and Beyond

[Syd Thrift]

I've played a *lot* of deadball era baseball with OOTP over the last few years for some reason. Actually, it all started as wanting to start a fictional-historical baseball league from way, way far back, then not liking the way the AI used players, and slowly getting more and more involved in things until I was Eugene Churching the games.

Anyway, point being, I've seen a lot of the game's issues with deadball era pitching in particular and it... well, it kinda blows. Most of the issues I see with deadball era pitchers stem from AI, not player creation. There are player creation issues as well (which I'll go into) but more than anything else it's not the players themselves, it's the way they're used. I'll just go point by point on this...

- First and foremost, pitchers - good pitchers especially - do not complete enough games and don't go far enough into games they don't complete, which in turn cuts down on the number of decisions they earn both on a per-game and a per-inning basis. I looked at Walter Johnson's career in my replay (see the Injury Log thread) as an indicative example. Although my Big Train started just 12 fewer games over the course of his career, he pitched 700 fewer innings - 2 Walter Johnson seasons worth of time lost.

- Adding to this, but admittedly a harder thing to solve, was the way that the real-life Johnson was used in relief over the course of his career. Like many aces of his day and up into the 1930s, he was used in the late innings of tie games started by lesser men in his team's rotation. Due to this, even on a per-9 innings basis, real Walter Johnson had 11% more decisions than the fake one. Overall, the 700 fewer innings, the 7-12 high-leverage outings per year, and the never being pulled except when he had completely blown a game combined to give the real Big Train 150(!) more decisions than my model.

- To exacerbate that issue WRT wins, there is *way* too much parity in this game, particularly in the early seasons. Yes, you do see individual 103-victory seasons, but there are also a *lot* of years where the best team in a league has fewer than 90 victories in a 154-game, 8 team setup. That occasionally happened IRL - I think 2 or 3 times in baseball history - but it happens perhaps every 3 or 4 years in OOTP.

This isn't just a realism issue, it's potentially a difficulty issue. Since you make the playoffs only by beating the top team in your league or division, whether they go 90-72 or 116-46, it's going to be a *lot* harder to put together a club that can get past a couple patsies and a couple dynasties than it is to emerge from a muddle of mediocre teams.

- Oh yeah... this applies to pitchers thusly: just as there's not a lot of difference in a lot of years between a great team and a bad one, there's often not a lot of difference between a great offense and an average one. I've seen seasons where the difference between the #1 offense and the worst was less than a run per game. Nice parity, I guess, but compare and contrast with the Giants in 1908 scoring 4.15 runs per game to the St. Louis Cardinals' 2.41 (the league average was 3.33). I'm not saying Christy Mathewson was not also a great pitcher, but clearly his 37-11 record was influenced at least a little bit helped out by a team that had an OPS+ of 117 (the batting equivalent of hitting in front of 9 Ichiro!s).

I also saw some big issues with BABIP (real-life Johnson's was a lot lower than the fake version for some reason... did big leaguers have a bigger effect on this than they do now?), but those are my main ones.

One other thing...

- I don't like the way the game treats old pitchers. I think it's way too nice to the bad ones. A team that's scuffling really shouldn't be starting a 1.5 star former great just because he happens to be the 5th best starter they could find. In that situation, most clubs would be looking at youngsters. This is a big part of why the game has guys finish their careers with lots and lots of decisions (in the modern era, anyway) but records hovering around .500.

- So accepting

[Mike Donlin]

I think the parity problem--and offense parity--has to do with the simplified financial system. If you play with finances off, it's basically a crapshoot where the best players end up. But the deadball era didn't have finances off, as OOTP imagines it. Some teams had wealthier owners than others, were willing to invest more in scouting, were willing to buy the best young players, etc. Without disparities in wealth, it's hard to produce disparities in talent or runs.

[silentrob]

Good point on the parity. In OOTP, owners don't sell their star players to finance other ventures like Broadway musicals. Nor can it simulate players being under the influence of gamblers and throwing games because of it.

Also a good point, though, Syd, on how star pitchers occasionally finished games in relief, thus boosting their career decisions. I'm not sure if you can adjust the game to simulate that or not. Maybe you can maneuver it to happen on teams you manage, but I don't see how you could get AI managed teams to do such a thing.

[Dagrims]

Originally Posted by Syd Thrift

- Adding to this, but admittedly a harder thing to solve, was the way that the real-life Johnson was used in relief over the course of his career. Like many aces of his day and up into the 1930s, he was used in the late innings of tie games started by lesser men in his team's rotation. Due to this, even on a per-9 innings basis, real Walter Johnson had 11% more decisions than the fake one. Overall, the 700 fewer innings, the 7-12 high-leverage outings per year, and the never being pulled except when he had completely blown a game combined to give the real Big Train 150(!) more decisions than my model.

What does the AI do if you put Johnson in the set-up/middle relief role as well as #1 starter role? Ignore it?

[Syd Thrift]

Currently, I'm pretty sure that if you try to stick him in relief, it will pull him out of the starting rotation automatically and vice versa. Perhaps if you set him as a spot starter and set the percentages to 100% that would work...

[Syd Thrift]

OOTP9 does allow you to play with full finances but no free agency. I haven't really dived in to see how the AI handles teams who don't have enough money to pay their stars.

That being said, I simmed 50 years of a fictional league using historical modifiers to mine the data and compare it to real life. Here are the competitive balance data I came up with for a couple of seasons picked more or less at random (more or less because I didn't even bother to look at seasons right at the beginning of the league, where initial player creation hasn't given way to player development ). Here's what I came up with:

Standard deviation of won/loss records: .076 / .064
Coefficient of runs scored (standard deviation divided by the average): 6.6% / 8.6%
Coefficient of runs allowed: 10.3% / 9.4%

Real life
----------------

1906
-------
W/L: .128
RS: 13.8%
RA: 15.9%

1916
--------
W/L: .102
RS: 10.7%
RA: 14.6%

1926
--------
W/L: .082
RS: 10.9%
RA: 13.4%

1936
--------
W/L: .089
RS: 14.6%
RA: 16.3%

1946
--------
W/L: .098
RS: 12.6%
RA: 9.9%

1956
--------
W/L: .090
RS: 13.7%
RA: 13.5%

1966
--------
W/L: .064
RS: 11.1%
RA: 12.0%

1976
-------
W/L: .073
RS: 12.3%
RA: 8.9%

1986
-------
W/L: .062
RS: 9.9%
RA: 8.9%

And this, folks, is why I find myself playing out each season when I do "serious" replays. Real-life competitive balance in terms of won-lost records doesn't achieve OOTP levels until the 1960s, and even then there's not the same sort of spread of teams achieving their results by, say, loading up on hitting at the expense of pitching. Hitting never really reaches the level of parity you get in OOTP. IME this is borne out by the fact that it seems to be too easy to pick up AAAA replacements of starters who get hurt. Bear in mind that I'm using a souped-up injury.txt file in conjunction with the "High" setting, so although there are a lot more injuries per year compared to what most players have, the game actually has fewer season-ending ones, which are mostly random and therefore also add entropy to a game as they increase.

Also, this sim had free agency turned off, financials turned on, and allowed teams to trade draft picks - all things that would either reduce entropy in a league or allow a good team to stay good through trade more easily (until their stars get old). Also, I noticed that even with the heightened injuries the best starting pitchers in my game pitched WAY more than their real-life counterparts, which also reduces entropy (although true to OOTPness, individual starters are just nowhere near as good on a per-game basis as real-life studs from that period). Several guys started more than 900 games (Cy Young is the real-life record holder with 811 GS). Despite all that, pitching coefficients didn't get down to my simmed numbers until the 1980s.

This is one of these "advanced topics" where there's no easy answer as to how to improve the game for the future. But this is definitely a major area of necessity IMO.

[silentrob]

Originally Posted by Syd Thrift

Bear in mind that I'm using a souped-up injury.txt file in conjunction with the "High" setting, so although there are a lot more injuries per year compared to what most players have, the game actually has fewer season-ending ones, which are mostly random and therefore also add entropy to a game as they increase.

Just wondering, have you ever posted this injury file? I've been reading the injury file thread with interest, but don't recall if you've made your latest improvements available. If I understand correctly what you are saying, it is pretty close to what I'd like to see in my leagues. Coming from a veteran such as yourself, it is probably superior to the default, and thus an upgrade over what I have right now.

[Syd Thrift]

I'll post it (perhaps along with my own Skydoggical settings) sooner or later. It's basically injury_log's file with one category of SEIs removed. Even with RonCo's development adjusts, you still get guys with Cy Young start numbers in the modern era (which I guess isn't toooo bad because Nolan Ryan did have 772 starts in his career).