I Hate Diamond Mind's Manager Profile System

Learning to Manage Carl Erskine

The Problem

The problem here is simple. Relievers in 1949 were used differently than relievers are used today.

That’s obvious from the statistics alone, right? Take somebody like Joe Page for example.

Page had 60 appearances in 1949, all in relief. He pitched 135 1/3 innings, or over 2 innings per appearance.

Now, if we were controlling the in-game roster moves for both teams, this would be an easy problem to solve. We’d start looking to bring Page in during the 7th inning or so, probably saving him until the Yankees starter started to falter. In my case, I’d be a tad more lenient with the starter, letting him get deep into trouble before going to the bullpen just in case.

Diamond Mind Baseball’s Managerial Tendencies

However, Diamond Mind Baseball’s computer manager simply isn’t built to do that.

As I understand it, the Diamond Mind Baseball computer manager was programmed to emulate the managerial styles of baseball in the era from the mid-1990s until sometime in the mid-2000s.

Now, this isn’t explicitly stated anywhere. The manual doesn’t tell you how the logic behind the computer manager works exactly. The statements we get (which I believe were composed 20 or more years ago) are all somewhat cryptic. Take this, under the Help entry for “Manager Tendencies:”

You’ll notice that Diamond Mind Baseball never tells you what years the play-by-play data encompassing “thousands of games” is from. It’s safe to assume, of course, that it is referring to the data available through Retrosheet — though I’m inclined to believe that this is the Retrosheet data that was available around 2006 or earlier, not the extensive data we have available today.

In other words, seasons like 1949 don’t really fit into the “neutral” picture.

So, then, what do we do to get the computer to act like it’s 1949?

We mess around with team-specific managerial and player tendencies.

Carl Erskine

This game inspired this post.

You might have forgotten what happened. The Brooklyn Dodgers, who have played poorly to date in my 1949 replay, were down by 3 in Philadelphia. They had two men on with two men out, and relief pitcher Carl Erskine up to bat.

Instead of pinch hitting for Erskine, which any sensible manager would do, the computer decided to let him hit for himself:

And this was the result:

I was confused as to why this would happen. I mean, I’ve seen bad computer manager decisions in the past, but this one was simply outrageous. I couldn’t come up with a single reason for it.

And then I decided to look into the managerial and player tendencies.

The Tendencies

First of all, it’s not easy to find these tendencies.

Diamond Mind Baseball is in desperate need of a user interface overhaul. You can’t chance the player tendencies by doing something as simple as looking at the player on the organizer screen, for example:

While you can view the player’s statistics here, and even modify those statistics and recalculate the all-important event tables, you can’t change how the player is managed.

That’s in a different screen, hidden under “Roster / Manager Profile:”

Click on the Dodgers, and you’ll get this:

This is the roster screen that you would use if you were managing rosters by hand. I’m not a fan of this screen at all, by the way. It’s cluttered, it’s complicated, and it’s very user unfriendly. But we’ll talk about that later.

What we are looking for is in those tabs at the top. First, let’s check out the Dodgers’ managerial tendencies, just in case something strange is happening there:

Now, this is where my issue with this system starts. Nothing is written in stone. Instead, we have a bunch of sliders that may or may not have a noticable effect in our replay.

Is it really all that important that “Pickoff throws” are set to “Most frequent?” Is there some sort of basis for this decision? We know that there are no records for attempted pickoff throws in 1949. I presume that this has something to do with the number of potential basestealers who attempted to steal against the Dodgers in 1949 — but wouldn’t it be nice to know what this is actually doing?

Anyway, there’s nothing strange about pinch hitting for pitchers here. Let’s take a look at the settings for Erskine himself:

And there you have it.

Erskine is set to never be removed for a pinch hitter. Ever. No matter what.

That’s the reason why he was left in even though the circumstance clearly called for him to be removed.

Also, by the way, Erskine will apparently never be removed for another reliever while he is in the game. That’s what that “Using relievers” and “Using closers” signifies, assuming I understand the corresponding section in “Player Tendencies” in Diamond Mind’s help page correctly:

Okay — maybe “never” here doesn’t actually mean “never” (nice semantics, eh?). It just means that Erskine is expected to finish out the game, unless he gets shelled.

By the way, those pitching tendencies in the “manager tendencies” section aren’t explained well, either:

Wait a second. If I understand this right, the decision to use relievers is complex — so complex, in fact, that the help tool doesn’t even try to explain it to the user? I mean, how are you supposed to know how to set your managerial tendencies correctly without simming through the entire season a few dozen times to see what happens?

Anyway, for the sake of completeness, here’s the relevant section on pinch hitting:

Again, it’s not really all that helpful, except to tell us that “never” means “never except in a blowout.”

Now, I want to note here that these are all default settings. I have not made any modifications to the 1949 season disk. This is how it came straight from the company.

Erskine In Real Life

The problem is that these settings don’t match Erskine’s real life performance.

Yes, Erskine did finish quite a few games. However, he was also taken out in games that were decidedly not blowouts.

In other words, Diamond Mind Baseball’s official 1949 Carl Erskine settings don’t reflect reality.

Let’s look at a few examples together. After all, we’ve got the play by play data, and we’ve also got the newspapers to make it fun.

Pinch Hitting for Erskine

Yes, the Dodgers did pinch hit for Erskine occasionally in 1949 — even if the game wasn’t a blowout.

For example, in the first game of the August 23 doubleheader at Ebbets against the Cardinals, Erskine was taken out for a pinch hitter in the bottom of the 7th inning, with the Dodgers down 5-2. Duke Snider had just hit a solo home run, and I’m guessing that Walter Alston thought a rally was imminent:

Sadly, it wasn’t enough in the end:

From Brooklyn Eagle, August 24, 1949, page 19

A few days later — August 26, to be precise — the Dodgers hosted the Cubs. Erskine relieved Joe Hatten in this one, pitching 4 1/3 innings of effective ball before being taken out for a pinch hitter:

Again, the score here was 4-1 — not a blowout by any reasonable definition.

Brooklyn lost this one as well:

From Brooklyn Eagle, August 27, 1949, page 6

It happened again. On September 3 in the Polo Grounds, Erskine was taken out for a pinch hitter in the top of the 7th, as the Dodgers had a runner on second base, down 4-2:

Erskine had pitched 4 innings in this game, giving up 0 runs on 2 hits and striking out 4. He had struck out the last two hitters he had faced, and probably could have stayed out there for more. And, of course, this game was clearly not a blowout.

From Sunday News (New York City), September 4, 1949, page 48

The Dodgers lost again (see the pattern?).

I could go on. Yes, that’s right: there are other examples of Erskine being lifted for pinch hitters, including ones in which he pitched only a single full inning beforehand. None of them came in blowouts.

Removing Erskine For A Reliever

Erskine was also taken out for a reliever from time to time.

Take the October 1st game at Philadelphia, for example. The Dodgers were in the thick of a huge pennant race, and were suddenly tied 3-3 in the bottom of the 6th inning. Ralph Branca had just given up a home run to Del Ennis, prompting Erskine’s entrance.

Erskine promptly gave up a home run to Andy Seminick to put the Phillies ahead. He gave up a base hit and a walk, and then was brought out for Jack Banta, who put out the fire:

Would this happen in Diamond Mind Baseball? Of course not. It’s not a blowout, and the computer has been told to keep Erskine in there no matter what.

From The Philadelphia Inquirer, October 2, 1949, page 49

Why?

I think the biggest question here is why in the world Diamond Mind has these settings for Carl Erskine in 1949.

Honestly, I think this is the result of extensive playtesting.

Erskine’s pitching line in 1949 is one of those difficult to reproduce lines: 22 appearances, 3 starts, 2 complete games, and 79 2/3 innings. Like Page, he averaged over 2 innings per relief appearance.

It’s hard to convince Diamond Mind’s computer manager to keep pitchers like this on the mound during close games. And that’s why we’ve got this extreme player tendency.

Now, I would prefer to see a game with a transparent computer manager. I’d prefer to see one that can read and adapt to the era that the game was played in, one that is not entirely dependent on play-by-play data from the 1990s and early 2000s in its decision making process.

Barring that, I would at least like to see a computer manager that can be prompted to make decisions when it makes sense, or that can be prevented from making poor decisions. None of that is possible with Diamond Mind Baseball, though it thankfully is with the other major computer baseball sims.

Honestly, I’d really prefer to manage both teams and be done with it. You can’t do that in Diamond Mind Baseball, however. That completely destroys the running game — and, yes, calling for a pitchout or throwing to first at the right time drastically impacts the stolen base rate.

Even without the pitch by pitch mode, there still is an argument to be made for the more simple card and dice games. Would you rather have a bunch of foul balls or the ability to manage in a manner appropriate to the era?