Tuesday, September 20, 2016

UPDATE: Baseball Mogul Player Ratings

Back in April, I announced changes to the Baseball Mogul rating system for Baseball Mogul 2016.

To summarize:

1) Player ratings had crept up over the years (to the point that the average major league pitcher had an Overall rating of about 83).

2) Furthermore, this rating creep varied by rating. For example, a pitcher in Baseball Mogul Diamond with an 83 Control was in the top 40% of major leaguers. But a pitcher with an 84 Power was actually below average.

3) Because of these problems, Baseball Mogul 2016 featured a new system that defined 75 as the average across all ratings, and standardized the way in which ratings were distributed (the degree to which they clumped around the average value).

(This change had other benefits, such as more accurate historical simulations, and more realistic results in league that combined players from different historical eras.)

4) However, this meant that virtually all player ratings went down by 5-7 points in Baseball Mogul 2016, whether you were starting a game from scratch or resuming a game from a previous version. This came as a shock to many of you, eliciting some complaints that the player ratings were "broken".

Because of item #4, the most recent official patch (Version 19.17) included changes to the Player Rating Scale options (available in Options, on the Tools Menu). These are the four options that are now available:

Rating Scale
MLB Average
Notes
50-100
80
Current default for Baseball Mogul 2016 (version 19.16 and later). The default rating scale for Baseball Mogul Diamond and Baseball Mogul 2015.
25-100
75
Default rating scale for Baseball Mogul 2016 (before Version 19.16).
20-80
50
Scale used by professional scouts (described in further detail on FanGraphs).
1-20
13
20-point scale provided for fans of Football Manager.

This means that if you started a game in Baseball Mogul 2016 before version 19.16, you will see a big jump (about 5 points across the board) when you resume this game with the current patch. If you would like to go back to the system with an average rating of 75, you can change this setting in the Options Dialog:


I'm sorry for the above confusion. I personally prefer the rating system centered around 75 (a "C" on the traditional academic scale). But I didn't foresee how annoying this would be to people who had become accustomed to having a team full of players ratings in the 80s and 90s.







Wednesday, August 17, 2016

The Only Rule Is It Has To Work


I just finished reading my favorite book of the year, The Only Rule Is It Has to Work: Our Wild Experiment Building a New Kind of Baseball Team, by Ben Lindbergh and Sam Miller. As it is about baseball -- and it helped refine the way that I think about the game and how to model it -- I figured I would share my thoughts here.

This book may not be for everyone. But as someone who writes baseball simulations for a living, I have to say that this book is perfect.

The authors start with the sequence of events that landed them with the Sonoma Stompers for the Summer of 2015. One key reason is something that hadn't occurred to me: the general manager of a low-budget team spends most of his time selling tickets and keeping the concessions flowing, so he is happy to get free help building the roster (the task we associate most often with a team's front office).

Then Ben and Sam dig into the nitty-gritty of building a team. They do a great job of laying out all the numbers that they had in front of them for such tasks as: choosing which players to sign; making lineup recommendations; employing extreme defensive shifts; and building detailed reports on opposing pitchers for use by the team's hitters. Seeing the raw data made the book much more enjoyable than if they had just jumped ahead to the conclusions that they reached.

The authors also do a great job of conveying the storylines and emotions associated with the team. It's reminiscient of a movie like Bull Durham: the overarching plot is about baseball players trying to get a crack at the majors, but the most interesting and amusing and important events revolve around the players' individual growth and interpersonal relationships.

Finally, I found this book inspiring as a personal story of humility and frustration, combined with some great insight into how to "make friends and influence people". I would honestly recommend this book to aspiring business leaders or consultants. Ben and Sam are two extremely bright guys with great communication skills. So one could assume that they had an easy time showing up at the Sonoma Stompers and turning the team around. But I know that my life is never that easy -- and I have to admit it's nice to see that their lives aren't either. Although Ben and Sam are nominally in charge of the roster, it's difficult for two guys that never played professional baseball to earn credibility in the clubhouse. But if you keep reading you see that more often than not they are able to succeed at what they try to do -- relying on respect and candor and true generosity, more than on any spreadsheet.

Wednesday, June 8, 2016

2016 Mock MLB Draft

Jason Groome
A.J. Puk
As many of you already know, this year’s version of Baseball Mogul includes most of the 100 top prospects that are eligible for this year’s amateur draft. As the 2016 database already includes the top 75-80 players for each franchise, we also have a decent understanding of each team’s specific organizational needs.

So I booted up Baseball Mogul 2016 to see how it would handle this year's draftOne advantage of having a full simulation at our disposal is that I was able to simulate the draft 10,000 times and record all the results. The following table shows the most likely result of the first round (although this exact order only occurred in 131 of the 10,000 simulations). The average draft position for each player is shown below as “ADP”.

Pick
Team
Player
Pos
Age
Bats
Throws
Hgt
Wgt
Scouting
ADP
1
Phillies
Jason Groome
LHP
17
L
L
6'6"
220
40/75
1.61
2
Reds
A.J. Puk
LHP
21
L
L
6'7"
230
45/70
1.78
3
Braves
Nick Senzel
3B
20
R
R
6'1"
205
40/70
3.31
4
Rockies
Mickey Moniak
LF
18
L
R
6'2"
190
40/70
4.36
5
Brewers
Corey Ray
CF
21
L
L
5'11"
185
40/75
5.46
6
Athletics
Delvin Perez
SS
17
R
R
6'3"
165
45/75
6.26
7
Marlins
Riley Pint
RHP
18
R
R
6'4"
195
40/75
8.31
8
Padres
Kyle Lewis
RF
20
R
R
6'4"
195
35/70
10.86
9
Tigers
Matt Manning
RHP
18
R
R
6'6"
195
35/70
10.41
10
White Sox
Dakota Hudson
RHP
21
R
R
6'5"
205
40/65
11.46
11
Mariners
Cal Quantrill
RHP
20
L
R
6'3"
185
40/70
11.01
12
Red Sox
Braxton Garrett
LHP
18
R
L
6'3"
190
40/65
12.06
13
Rays
Blake Rutherford
LF
19
L
R
6'2"
190
45/70
10.86
14
Indians
Zack Collins
C
21
L
R
6'3"
220
40/65
14.16
15
Twins
Forrest Whitley
RHP
18
R
R
6'6"
250
40/70
15.21
16
Angels
Bryan Reynolds
CF
21
S
R
6'2"
210
35/70
16.26
17
Astros
Alex Kirilloff
CF
18
L
L
6'2"
195
40/65
16.06
18
Yankees
Kevin Gowdy
SP
18
R
R
6'4"
170
35/70
18.61
19
Mets
Will Craig
3B
18
R
R
6'3"
225
45/65
17.16
20
Dodgers
Jordan Sheffield
RHP
21
R
R
6'0"
185
40/70
18.71
21
Blue Jays
Buddy Reed
CF
21
S
R
6'4"
185
35/70
20.51
22
Pirates
Ian Anderson
RHP
18
R
R
6'3"
170
40/65
20.56
23
Cardinals
Cody Sedlock
RHP
20
R
R
6'4"
210
45/65
19.86
Scouting shows Baseball Mogul’s “Overall” and “Peak” ratings for each player (on the standard 20-80 scouting scale).

... and this table shows how frequently each player was taken with one of the top 3 picks (rounded to the nearest 0.1%):

Player
Pos
Age
Pick #1
Pick #2
Pick #3
Jason Groome
LHP
17
50.1%
39.1%
10.6%
A.J. Puk
LHP
21
39.2%
45.5%
14.4%
Nick Senzel
3B
20
6.8%
10.2%
42.1%
Mickey Moniak
LF
18
1.4%
2.6%
11.3%
Corey Ray
CF
21
1.0%
0.9%
9.0%
Delvin Perez
SS
17
0.7%
0.5%
6.2%
Riley Pint
RHP
18
0.5%
0.8%
3.5%
Kyle Lewis
RF
20
0.2%

0.7%
Matt Manning
RHP
18

0.3%
0.4%
Dakota Hudson
RHP
21


0.3%
Cal Quantrill
RHP
20
0.1%
0.1%
0.4%
Braxton Garrett
LHP
18


0.3%
Blake Rutherford
LF
19


0.2%

As you can see, Baseball Mogul was ambivalent about whether to take Puk or Groome with the first pick. Twelve years after Moneyball publicized the argument that college players had been perennially under-valued, front offices still frequently use top picks on high school players  so Mogul's drafting AI models that behavior.