[IBL] Fun With Numbers -- What Is Wrong With Your Team's Offense

[Chris Hartjes]

Hello friends. Today I wanted to share some stuff I had been looking at in a continued quest to torture myself about the lack of success of the Madness. In this edition, I am going to talk about First Order wins, Second Order wins, and Base Runs.

The purpose of this is to look at how teams are performing (usually mine) and to see if there is Something I Am Doing Poorly And Correct it.

First, here is a chart showing the actual win-loss records, First Order records, and Second Order records.

BAL 26-50 | 23-53 | 26-50
BFD 36-42 | 35-43 | 37-41
BOW 40-35 | 43-32 | 42-33
BUF 33-44 | 32-45 | 29-48
CAN 43-31 | 43-31 | 43-31
EVE 34-43 | 40-37 | 42-35
GTY 40-37 | 41-36 | 43-34
KAT 38-44 | 31-51 | 34-48
MAD 34-48 | 33-49 | 35-47
MCM 52-30 | 50-32 | 46-36
NJR 40-38 | 38-40 | 42-36
NYK 36-42 | 37-41 | 34-44
OXY 36-40 | 38-38 | 37-39
PHI 35-43 | 41-37 | 40-38
RRG 35-39 | 31-43 | 30-44
SCS 38-40 | 37-41 | 37-41
SDQ 37-37 | 36-38 | 34-40
SEA 53-25 | 49-29 | 43-35
SFL 47-35 | 47-35 | 47-35
SFP 23-51 | 28-46 | 30-44
SKY 45-36 | 48-33 | 48-33
SLG 37-41 | 30-48 | 30-48
SNH 52-34 | 54-32 | 57-29
WMS 46-31 | 42-35 | 41-36

The actual win-loss records are as of this morning when I grabbed a database dump.

First Order wins are just a fancy way of saying "wins and losses based on runs scored and runs allowed" method. Rusty uses it on the Almanac site. I am using a similar version that makes a tiny adjustment to the exponent used based on the runs scored in a team's games. The "official" exponent is 1.87 while the "adjusted" ones range from 1.73 to 1.93. For those wondering why I did it, there is an explanation of this approach in Wikipedia articles about using the Pythagorean Projection method.

Second Order wins are a fancy was of saying "wins and losses based on projected runs scored and projected runs allowed". In this case, I am using Base Runs to project them. There is a great article on Wikipedia about Base Runs and the various formulas you can use for it. I picked the one that used the most data available to me.

Base Runs estimates runs scored using hits, extra base hits, and walks. When there are significant differences between actual runs and Base Runs, it can be usually chalked up to a few factors:

1) sequencing
2) quality of opposition
3) luck

In my case, my team's performance appears to be exactly where it should be based on actual batting and pitching performance. Which is disappointing if I was looking for something like "I would be .500 if not for really bad luck". This isn't to say that rollbot is not playing a HUMUNGOUS role in generating outcomes. We all have players on our teams producing at levels way below what is expected, and this is a major contributor to teams feeling like they are "underperforming". It is not surprising, for example, that MAD is struggling to score runs when 8 of my top 11 hitters in terms of playing time are not even generating league-average results. Add to this a starting staff with only 2 of 5 members producing at a better-than-average rate, and this is exactly why my team is 34-48, on it's way to another terrible season.

I will continue to produce this data as the season progresses and I encourage you to look at it as a tool to try and make sense of where you team is. It can be very hard to answer the question "is my team doing what it is supposed to do" and avoid making team-construction decisions based on emotional responses. We can't control rollbot but we can learn to better evaluate our teams when things go right and go wrong, and make better decisions going forward.

There is a thing called Third Order wins, where you take the Base Runs generated and adjust them based on the quality of the opposition but that math is beyond my current level of knowledge to figure out how to proceed. 

Anyway, thanks for reading all the way to the bottom and good luck in the second half of the year.

-- 
  Chris Hartjes
  chartjes at littlehart.net