Revision as of 00:10, 6 March 2007

This page is for use in keeping notes about my PSBL simulation baseball league team. Rather than setting up another wiki, I figured this was a better way to do it.

Roster

The following players are the ones that matter, Im skipping the scrubs.

C John Buck
C Kenji Johjima
1B Albert Pujols
2B Tadahito Iguchi
3B Corey Koskie
SS Orlando Cabrera
LF Moises Alou
RF JD Drew
CF Kenny Lofton
CF David DeJesus
DH Travis Hafner
MI Ramon Martinez
3B Eric Hinske
OF Joe Borchard
UTL Eric Young
P John Lackey
P Greg Maddux
P Jake Westbrook
P Josh Beckett
P Jeff Weaver
P Scott Downs
P Jason Frasor
P Akinori Otsuka
P Brian Fuentes
P Joe Nathan

Reserve Roster
OF Sammy Sosa
C Humberto Quintero
P Chris George
OF Tim Raines Jr
C Brayan Pena

I have cleaned up this page and moved all the old information to PSBL Archive.

Pre Season Deals

We need to get a backup IF who can hit, an OF who can hit lefties and has a lot of plate apps, and to replace Jeff Weaver.

Claudio Vargas

I have no idea what to offer yet.

Curt Schilling

I offered Jeff Weaver for Schilling straight up. I would be willing to include a prospect as well, but I dont think its worth that much. I think Weaver can have a big year in Safeco.

Kenny Rogers

I am thinking Johjima and Weaver for Rogers and Posada. This would greatly improve their youth and help us immensely this year. That would catapult us up near the top I think.

Woody Williams

I dont know if he is worth Weaver by himself.

John Smoltz

He has overvalued Smoltz in the past, but who knows?

Kevin Mench

He fits the bill in the OF.

Juan Rivera

Why the hell is his name constantly on this list? I think he would be great filling in all over the place in the OF and even DHing now and again. With his injury, his value should be down. I have my fingers crossed.

Marco Scutaro/Pablo Ozuna

The Cubs have both of these guys, and getting one would help a lot. He is unpredictable to trade with, however.

Team Building Philosophy

Since we had that mini-discussion about team stuff, I figured it might be good to get my thoughts written. This way we can find where we differ, and then bridge that gap and hopefully improve our ability to field a great team and win games.

Offense

I prefer to have a lot of speed at the top of the lineup, but OBP is more important. I prefer no more than 3-4 right handed hitters if possible. The #4 hitter should be a RH with a lot of power and good OBP, Pujols is the ideal #4 in my opinion. The #3 hitter should have great OBP, good average and good power and be a switch hitter or LH. The #5 hitter should be a LH batter with great power and OBP. I prefer not to stack more than 2 RH batters in a row, there are so many relievers that get them out, make the opponent either have a guy who is good vs both sides or use 2 relievers.

Defense

I prefer range over sure-handedness. I dont mind some errors if it means the guy gets to a lot of balls. In the OF I really like guys who can throw. With as much value as I place on taking an extra base for free, having an OF who can prevent it is a nice bonus, and often isnt too expensive. I really prefer a C who can throw, if he cant the pitching staff needs to be tailored.

Pitching

I prefer to have one left handed starter and one left handed reliever. Its crucial that both be good vs both sides, because they are so easy to min-max. The #1 starter must be right handed and dominant both ways. The closer must not blow games, those are wins thrown away. A bad start can be overcome, but a blown save is very difficult (less innings to work with) to beat.

Bench

I prefer to only carry 2 C so there are enough spaces to platoon a couple of guys and have some bats when needed. No more than one of the reserve OF should be veterans, and all of the reserve OF should have big bats vs at least one side. Its great to have a real fast pinch runner too.

Overall Philosophy

I prefer a well rounded approach to a team with one big strength. It seems to be much more difficult to stop a team that does a lot of things, than one that concentrates on one.

Mathematical Analysis of Lineup Choices

This should not be construed as a justification of why my method is better than yours. I dont believe it could ever be proved either way. What I have done his is offer a mathematical analysis of the foundation of my reasoning.

Assumptions

The analysis ceases to be simple of some assumptions concerning remote possibilities are included, therefore I have made this section to elucidate the foundation assumptions of this analysis.

No one hits into a double play
No one is thrown out stealing
No errors occur
Big innings are ignored (5+ batters come to the plate)
Everyone reaches base at a constant rate based solely on their OB% vs right handed pitchers

Definitions

I am defining a run scoring opportunity as one where Pujols comes to bat with someone on base. This will also include the times when one of the players preceding Albert has hit a HR or been driven in by one of the other batters. Rather than completing the analysis that is required to remove those, I will have them included because those were also run scoring opportunities, and they were converted to runs.

Run producers are defined by the magnitude of their OPS. In actuality I have been working on a better formula that takes into account the league averages, and weights OB differently than SLG compared to their respective norms. I also would need to weight AVG differently than the BB portion of OB, since nobody scores from 2nd on a BB. I havent finished that yet, so I will use simple OPS.

Hypothesis

It is most effective to put the three best OB players in front of the two greatest run producers to increase run production.

Analysis

In this model the lineup is structured as in the MP I made. In any inning where Lofton leads off, Pujols has a 76.7% chance of coming to bat, and this would, by definition be a run scoring opportunity. In any inning Iguchi leads off, Pujols has a 62.5% chance of coming to bat with a run scoring opportunity. When Drew leads off an inning, the chance is 41.0%. If we assume all of these instances are equally likely (the chance of any batter making the final out is equally distributed), the chance of Albert having a run scoring opportunity is 60.1%. This means that if he receives 700 plate apps batting 4th, he will have 421 run scoring opportunities.

Weaknesses

The highest probability of Albert getting a run scoring opportunity is when Lofton leads off. This is given equal weight in the analysis because we dont know how many right handed pitchers we will face, even though it is the most likely occurrance, because every one of these games begins with it, then the rest are of equal probability.

Hypothesis

The best hitter should bat first to increase his plate appearances.

Additional Assumptions

The bottom portion of the lineup is equivalent to the MP I generated.

Analysis

Albert bats first, so never has a run scoring opportunity during the 1st inning. His chance of having a run scoring opportunity when Koskie leads off an inning is 71.9%. The chance when Alou leads off is 56.2% and when Cabrera leads off it is 34.2%. To give them equal weight his run scoring opportunity chance is 54.1%.

If we assume he will get 750 plate appearances over the course of the year, this yields 406 run scoring opportunities. This is actually flawed because we know that 160 of the plate apps will not take place with a run scoring opportunity, because they lead off the game.

Discussion

Is this analysis accurate? I dont know. All I know is this is the logic I have always used, though I never understood it as fully as I do now. Before you mentioned it, I just did what I saw MLB managers do and sprinkled in my own ideas. I remember seeing your lineup with Bonds leading off and wondering why you had done it. It never occurred to me that increasing his plate apps might generate more run production.

Data

I dont know if I mentioned that I am insane, if I didnt, consider yourself warned. I continued to research the best possible lineup and eventually found this [1]. It allows you to input a lineup and switch the players around and see how many runs this lineup will score. If I am understanding it correctly it places the lineup against MLB teams, probably National League. So it wont predict PSBL runs, but it can be used to extrapolate the number of runs one lineup will score relative to another.

I entered 3 different lineups, the standard one I made previously, the OPS model and the OB model similar to those I emailed you. I then ran them through the simulator in 10 batches of 1000 trials each, and came up with the following numbers:

Standard: 6.898
OPS model: 6.882
OB model: 6.992

So the OB model has fared the best. Of course, this cannot take into account the handedness of the batters. Another thing I did is compare the standard lineup with the most common substitutions, which brought these numbers:

Hinske instead of Koskie: 6.990
Borchard instead of Alou: 6.822
DeJesus instead of Alou: 6.777

This says to me that We should not use DeJesus in LF unless both Alou and Borchard are used, or in blowouts. It also brings Hinske up as the #1 3B over Koskie. Unless of course, the difference can be made up by defense. My intuition is it could, but when I did the numbers if both were used equally for half a season, Hinske would drive in almost 7 more weighted runs. I doubt Koskie will save 7 runs with his glove in that time period, though it is possible. I might do some analysis to this effect later.

Defensive Conjectures

I dont know how accurately DMB simulates actual defense, but if I assume it will perfectly simulate the differences between Hinske and Koskie, the following will occur (in half a season):

Koskie commits 12 errors, Hinske will commit none
Koskie will get 24 more putouts
Koskie will get 56 more assists

That is basically giving the opponent almost one extra baserunner every game, about 6 per series. Maybe it is worth those 7 runs.

Analysis

These numbers seem to indicate that the OB model could be superior to the model I normally use, to the tune of a little over 15 runs per season. It does not take into account things such as speed (how useful are Loftons steals in the middle of the lineup?), or ease of the opponent to manage against our lineup (left handed batters hit significantly lower vs left handers than the analog for right handers). Can these two factors combine to produce more than 15 runs over the course of a season?

I say yes. Mostly because of the lack of good pinch hitters on our team and the significant decrease in OPS versus lefties I believe that the relief matchups alone will accomplish this. Only once per 10 games would a left hander need to come in for multiple batters and cost a single run for the model to underperform enough to negate its advantage.

I had a great time doing this analysis. If you have any more for me to do, by all means, suggest it.