Machine Choice Postseason Crystall Ball Scoreboard

[Machine Choice]

If any of you have ever done spreadsheet business modeling, you've probably dealt with a software program called "Crystal Ball." I have been using that program to model baseball games and predict scores.

If anyone has any technical questions about how I modeled the parameters, feel free to ask. Otherwise, I will not belabor this thread with all the data inputs.

Each day I will provide the predicted scores and then post a "best bet" which is what I am actually wagering on.

[BB_Mariners]

Would you tell us what are the main parameters and variables and how values are assigned to each variable?

[Machine Choice]

Monday 10/3/11

Tigers 7
Yankees 3

Rangers 6
Rays 5

Best bet: Tigers -132

[Machine Choice]

The best bet will be the "graded" play, but I'll also track the record for the whole set-up just for information purposes.

[Machine Choice]

Paramaters broken into pitching and hitting

Pitching (parameter/relative weight for category)
-pitcher's last four starts either home/away (depending on which situtauion he's pitching in for the game)(.286)
-pitcher's previous ERA vs opponent in a home/away situation (depending on the current game) in the same calendar year(.228)
-Pitcher's current calendar year ERA in a day/night setting (use whichever one the current game will be in)(.186)
-Bullpen ERA as an average of two parameters: 1. total bullpen ERA at home (or away, depending on team's game situation) and 2. Bullpen ERA in the last 10 games (.3)
Cumulative weight: 1.0

Hitting (parameter/relative weight)
-Average of team's last 7 games home/away runs scored (depending on team's situation for the current game)(.6)
-Avg runs scored in support of starting pitcher for the pitcher's last four starts in a home/away situation (depends on the starter's situation in the current game)(.4)
cumulative weight: 1.0

I do the above parameters for each team, but the input variables are modeled on a normal distribution. The mean and standard deviations are pulled from Excel formulas, and then I run 10,000 trials in Crystal Ball.

The next step is to set up ratios that account for variations in team performance. For instance, a pitcher who allows 1 run in six innings vs the Seattle Mariners on the road, probably will not be able to duplicate that performance on the road at Rangers ballpark. Therefore, once I get the average pitching performance parameter (described above) I set up a ratio that factors in how the opposing team's pitcher could be expected to do against a team that produces runs in accordance with the hitting parameter described above, factored against the "league average." I then do the same thing for the hitting "aggregate". Once I get these modified aggregates, I get to....

The game score: average the Team A's aggregated pitching parameter with Team B's aggregated hitting paramater. Then do the same for Team B. This produces the predicted game score.

[mattmc419]

Have you used this for any previous games?

You don't know how much I'm hoping you're right about the Tigers game

[Machine Choice]

Have you used this for any previous games?

No, not in this exact set-up. I used a somewhat similar set-up since late August and it had about a 58-60% hit rate on totals, I can't rememebr what the ML rate was. I posted those results in this forum but I can't find it. Anyway, this model is different in a few ways so it may do better or worse. We'll see.

[firehoyt]

This Prediction program reminds me of the movie "Along Came Polly"

Paramaters broken into pitching and hitting

Pitching (parameter/relative weight for category)
-pitcher's last four starts either home/away (depending on which situtauion he's pitching in for the game)(.286)
-pitcher's previous ERA vs opponent in a home/away situation (depending on the current game) in the same calendar year(.228)
-Pitcher's current calendar year ERA in a day/night setting (use whichever one the current game will be in)(.186)
-Bullpen ERA as an average of two parameters: 1. total bullpen ERA at home (or away, depending on team's game situation) and 2. Bullpen ERA in the last 10 games (.3)
Cumulative weight: 1.0

Hitting (parameter/relative weight)
-Average of team's last 7 games home/away runs scored (depending on team's situation for the current game)(.6)
-Avg runs scored in support of starting pitcher for the pitcher's last four starts in a home/away situation (depends on the starter's situation in the current game)(.4)
cumulative weight: 1.0

I do the above parameters for each team, but the input variables are modeled on a normal distribution. The mean and standard deviations are pulled from Excel formulas, and then I run 10,000 trials in Crystal Ball.

The next step is to set up ratios that account for variations in team performance. For instance, a pitcher who allows 1 run in six innings vs the Seattle Mariners on the road, probably will not be able to duplicate that performance on the road at Rangers ballpark. Therefore, once I get the average pitching performance parameter (described above) I set up a ratio that factors in how the opposing team's pitcher could be expected to do against a team that produces runs in accordance with the hitting parameter described above, factored against the "league average." I then do the same thing for the hitting "aggregate". Once I get these modified aggregates, I get to....

The game score: average the Team A's aggregated pitching parameter with Team B's aggregated hitting paramater. Then do the same for Team B. This produces the predicted game score.

[Machine Choice]

Graded Plays: 1-0

System Stats:
ML: 2-0
Totals: 1-1
Overall: 3-1

[Machine Choice]

Tuesday 10/4/11

Rangers 4
Rays 3

Yankees 4
Tigers 13

Brewers pk
Diamonds pk
Total runs: 6

Cards/Phillies: no plays (Holliday out)

Best Bet: Tigers -115

[I'm_Lucky]

Hi Machine Choice,

Good to see that your still here, and best of luck on your picks .. you don't know me, but I still remember you from two years ago when I lurked around the forum and tailed you on your NCAA basketball picks; you had a ridiculously good streak going! Anyways, this was totally random and maybe I should have pm'd you instead of derailing the thread, lol .. Good Luck!!

[Aztec Empire2]

Good Luck Machine!!

[btd]

Thanx for this info, didn't know Holliday was out .

[figue]

[Machine Choice]

Tuesday 10/4/11

Rangers 4
Rays 3

Yankees 4
Tigers 13

Brewers pk
Diamonds pk
Total runs: 6

Cards/Phillies: no plays (Holliday out)

Best Bet: Tigers -115

LOL, I nailed the Rangers game exactly

[Goodluck]

gj machine!

[Goodluck]

machine, any thought about the NYY line going up? I like Detroit here too, but whats with the line movement? People don't think NYY will be done in game 4?

[Machine Choice]

machine, any thought about the NYY line going up? I like Detroit here too, but whats with the line movement? People don't think NYY will be done in game 4?

New York may use 10 pitchers today to keep from getting eliminated. Burnett will not be in the game long, that's why people are backing New York. I am staying with Detroit for the simple reason that they are lights-out on offense at home, they do not want to go back to New York, and they also will use any amount of pitchers necessary to END THIS tonight!

[Machine Choice]

Tuesday 10/4/11

Rangers 4
Rays 3

Yankees 4
Tigers 13

Brewers pk
Diamonds pk
Total runs: 6

Cards/Phillies: no plays (Holliday out)

Best Bet: Tigers -115

Graded Plays: 1-1

System Stats:
ML: 3-1
Totals: 3-2
Overall: 6-3

[Machine Choice]

Wed 10/05/11

Brewers 4
Diamonds 5

Phillies pk
Cards pk
Total runs: 8

Best Bet: Snakes 1st 5 innings (Brewers superior bullpen)

[ncsubowen]

Threw a half unit on both for the heck of it - lets go Diamondbacks!

[Machine Choice]

Wed 10/05/11

Brewers 4
Diamonds 5

Phillies pk
Cards pk
Total runs: 8

Best Bet: Snakes 1st 5 innings (Brewers superior bullpen)

LOL, nailed the total runs in the Cards game exactly. I wish I had this model during the regular season. That's okay, I will have it for next year.

[Machine Choice]

Wed 10/05/11

Brewers 4
Diamonds 5

Phillies pk
Cards pk
Total runs: 8

Best Bet: Snakes 1st 5 innings (Brewers superior bullpen)

Graded Plays: 2-1

System Stats:
ML: 4-1
Totals: 5-2
Overall: 9-3

[gunnersgun]

keep up the good work, what u got in mind for the game 5s?

[Machine Choice]

keep up the good work, what u got in mind for the game 5s?

I don't have anything in mind. The spreadsheet tells me what to do. I'll let you know, I'm loading it right now.

[Machine Choice]

Thurs: 10/06/11

Detroit 3
Yankees 9

Best bet: Yankees ML (I will form a -1 line by combining RL/ML to save juice)

[ncsubowen]

Nice call yesterday Machine. I like it.

[Raybolts]

The machine is a machine, TY

[ncsubowen]

LOL, nailed the total runs in the Cards game exactly. I wish I had this model during the regular season. That's okay, I will have it for next year.

As with most systems, I think you'll have better luck in the playoffs on this one. I think that on the whole its a lot harder to predict regular season, because the people that are out there playing may statistically have the ability to do something, but you never know what nights they're actually going to be trying out there.

The playoffs though, they're a whole different story

Excellent calls last night.

[Machine Choice]

System record is actually 9-3, not 8-3. I edited in the correction

[figue]

[No coincidences]

Thurs: 10/06/11

Detroit 3
Yankees 9

Best bet: Yankees ML (I will form a -1 line by combining RL/ML to save juice)

You think the Yankees are going to win by 6 but you aren't comfortable in just taking the RL?

[brewcrew2k]

You think the Yankees are going to win by 6 but you aren't comfortable in just taking the RL?

Crap in this game I wouldn't RL either, Fister had a bad year vs yanks But this is the playoffs

[Machine Choice]

You think the Yankees are going to win by 6 but you aren't comfortable in just taking the RL?

I have 33% of my wager on the RL, 66% on the ML. This creates a -1 line at juice of -110

[Machine Choice]

Graded Plays: 2-2

System Stats:
ML: 4-2
Totals: 5-3
Overall: 9-5