MLB Prediction Formula

Forum: Systems & Strategies Page 1 of 2  1 2  
Author: [Systems & Strategies] Topic: MLB Prediction Formula
marko123 PM marko123
Joined: Jan 2012
Posts: 120
 
marko123
Participation Meter
Prospect
Posted: 6/19/2012 1:01:07 PM
I'm not going to bother explaining and showing what the formula/system is until I notice good results. I'll post the daily plays here and see how it does and if it is hitting over 55% daily I'll fully explain it by the end of the week.

There are a ton of adjustments that I can add on to it so if it starts off horrible, I'll be changing it around. I'm starting off with an incredibly basic model. 

Yesterday I did a test run on CLE vs CIN and it predicted this:

CIN ; 5.8
CLE; 6.6

So, currently it is 1-0, +100 but obviously that is just one game.

Here are today's predictions:
(the bolded teams are the plays)

rockies   6.1
phillies  9.1

reds      7.6
indians   7.1

cards     4.8
tigers    3.9

braves    5.5
yankees   6.5

marlins   6.1
red sox   6.1

cubs      4.1
white sox 7.2

giants    3.3
angels    5.1

dodgers   4.4
athletics 4.8

rays      6.7 
nationals 4.2

twins     6.5
pirates   2.6

orioles   5.2
mets      7.6

royals    4.8
astros    7.8

blue jays -- 
brewers   --(first game for SP so not enough data)

mariners  6.9
dbacks    6.3


All plays are ML

marko123 PM marko123
Joined: Jan 2012
Posts: 120
 
marko123
Participation Meter
Prospect
Posted: 6/19/2012 1:03:34 PM
Forgot to add :

Rangers  6.9
Padres  6.2
DegenGamble PM DegenGamble
Joined: Oct 2011
Posts: 1070
 
DegenGamble
Participation Meter
Veteran
Posted: 6/19/2012 1:13:44 PM
If it pans out Marko - I can automate it
Posted using a mobile device.
DegenGamble PM DegenGamble
Joined: Oct 2011
Posts: 1070
 
DegenGamble
Participation Meter
Veteran
Posted: 6/19/2012 1:13:58 PM
Good Luck!
Posted using a mobile device.
marko123 PM marko123
Joined: Jan 2012
Posts: 120
 
marko123
Participation Meter
Prospect
Posted: 6/19/2012 1:17:46 PM
Thanks a lot Degen, this system is similar to what I first brought up in your Database thread. If it goes well tonight I'll PM you how it all works.
DegenGamble PM DegenGamble
Joined: Oct 2011
Posts: 1070
 
DegenGamble
Participation Meter
Veteran
Posted: 6/19/2012 3:57:06 PM
Yeah - that's what I figured. Good work - I'm rooting for ya
Posted using a mobile device.
marko123 PM marko123
Joined: Jan 2012
Posts: 120
 
marko123
Participation Meter
Prospect
Posted: 6/19/2012 4:33:34 PM
Alright so I did an incredibly small sample test just on yesterday (9 games) and these were the results.

Keep in mind, I used the SP's 2012 season data so I don't know how much yesterday's game would've affected the stats enough for it to change the predictions.

The fraction looking thing on the right of each match is this:
predicted winning margin/actual winning margin

While it doesn't seem to give accurate runs scored, for the most part it gives decent margins. This could be used for RL to lower juice whenever needed.

braves 5.7
yankees 8.6     +100     2.9/4

reds   5.8
indians  6.6     +100    0.8/1

cubs       6.8   +113    0.2/9
white sox  6.6

giants    6      +106    2.7/2
angels    3.3

orioles   3.1    +100    4.7/5
mets      7.8

royals    6.5    +100    1/2
astros    7.5

jays    5.5   +100      0.1/1
brewers 5.6

mariners  3.3   +100    4/6
dbacks    7.3

rangers  9.5    +100   5.4/1
padres   4.1

9-0    +919

This seems incredible but I'm going to wait for tonight before jumping to any conclusions prematurely. 

Btw, for anyone reading this, I am not playing tonight's picks, too early to be putting money on this.

marko123 PM marko123
Joined: Jan 2012
Posts: 120
 
marko123
Participation Meter
Prospect
Posted: 6/19/2012 9:43:36 PM
2-0 so far, not too confident on the others though.
marko123 PM marko123
Joined: Jan 2012
Posts: 120
 
marko123
Participation Meter
Prospect
Posted: 6/19/2012 10:15:39 PM
3-2, a few games still pending, hopefully its a positive return tonight with over 55% accuracy
glefkb01 PM glefkb01
Joined: Oct 2008
Posts: 2540
 
glefkb01
Participation Meter
Veteran
Posted: 6/20/2012 10:19:13 AM
six and six.....
DegenGamble PM DegenGamble
Joined: Oct 2011
Posts: 1070
 
DegenGamble
Participation Meter
Veteran
Posted: 6/20/2012 10:45:47 AM

Still strong work Marko!!! 

marko123 PM marko123
Joined: Jan 2012
Posts: 120
 
marko123
Participation Meter
Prospect
Posted: 6/20/2012 11:17:55 AM
rockies   6.1      +100                
phillies  9.1

reds      7.6      -100                 
indians   7.1

cards     4.8      -100              
tigers    3.9

braves    5.5      -134             
yankees   6.5


marko123 PM marko123
Joined: Jan 2012
Posts: 120
 
marko123
Participation Meter
Prospect
Posted: 6/20/2012 11:18:05 AM
cubs      4.1      -150 (-1 RL)    
white sox 7.2

giants    3.3     +100             
angels    5.1

dodgers   4.4   +100     
athletics 4.8

rays      6.7    +100      
nationals 4.2
 
twins     6.5   -105        
pirates   2.6

orioles   5.2    +100       
mets      7.6
 
royals    4.8    -115
astros    7.8

mariners  6.9  +150
dbacks    6.3

Rangers  6.9   +100
Padres  6.2
marko123 PM marko123
Joined: Jan 2012
Posts: 120
 
marko123
Participation Meter
Prospect
Posted: 6/20/2012 11:18:56 AM
Didn't let me put that all in one post.... weird.

7-6 +46  53%

IMO this is not a bad start considering the only stats I used are R/G and ERA. 

I will see how it does tonight but if it shows a loss or a small gain like this one I am going to upgrade the whole equation to use stats like component ERA, Form Reversal ERA, RC, Base Runs and other run estimators. 

Currently, the bullpen is not considered at all, so I will probably set it up so the SP is accountable for 6 innings and the BP for 3. 

I see no point in considering the indivdual contribution of each lineup but if anyone has any input feel free to respond.

I'll be posting today's day games soon and the rest later this afternoon.

Lastly, I think it is appropriate to play RL -1 on teams that are higher than -200 as it'll bring down the odds to about -150ish and theres no loss if they push. 
DegenGamble PM DegenGamble
Joined: Oct 2011
Posts: 1070
 
DegenGamble
Participation Meter
Veteran
Posted: 6/20/2012 11:41:20 AM
QUOTE Originally Posted by marko123:


 
I see no point in considering the indivdual contribution of each lineup but if anyone has any input feel free to respond.

Yeah - marko - the problem with using any run estimator (i.e. Base Runs, Runs Created, Extrapolated Runs) with individual batter vs pitcher metrics is the missing player data in matchups.  As you know some pitchers don't face batters in their career.

You have me intrigued on how you're calculating everything though.

 

marko123 PM marko123
Joined: Jan 2012
Posts: 120
 
marko123
Participation Meter
Prospect
Posted: 6/20/2012 11:52:12 AM
I know that most run estimators are created to be used only a team level except Extrapolated I think.

I'll PM you what I have so far but I am always playing around with it so I wouldn't take it too seriously. 

Today's Day Games:

braves    6.7
yankees   5.7

royals     6.7     ** I don't know if this is a large enough difference**
astros     6.8    

jays       7.0
brewers    6.7

mariners   4.6
dbacks     5.8

marko123 PM marko123
Joined: Jan 2012
Posts: 120
 
marko123
Participation Meter
Prospect
Posted: 6/20/2012 11:55:23 AM
Degen, check ur inbox.
marko123 PM marko123
Joined: Jan 2012
Posts: 120
 
marko123
Participation Meter
Prospect
Posted: 6/20/2012 12:15:56 PM
braves    6.7
yankees   5.7

jays       7.0
brewers    6.7

mariners   4.6
dbacks     5.8

I'm taking out the Astros play as the run difference was only 0.1 and with the rounding and whatnot, there is no value in the Astros. 
DegenGamble PM DegenGamble
Joined: Oct 2011
Posts: 1070
 
DegenGamble
Participation Meter
Veteran
Posted: 6/20/2012 1:23:17 PM
QUOTE

Originally Posted by marko123:

Degen, check ur inbox.

Marko - I sent you a reply with some questions - can you take a look and get back to me.

Thanks

marko123 PM marko123
Joined: Jan 2012
Posts: 120
 
marko123
Participation Meter
Prospect
Posted: 6/20/2012 4:52:55 PM
Braves win 1-0 so far. Glad I took out the Astros play. The jays are pissing me off though. They get 2-3 men on base each inning but can't finish anything. 
marko123 PM marko123
Joined: Jan 2012
Posts: 120
 
marko123
Participation Meter
Prospect
Posted: 6/20/2012 5:35:37 PM
Here are today's plays including the day games

rockies   8.0
phillies  7.4

braves    6.7       
yankees   5.7

cards     8.0
tigers    6.5

reds      6.5
indians   5.9

marlins   5.2
red sox   7.4

cubs      6.5
white sox 7.0

giants    3.9
angels    3.4

jays      7.0
brewers   6.7

mariners  4.6
dbacks    5.8

rangers   8.4
padres    4.2

twins     6.2
pirates   6.4

orioles   6.6
mets      7.1


1-1 so far, D'backs look good though.
DegenGamble PM DegenGamble
Joined: Oct 2011
Posts: 1070
 
DegenGamble
Participation Meter
Veteran
Posted: 6/20/2012 9:03:16 PM
Marko - check your email.
marko123 PM marko123
Joined: Jan 2012
Posts: 120
 
marko123
Participation Meter
Prospect
Posted: 6/21/2012 12:57:16 AM
Today's results:

7-5  58%  +232

YTD:    14-11     56%     +278


I am also going to start running a 2.0 version where a team qualifies as a play only if the predicted point differential is 1.0 or larger. This did well today so I wanna see if it is random or actually has value.

v2.0's YTD would be: 

9-4   69%   +454

Not many games tomorrow but hopefully we get some plays.
I wouldn't put money on this yet but I think by Monday we should know what we are dealing with here. I don't see it hitting above 60% but a 1-2 unit return a day is nice since you can increase your unit sizes instead of putting smaller units on more games that you may be unsure about. 

Good night to all and see you again tomorrow. BOL


marko123 PM marko123
Joined: Jan 2012
Posts: 120
 
marko123
Participation Meter
Prospect
Posted: 6/21/2012 11:09:55 AM
Today I want to add two more variations of the system.

v3.0 will use Pitcher Form Reversal ERA instead of the regular season ERA to take into account streaks and possible regression a pitcher might be doing.

v4.0 will simply take the average of season ERA and career ERA to take care of small data sets. A good example of this is today's game of Rockies vs Phillies. Francis has a 12.46 season ERA but using that ERA would completely throw everything off since I doubt he will have another 12+ ERA game. 
marko123 PM marko123
Joined: Jan 2012
Posts: 120
 
marko123
Participation Meter
Prospect
Posted: 6/21/2012 11:21:16 AM
v1.0 Plays

YTD:     14-11  56%    +2.78u


rockies     5.2
phillies    10.9  

dodgers      5.6
athletics    3.9

marlins      6.4
red sox      7.1

rays         4.0
nationals    5.5

twins        3.5
pirates      7.1

Forum: Systems & Strategies Page 1 of 2  1 2  
You have entered the forum as a GUEST. 
You must login/register to post or reply.