TOTALCON Mark II - MLB Total Calculator

[frogsrangers]

frogger, I'm curious what you meant when you said all of the signs in STL/SFG point towards over? What signs did you mean exactly?

Bumgarner pitches worse on the road
Giants hit better away from home
Cards hit better at home
Wainwright pitches worse at home
Weather favored the over
Both pitchers had adjusted averages and career averages over .300

[frogsrangers]

Today's stats:

If you played every capped total, you would have gone 5-4 (3-3 unders, 2-1 overs)
If you only played capped totals with a difference of one run or more, you would have gone 3-4 (2-3 unders, 1-1 overs)

Life to date stats:

If you played every capped total, you would have gone 19-15 (13-9 unders, 6-6 overs) 55.8%
If you only played capped totals with a difference of one run or more, you would have gone 9-10 (6-7 unders, 3-3 overs) 47.3%

[frogsrangers]

CIN @ CHC
---------
OFF


STL @ PHI
---------
Crude Total: 7.5 -115o/-105u
D1: +1.53
D2: -0.30
D3: -0.12
Capped Line: 8.0 +100o/-120u
Actual Line: 7.5 -110o/-110u


SD @ PIT
--------
Crude Total: 6.0 -105o/-115u
D1: -3.71
D2: -1.44
D3: -0.80
Capped Line: 5.5 -110o/-110u
Actual Line: 8.0 -120o/+100u


ATL @ NYM
---------
Crude Total: 9.0 -115o/-105u
D1: +1.47
D2: -1.42
D3: -0.731
Capped Line: 9.0 -125o/+105u
Actual Line: 8.0 -110o/-110u


LAD @ MIA
---------
Crude Total: 6.5 -105o/-115u
D1: -1.75
D2: -0.81
D3: +0.12
Capped Line: 6.0 -120o/+100u
Actual Line: 7.0 +100o/-120u


MIL @ HOU
---------
OFF


WAS @ ARI
---------
Crude Total: 8.5 +105o/-125u
D1: +2.31
D2: -0.05
D3: +0.403
Capped Line: 8.5 -115o/-105u
Actual Line: 8.0 -120o/+100u


COL @ SF
--------
Crude Total: 6.5 -110o/-110u
D1: +0.54
D2: +1.57
D3: -0.89
Capped Line: 7.0 -105o/-110u
Actual Line: 7.0 -120o/+100u


KC @ BAL
---------
Crude Total: 10.0 -115o/105u
D1: +2.43
D2: +1.17
D3: +0.29
Capped Line: 10.5 -110o/-110u
Actual Line: 9.0 -115o/-105u




BOS @ CLE
---------
OFF




NYY @ TOR
---------
Crude Total: 10.0 -120o/+100u
D1: +1.72
D2: -0.79
D3: +0.02
Capped Line: 10.5 +105o/-125u
Actual Line: 9.5 -110o/-110u


DET @ TEX
---------
Crude Total: 9.0 -110o/-110u
D1: +2.61
D2: +2.96
D3: +1.16
Capped Line: 10.0 -110o/-110u
Actual Line: 10.0 -110o/-110u


OAK @ CHW
---------
Crude Total: 8.5 -105o/-115u
D1: -0.87
D2: -3.17
D3: -0.20
Capped Line: 8.0 -110o/-110u
Actual Line: 8.5 -105o/-115u


MIN @ TAM
---------
Crude Total: 7.5 -110o/-110u
D1: -1.38
D2: -2.27
D3: -0.36
Capped Line: 7.0 +100o/-120u
Actual Line: 9.0 -105o/-115u


SEA @ LAA
---------
Crude Total: 8.0 -115o/-105u
D1: -0.41
D2: -0.62
D3: +0.32
Capped Line: 8.0 -105o/-115u
Actual Line: 7.5 -105o/-115u

[frogsrangers]

My plays today:

MIN/TAM U 9
KC/BAL O 9
LAD/MIA U 7

[mathdotcom]

frogs you're killing me

you have the programming talent to actually make it big here but your model isn't doing any better evaluating meaningless stats than some random degenerate sitting with the newspaper in front of him at a sportsbook

[mathdotcom]

Your first step should be to test what variables affect the game results, not just assuming everything does and throwing it into a blender

[frogsrangers]

Your first step should be to test what variables affect the game results, not just assuming everything does and throwing it into a blender

I'll be the first to say that lead-off extra base hits and 2 out base hits are major factors

[phillybadboy]

free play take the reds to win 10 dimes gl

[Utah_Jay]

free play take the reds to win 10 dimes gl

Thanks for the free play now that the game is in the bottom of the 6th! How did you ever come up with this winner.

[mathdotcom]

I'll be the first to say that lead-off extra base hits and 2 out base hits are major factors

Sigh..

I highly recommend you read http://en.wikipedia.org/wiki/Kitchen_sink_regression and the associated links

I'm trying to save you money

[The Kraken]

frogs if you want to make money, I'd recommend you take Mathy's advice and use this as a learning opportunity. It's not often Mathy is helpful anymore but apparently he sees something in you that he is willing to take that step and from what i can tell, you're either disinterested or egotistical.

And by the turnout of your last thread and the direction of this one, you don't have a reason for either.

[mathdotcom]

frogs if you want to make money, I'd recommend you take Mathy's advice and use this as a learning opportunity. It's not often Mathy is helpful anymore but apparently he sees something in you that he is willing to take that step and from what i can tell, you're either disinterested or egotistical.

And by the turnout of your last thread and the direction of this one, you don't have a reason for either.

I have 0 advice to handicappers... I mostly just feel sorry for them

I try to help modelers

Unfortunately SBR is 99.99% handicappers and they all lose

[The Kraken]

All programmer, no math.

[frogsrangers]

Looks like I am about to throw in the towel here

I agree with mathdotcom and Kraken, I don't think there is enough mathematical integrity in this and after reading that wikipedia article about kitchen sink regression, that is pretty spot on as to what this is

So I am going to shut down operations and begin to study

mathdotcom, I am not asking for much here, I don't want to hound you and be all over you asking on how stuff like this works. But, one favor I could ask is if you could provide me links to some websites that can introduce me to models and regression and links that will help me get started on studying how to produce some kind of model.

I consider myself somewhat good at math, I took calculus and all that in high school and some classes in college, but not well versed in it either.

So if you know of anything that will help me get on the right track, I would like to see it

I assume Nate Silver is a baseball modeler, so anything to do with him would be great too

[Extra Innings]

Looks like I am about to throw in the towel here

I agree with mathdotcom and Kraken, I don't think there is enough mathematical integrity in this and after reading that wikipedia article about kitchen sink regression, that is pretty spot on as to what this is

So I am going to shut down operations and begin to study

mathdotcom, I am not asking for much here, I don't want to hound you and be all over you asking on how stuff like this works. But, one favor I could ask is if you could provide me links to some websites that can introduce me to models and regression and links that will help me get started on studying how to produce some kind of model.

I consider myself somewhat good at math, I took calculus and all that in high school and some classes in college, but not well versed in it either.

So if you know of anything that will help me get on the right track, I would like to see it

I assume Nate Silver is a baseball modeler, so anything to do with him would be great too

Good decision, sure there are a couple of folks that would be willing to help guide you. All the best!

[shooms79]

I say just continue to track it and tweak

[The Kraken]

I say just continue to track it and tweak

Shooms, tweak what? He doesn't have a good model. Not saying he's not capable but what he has right now is not it. Frogs, you ever had a college level stats course?

You can polish a turd all day but in the end you still have a piece of shit

[Smutbucket]

Throw in the towel this early? You can do what you want with your time, but honestly I dont see why mathdotcoms link about kitchen sink regression should scare you, I thought that link was common sense/well known, the key to this and handicapping obviously is knowing which are the right stats/variables to evaluate the game. you just need to tweak yours around to find the right mix but the only way you can truely learn is by trying (and failing)....mathdotcom doesnt know shit about handicapping, thats why he has 0 advice to offer about it....and honestly dont think he knows that much about programming either. In the last thread he started off saying "YOU HAVE NO IDEA WHAT YOUR DOING" and now he says "you have the programming talent" My question is does mathdotcom actually know what good programming talent is or does he know the difference between important variables/stats and not important ones? my guess is no, but then again I may be wrong, just my guess from the little snippets Ive seen from him here and there..

anyways good luck, frogs, whatever you decide to do

[mathdotcom]

Looks like I am about to throw in the towel here

I agree with mathdotcom and Kraken, I don't think there is enough mathematical integrity in this and after reading that wikipedia article about kitchen sink regression, that is pretty spot on as to what this is

So I am going to shut down operations and begin to study

mathdotcom, I am not asking for much here, I don't want to hound you and be all over you asking on how stuff like this works. But, one favor I could ask is if you could provide me links to some websites that can introduce me to models and regression and links that will help me get started on studying how to produce some kind of model.

I consider myself somewhat good at math, I took calculus and all that in high school and some classes in college, but not well versed in it either.

So if you know of anything that will help me get on the right track, I would like to see it

I assume Nate Silver is a baseball modeler, so anything to do with him would be great too

First I will say good programming skills are more a bonus than a requirement. They will help you arrange the data quickly and easily into the format you want, and a good programmer will write efficient code so that if they want to change the model it can be done easily. A poor programmer will probably have to re-write most of it. Personally I am somewhere in between. The time saved through good programming just allows you to tackle more things. That's why I feel you have some potential.

The reason I blabbed about programming is that the best way to model baseball is probably via Monte Carlo simulation. However that is probably over your head given what you've written.

That said, why do you feel you have to create a model for baseball totals? You're competing against hundreds of very sharp guys with their own models in this area that they've probably been tweaking for years. At best if your model works out you'll have an edge of a few cents. The reason these guys bother with a totals model is they have acquired a large BR over time and need to bet in large markets in order to get enough money down. But if your BR is below $50,000, why not go after lower hanging fruit? Go after props, derivatives, smaller market sports. These will all have softer lines, and the lower limits may not be an issue for you because you may not want to bet more than $500-$1000 anyways. Maybe you can focus on 1H or 2H markets and hope for a bigger edge. Sometimes for these markets you may not even have to run a regression; simply looking at an average or median can be enough. But for anything remotely complicated, you will probably need to use regression tools and for that I just recommend googling Intro to Regression and spending a lot of time reading. Unfortunately there are no shortcuts, but there will be a lot of examples out there of why you don't want to just put any variables mindlessly into a regression. A general rule is be careful about testing what variables matter, and if you are confident they don't matter, don't include them in your final regression. How do you know what variables should be tested? Well you need some knowledge of the sport.

Good luck

[fitguy67]

reading thru this thread: http://forum.sbrforum.com/handicappe...-research.html

and contacting some of the sharp contributors to it couldn't hurt...

Two guys who've made constructive contributions to the TOTALCON threads would also be very worth pm'ing...Inspirited and Ex-Hoosier

One thing I very much agree with Mathy about is that Statistics & Probability are the key quant areas...combined with a knowledge of the sport...programming skills can certainly help propel things along (especially in back-testing)

[EXhoosier10]

reading thru this thread: http://forum.sbrforum.com/handicappe...-research.html

and contacting some of the sharp contributors to it couldn't hurt...

Two guys who've made constructive contributions to the TOTALCON threads would also be very worth pm'ing...Inspirited and Ex-Hoosier

One thing I very much agree with Mathy about is that Statistics & Probability are the key quant areas...combined with a knowledge of the sport...programming skills can certainly help propel things along (especially in back-testing)

WOO thanks for the shoutout fit!!!

Any update on this model, frogs?

Fit, did you do any work in the offseason on the stuff you and Pacocn were working on?