Vegas vs KenPom Week 9

Correct.

Here is a calculator where you can change the inputs to match your desires.

Gambling, in general, is questionable

Portland St. shouldn't be laying DD to anyone.

Kelly Criterion Method
Yesterday: 0-2
Since 1/13/2012: 61% or 14-11
Total: +164.82

Week 9 Inputs
Starting bankroll: 500.00
Home, Favorite, Away (HFA) 56.4% or 42
Home, Dog, Away (HDA) 57.4% or53
Road, Fav, Away (RFA) 59.5% or 75
Road, Dog, Away (RDA) 60.2% or 82
Home, Fav, Towards (HFT) 52.8% or 4
Home, Dog, Towards (HDT) 53.9% or 16
Road, Fav, Towards (RFT) 54.5%% or 22
Road, Dog, Towards RDT) 55.5% or 33

Note, I left the starting bankroll input at 500.00 for week 9.  I'm going to wait for a full week to change this input, so all inputs will be refreshed at the start of week 10.

Home Teams: 75-68 (52%)
Road Teams: 58-41 (59%)

Favorites: 64-57 (53%)
Dogs:      88-68 (56%)

Line moved toward KP:  103-97 (52%)
Line moved away KP:  40-17 (70%)

Your weighted averages would be:

• Home - Fave - Away = ((52%*143)+(53%*121)+(70%*57))/(143+121+57) = 55% OR ((Home % * Total Home games)+(Fave % * Total Fave games)+(Away % * Total Away games)) / (Total Home games + Total Fave games + Total away games)
• Home - Dog - Away 57%
• Road - Fave - Away 59%
• Road - Dog - Away 60%

The only problem w/ that is that w/o knowing where the overlap in buckets are, you may be double counting some bad (or good) %'s.
eg.  Let's suppose 10 of the 57 away bets came from Dogs, then you're calc for Home teams should not be using the same %. 

I think to use your calcs the way you want (which btw is a great idea) - you have to really isolate the buckets, i.e.  7 Home teams were favs and moved away, vs maybe 35 away teams were faves & moved away. (These numbers are hypothetical, I don't calc the actual history)

Does that make sense?  Basically some of your calcs *should* get closer to the 70% hit rate that we're seeing on the Away numbers.

Or maybe I'm totally missing the logic and am way off

Agree to get the proper %'s the records sb broken out into the 8 different possabilities.

HFA

HDA

RFA

RDA

HFT

HDT

RFT

RDT

This is a lot of research(8 weeks of threads) unless someone has kept track by category.Agree with Schiman that we could uncover some better %'s this way

I agree with the statement that the sound method to determine the accuracy levels of 8 distinct combinations is to track and record each combination individually. 

However, I was not sold that this was the most efficient and effective way to determine the inputs into the Kelly calculator, at this point in time.

Some thoughts on why:
1. I don't have access to the required data, at least not in a semi-formatted manner.
2. I doubted that some combinations (e.g. RFA) would have had totalled enough plays to justify their percentage inputs (i.e. too small of sample).

Thus, using a weighted average right now:
1. I personally think is better than using a simple average.
2. Allows for the usage of the Kelly Criterion, instead of selecting all 8 combinations as equal.  Again, personal preference.
3. Most importantly, allows me to leverage the high totals of the individual components to "massage" the data into what it will most likely look like.  Theory of large numbers.

I agree that isolated buckets is the correct way to go, but I feel more comfortable with the weighted average until each isolated bucket has a larger sample of games.

Quote Originally Posted by beaverfan23:

Stiff and flyer,

you guys are right about michigan.  In the live odds section under the ncaab tab, it has the opening line listed at +2.  I wish they could get this right.



Official Plays are:
Depaul +9 (HDT)
Michigan +1 (HDA) - this line may still move toward the kp #

Hate the depaul play (which means its probably right...haha) Good stuff Beaver, thanks again. Looking forward to tomorrows card.