Playing favs only on ML and dogs on ML/spread blend....

[curious]

I have stopped playing favs on the spread and play them only on the ML, if the value is there. By "if the value is there" I mean that my calculation of what the ML odds should be is <= what the sportsbooks are actually setting the ML at. If the sports books are setting the ML at > what my calculation shows that it should be then I take the dog on the ML and I hedge the dog ML with the dog plus the points.

To determine the value I use the pythag theorem and the log5 formula with the pythag adjusted for strength of schedule.

This has been absolutely on fire, and so far I have not stepped on any land mines (large favs on the moneyline that lose straight up).

Any criticisms of playing medium to large favs on the moneyline using the above approach?

Am I risking walking into a minefield?

[Ganchrow]

Check out this spreadsheet I put together a little while back in order to perform just these sorts of ML/Spread optimizations.

Shaded cells are the ones intended to be modified by the user.

Bugs may well exist. If anyone finds any please do post them here.

[therber2]

Hello there. Could you elaborate or give an example of how you "determine the value"?

I have stopped playing favs on the spread and play them only on the ML, if the value is there. By "if the value is there" I mean that my calculation of what the ML odds should be is <= what the sportsbooks are actually setting the ML at. If the sports books are setting the ML at > what my calculation shows that it should be then I take the dog on the ML and I hedge the dog ML with the dog plus the points.

To determine the value I use the pythag theorem and the log5 formula with the pythag adjusted for strength of schedule.

This has been absolutely on fire, and so far I have not stepped on any land mines (large favs on the moneyline that lose straight up).

Any criticisms of playing medium to large favs on the moneyline using the above approach?

Am I risking walking into a minefield?

[curious]

Hello there. Could you elaborate or give an example of how you "determine the value"?

Use the pythagorean theorem adjusted for strength of schedule, then use the log 5 formula to determine probability of winning a given matchup then convert the probability into moneyline odds.

[therber2]

Sorry I got that before. Could you explain how to do this? That is what I meant.

[curious]

Sorry I got that before. Could you explain how to do this? That is what I meant.

http://www.rockytoptalk.com/story/2008/2/27/165520/578

I just use the pythag and log5, I don't do the predict the score of the games part. I only play the favs on the ml so I don't care what the spread is.

[therber2]

Looks useful. I will take a look. You are a gentleman kind sir.

[therber2]

curious....I am sorry, about Gonzaga. Ouch. But happened to doing the dogs with this system. Not to rub salt in the wound, but I did Illinois tonight using this idea, and it payed of really well. Its not a bad strat at all. Thanks brutha.

[curious]

curious....I am sorry, about Gonzaga. Ouch. But happened to doing the dogs with this system. Not to rub salt in the wound, but I did Illinois tonight using this idea, and it payed of really well. Its not a bad strat at all. Thanks brutha.

My YTD is 253-211 +3600 Units. And that includes a 15,000 unit loss. LOL I'm not worried.

Merry Christmas!!

[therber2]

Hey Curious,

Do you know of any of sites that do the same thing KENPOM does...except for other sports? I'm not good with this math stuff.

Thanks and glad to hear you will be alright!

Merry XMAS to you too!

[curious]

Hey Curious,

Do you know of any of sites that do the same thing KENPOM does...except for other sports? I'm not good with this math stuff.

Thanks and glad to hear you will be alright!

Merry XMAS to you too!

There are several baseball sites. I haven't found any others. You'll have to do a google search to find the baseball sites.

For NFL and NCAAF I use plays per TD and big play capability. Big play capability is a stat that shows the premium that the passing game gives the offense. I didn't do any football this year because I have to keep the stats myself to calculate plays per TD and big play capability, no one publishes it. I just started fooling with sports again in November so I didn't have the stats for football.

Last year I totally tore up NCAAF. Playing mostly big dogs on the moneyline with a hedge on the spread. Not blind, using my models.

I do really well at baseball.

[therber2]

Curious,

I like how KENPOM's page for NCAA-B is so clear. Is there anyone the does the same system (publishes it) for NBA or any other sport.

I specifically enjoy the fact that you can check probabilities sooooo quickly

KENPOM is the only one I can find that does this kind of thing, and it seems to be spot on correct. I love his system. I dread NCAA-b season ending.

Please let me know if you know of anything or does KENPOM do other sports???

[smokin1]

Here's an interesting question. How about playing both Fav ML and dog spread. Most of the time it will be a loser, but only time its significant is when you lose the ML Fav bet. However, sometimes when the fav wins but doesnt cover the spread you win both.

Anyone have stats on how many times the dog won su on various spread values. I bet at spreads greater then 7 it may be possible to make money? Obviously excluding spreads > -1500 ML on Fav would prolly be advisable as the ML has become overinflated and not representative of the true odds of the fav winning imo.

Just a theaory I had with no long term analysis

[curious]

Here's an interesting question. How about playing both Fav ML and dog spread. Most of the time it will be a loser, but only time its significant is when you lose the ML Fav bet. However, sometimes when the fav wins but doesnt cover the spread you win both.

Anyone have stats on how many times the dog won su on various spread values. I bet at spreads greater then 7 it may be possible to make money? Obviously excluding spreads > -1500 ML on Fav would prolly be advisable as the ML has become overinflated and not representative of the true odds of the fav winning imo.

Just a theaory I had with no long term analysis

Well my big favs are 48-3. Small favs are roughly 40-60. ML medium sized to big dogs are 50-50 which means ML dogs are a huge winner. Small ML dogs are awful.

Medium and big dogs on the spreads are way up.

I don't think fav ML and dog on the spread is a good play.

[FreeFall]

curious I have calculated the phytag for todays game. With P(W) for Connecticut being .9832. How do I turn that into a money line to see if there is an edge compared to what the books offer?

Thank you.


EDIT:

Where can I find the average points per 100 possessions nationwide?

For each team we'll need to compare their offense and defense to the average, just as we did with tempo. The average points per 100 possessions nationwide is 101.5. That means that Tennessee's offense is 117.6% of the average. Its defense is 86.9% of the average. For Kentucky, these numbers are 104.9% and 92.7%. Remember, we already added in the homecourt advantage, so these really represent how we expect UT to do at home and how we expect UK to do on the road.

Now getting each team's expected output is easy. We simply multiply their offense by the opposing defense and then by the average output:

Tennessee expected output = 117.6% * 92.7% * 101.5 = 110.6
Kentucky expected output = 104.9% * 86.9% * 101.5 = 92.6

I'm having trouble obtaining these numbers of 100 possessions for todays game using that website. Any help would be appreciated.

[curious]

curious I have calculated the phytag for todays game. With P(W) for Connecticut being .9832. How do I turn that into a money line to see if there is an edge compared to what the books offer?

Thank you.


EDIT:

Where can I find the average points per 100 possessions nationwide?

For each team we'll need to compare their offense and defense to the average, just as we did with tempo. The average points per 100 possessions nationwide is 101.5. That means that Tennessee's offense is 117.6% of the average. Its defense is 86.9% of the average. For Kentucky, these numbers are 104.9% and 92.7%. Remember, we already added in the homecourt advantage, so these really represent how we expect UT to do at home and how we expect UK to do on the road.

Now getting each team's expected output is easy. We simply multiply their offense by the opposing defense and then by the average output:

Tennessee expected output = 117.6% * 92.7% * 101.5 = 110.6
Kentucky expected output = 104.9% * 86.9% * 101.5 = 92.6

I'm having trouble obtaining these numbers of 100 possessions for todays game using that website. Any help would be appreciated.

To calculate probability of winning you need Fairfield's pythag then you use the log5 formula to get a probability, then you convert that to a line. If the line that the books are offering is better than that line you have a play, if the line the books offer is worse than that line then Fairfield is the play. The bigger the difference between the two lines the stronger the Fairfield play is. If the difference is really large then you take Fairfield on both the spread and the ML.

www.kenpom.com gives all the stats you need including average points per game nationwide.

[curious]

Here are the numbers I got for the UConn game

Pythag
Fairfield .57
UConn .98

Probability of winning
F 2.63%
UConn 97.37%

Value line -3700

Actual line -6000

Play is Fairfield + the points and 1/2 that on Fairfield ML

[FreeFall]

Pythag
Fairfield .57
UConn .98

Probability of winning
F 2.63%
UConn 97.37%

Value line -3700

Actual line -6000

Play is Fairfield + the points and 1/2 that on Fairfield ML

What numbers did you use for fairfield. I got the same numbers for Conn, but for Fair Field I got .4865.

I used

(102.7412)^11.5
_________________

(102.7412^11.5 + 103.2252^11.5)

that gave me .4865


also how do I turn the probability for Conn to win of .9832 into a money line?

[curious]

What numbers did you use for fairfield. I got the same numbers for Conn, but for Fair Field I got .4865.

I used

(102.7412)^11.5
_________________

(102.7412^11.5 + 103.2252^11.5)

that gave me .4865


also how do I turn the probability for Conn to win of .9832 into a money line?

You are not adjusting for strength of schedule. I adjust for strength of schedule. That is why our numbers are different.

Here is the excel spreadsheet formula to turn the probability into a value line:
ROUNDUP((A1/(1-A1))*-100,0)

Assuming A1 contains a probability which is >= 50%.

[smokin1]

My YTD is 253-211 +3600 Units. And that includes a 15,000 unit loss. LOL I'm not worried.

Merry Christmas!!

I did a little research and my above theory of playing bothfav ml and dog spread turns out to break even for this season.

The reason for my alternative thought pattern remains tho. I was looking for a way to hedge your bets or limit your results to particular spreads because a 15000 unit loss and only positive a +3600 return thus far are kind of scary numbers to me. To be able to lose almost 5x your gain for the season on 1 play

Why kind of play was this? What is your standard unit size for a game. I mean it wasnt a -15000 odd bad beat I assume. Was it a -1500 bet @ 10 units? Have you looked into ways to limit your risk and maintain your profits?

[curious]

I did a little research and my above theory of playing bothfav ml and dog spread turns out to break even for this season.

The reason for my alternative thought pattern remains tho. I was looking for a way to hedge your bets or limit your results to particular spreads because a 15000 unit loss and only positive a +3600 return thus far are kind of scary numbers to me. To be able to lose almost 5x your gain for the season on 1 play

Why kind of play was this? What is your standard unit size for a game. I mean it wasnt a -15000 odd bad beat I assume. Was it a -1500 bet @ 10 units? Have you looked into ways to limit your risk and maintain your profits?

the odds were -5000 on Gonzaga vs Portland St. I went against the prediction model because the value line was -3667 so the play was actually portland state. The play was -5000 ML 15000 units to win 300 units.

Some colleagues have said not to play lines > -1000.

There is no way to hedge this big ML fav plays because the vig on the dog is so large. If the dog was the mirror of the fav then you could take the fav the moment the line opened and take the dog right before game time, but with so much vig that won't work.

The only way to limit risk is to cut the size of the fav that youa re willing to play at some arbitrarily line like -1000. I'm not sure I agree with that.

Even after losing 15000 units I was still up 3300 units for the month.

[smokin1]

the odds were -5000 on Gonzaga vs Portland St. I went against the prediction model because the value line was -3667 so the play was actually portland state. The play was -5000 ML 15000 units to win 300 units.

Some colleagues have said not to play lines > -1000.

There is no way to hedge this big ML fav plays because the vig on the dog is so large. If the dog was the mirror of the fav then you could take the fav the moment the line opened and take the dog right before game time, but with so much vig that won't work.

The only way to limit risk is to cut the size of the fav that youa re willing to play at some arbitrarily line like -1000. I'm not sure I agree with that.

Even after losing 15000 units I was still up 3300 units for the month.

If your system is sound, which Im not arguing at all (I like it alot actually), I completely agree that in the long run playing -5000 lines would produce positive results. My only fear is there are not enough games like that in the season to allow the laws of probability to even out.

Im just curious, can you look at your plays for the season and just eliminate anything you bet on a fav at odds over 1000 or 1500 or whatever. Pick a few different cutoff points and see if you retain your positive gain. When I looked at games this season using the theory I mentioned, I only broke even when eliminating games with spreads under 6 and over 13. When I included those plays I started seeing negative results for the season.

While the theory is completely different, those cutoffs on spread values do make a little sense, especially the upper limit of 13 since it takes so many positive results to balance 1 bad result.

Good luck on your plays tonight tho, I had no interest in that game till this thread

[curious]

If your system is sound, which Im not arguing at all (I like it alot actually), I completely agree that in the long run playing -5000 lines would produce positive results. My only fear is there are not enough games like that in the season to allow the laws of probability to even out.

Im just curious, can you look at your plays for the season and just eliminate anything you bet on a fav at odds over 1000 or 1500 or whatever. Pick a few and see if you retain your positive gain. When I looked at games this season using the theory I mentioned, I only broke even when eliminating games with spreads under 6 and over 13. When I included those plays I started seeing negative results for the season.

While the theory is completely different, those cutoffs on spread values do make a little sense, especially the upper limit of 13 since it takes so many positive results to balance 1 bad result.

Good luck on your plays tonight tho, I had no interest in that game till this thread

Here are all my fav ML plays (well I might be missing one or two). I had a whole bunch of big dog ML wins which is why I am still up after losing that 15000 play. The dogs are not in this list.

Loyola Chicago -150 win
USF -155 loss
Seton Hall -155 win
Oklahoma St -160 win
Riverside -160 win
Samford -160 win
Utah St -170 win
Morehead St -180 loss
Purdue -200 win
St Mary's -200 win
Cinci -200 win
Riverside -200 win
Gonzaga -200 win
South Alabama -215 win
Boston College -220 win
Utah St -225 win
Templ -235 loss
Tenn -240 win
Illinois Chicago -290 win
Houston -300 win
Irvine -315 win
Creighton -325 win
Cleveland St -360 win
Pittsburgh -360 win
Portland St -375 win
Kansas -440 win
Wake Forest -475 win
MTSU -500 win
California -500 win
George Washington -550 win
Wichita St -525 win
Washington St -650 win
Georgia -800 win
Oklahoma -850 win
Tulsa -1000 win
SD State -1000 win
Northridge -1000 win
Florida -1100 win
Vanderbilt -1100 win
ST Johns -1200 win
Missouri St -1400 win
Charlotte -1400 win
GA Tech -1500 win
MTSU -1600 win
VA Tech -1800 win
Arkansas St -2200 win
Wyoming -2400 win
Washington -2500 win
Ohio St -2500 win
Kentucky -2500 win
Oklahoma -3000 win
Northwestern -3000 win
Gonzaga -5000 loss
Georgetown -15000 win

[smokin1]

Arkansas St -2200 win
Wyoming -2400 win
Washington -2500 win
Ohio St -2500 win
Kentucky -2500 win
Oklahoma -3000 win
Northwestern -3000 win
Gonzaga -5000 loss
Georgetown -15000 win

My guess is if you eliminated all of these plays from your results your units for the season would be > +3300

Just something to consider. Im planning on playing with your strat this weekend, thank you for sharing and good luck!

[therber2]

Curious, when you pick your favs do you weigh in regional finals playoffs. Like do you weigh in one team playing harder because they need it to move on and the other taking it easy.

For example, NBA: Boston and Cleveland games ....wow what bums!

Your system is great, but this is the one thing that I think might throw it off coming up soon.

[curious]

Curious, when you pick your favs do you weigh in regional finals playoffs. Like do you weigh in one team playing harder because they need it to move on and the other taking it easy.

For example, NBA: Boston and Cleveland games ....wow what bums!

Your system is great, but this is the one thing that I think might throw it off coming up soon.

I weigh strength of schedule the most, then pythag, then HCA, then regional rivalry.

I only look at HCA and regional rivalry if the value line and the line offered by the books are close. If the value line shows a clear advantage then I don't look at HCA or rivalry.

I think this "Team A will play harder..." stuff is a myth.

When I say regional rivalry I am talking about teams like W Kentucky at home against any of their nearby rivals where the fans hate the other team.

[andywend]

Did you actually lay 150-1 odds on Georgetown as indicated on your list?

If you are doing your betting through a regular sportsbook as opposed to a betting exchange, I can almost guarantee you will lose money long term betting huge favorites to win on the moneyline.

The reason being the "vig" charged on those type games is huge.

[curious]

Did you actually lay 150-1 odds on Georgetown as indicated on your list?

Yes

If you are doing your betting through a regular sportsbook as opposed to a betting exchange, I can almost guarantee you will lose money long term betting huge favorites to win on the moneyline.


The reason being the "vig" charged on those type games is huge.

Perhaps.

I calculate a value line and if the line offered by the books is better than that I take it. My only loss has happened when I let my personal bias overrule what the prediction model said.

[willpies]

Hey Curious obviously this a great formula and props to you. Could you please upload a spread sheet where you just punch in values and it does the rest for you? If its easy i mean of course. Either way good work

[curious]

Hey Curious obviously this a great formula and props to you. Could you please upload a spread sheet where you just punch in values and it does the rest for you? If its easy i mean of course. Either way good work

Sure, I will just give away the system that I spent hundreds of hours creating. Sure I will....Okay hold your breath until I upload it...

[innovator]

curious will you be my Obi Wan Kenobi?

[therber2]

Sure, I will just give away the system that I spent hundreds of hours creating. Sure I will....Okay hold your breath until I upload it...

Why don't you sell it?

Just make it look nice. Don't put that picture of the Wookie on there.

Well come on now, Curious. This is a friendly....ehm sharing forum

On a serious note: Curious, or (calling) Ganchrow. I have done some data mining public trends (spent hours looking for patterns and compiling data into an excel file). I can't make anything out of it. What I have been noticing though (on bettracker) is that usually when the public thinks they're right (on a bookie's spread play) they are wrong.

It makes sense. Explains why the bookies are so hard to beat and we need geniuses like you guys to figure out these overly complicated ways to win some fair cash.

Would you like to take a look at my file? Maybe you'll notice something that I didn't.

[therber2]

Here is what I am thinking plain and simple.

High market volume and weight on a spread play + frequent line move in a favorable direction as more bets comes in = bet that 10% spread play that no one wants to lay their hands on (or if you like the popular bet tease it with another low stakes).

I wish I had like maybe $50,000 to test my whims out. It would be uhm.....fun. Yeah.

I think I'll watch this. I just wish bettracker and bookies archived things. POMROY too. Geead damn I hate doing all nighters to data mine a whim. In the end it is paying me $-5/hr. Ugh!

[pico]

some links i found

http://www.rockytoptalk.com/story/2008/2/27/165520/578

http://kenpom.com/rate.php

http://kenpom.com/blog/index.php/weblog/ratings_explanation/

http://www.diamond-mind.com/articles/playoff2002.htm

http://en.wikipedia.org/wiki/Pythagorean_expectation

[Axis]

I've read through this thread, and maybe I really did miss it, but what is the Log5 formula that is being discussed? Or if you don't want to explain it, is there a link where I can find it?

I made my own spread sheet this morning for the Pythag, but I'm interested in trying Curious' method, but I'm still missing that Log5 Formula.

Thank you to anyone who can help.