Buying Points and Related Issues/Sports Betting Theory Question

[YoureToast]

"Hypothetically", Team A is favored against Team B by 5. I think Team A has a good chance of beating Team B by 20 points. Therefore, I am looking to get action on Team A at a point spread higher than 5. Suppose I could bet on Team A, give 5 points at -110. How do I go about determining at what number I should get to increase the line to 7? Related, if I can get the line at -10.5, how do I figure what fair value is?

I'm looking here not for some sort of general "feeling" type answer, but more of a mathematical approach.

Syg? Anyone?

Check out the NFL Pointspread Ties thread if you haven't yet.

Part 1 -- Using Tradesports as example, you can currently buy Atlanta -5.5 for -105.7 (51+(0.8 commission))/49 meaning the wager price converted to a percent is 51.39% (518/1008).

Part 2 -- you can expect the game to end on 6 and 7 with rough frequency of 3.10% and 5.36%, respectively. This means the value of Atlanta -7.5, offered on Tradesports, is equivalent to (51.39%-3.10-5.36)=42.93%. Why subtract the numbers in entirety? Because the new line moves over these numbers in whole and shifts 6 and 7 directly from win column to loss column.

Part 3 -- therefore if you made buy offers of 42.0 or fewer (or market making offers of 42.5 or fewer) you will be paying a lower amount for the -7.5 bet than you would for the -5.5 bet. (ed Non-market making offer of 42 is equivalent to (42.0+(0.8 comm))/(58.0+(42.0+(0.8 comm)) or 42.46%, below the relative expected frequency of 42.93%.

I don't compute values for landing on 8, 9 & 10 so can't estimate the Atlanta -10.5 contract on TS, nor is the market for that contract particularly liquid.

[YoureToast]

[ QUOTE ]
Check out the NFL Pointspread Ties thread if you haven't yet.

Part 1 -- Using Tradesports as example, you can currently buy Atlanta -5.5 for -105.7 (51+(0.8 commission))/49 meaning the wager price converted to a percent is 51.39% (518/1008).

Part 2 -- you can expect the game to end on 6 and 7 with rough frequency of 3.10% and 5.36%, respectively. This means the value of Atlanta -7.5, offered on Tradesports, is equivalent to (51.39%-3.10-5.36)=42.93%. Why subtract the numbers in entirety? Because the new line moves over these numbers in whole and shifts 6 and 7 directly from win column to loss column.

Part 3 -- therefore if you made buy offers of 42.0 or fewer (or market making offers of 42.5 or fewer) you will be paying a lower amount for the -7.5 bet than you would for the -5.5 bet. (ed Non-market making offer of 42 is equivalent to (42.0+(0.8 comm))/(58.0+(42.0+(0.8 comm)) or 42.46%, below the relative expected frequency of 42.93%.

I don't compute values for landing on 8, 9 & 10 so can't value the Atlanta -10.5 contract on TS, nor is the market for that contract particularly liquid.

[/ QUOTE ]

Thanks Syg.

First, what makes you think I'm talking about the Atlanta game? [img]/images/graemlins/smile.gif[/img]

Second, the fact that its not liquid is a good reason to pay attention to it -- ie. theres a greater likelihood of finding real value.

Third, are the tools available to determine how often a team wins by more than 10 when that team is favored by 5 or 6 or 7? I know you've done this analysis on the small lines (ie. how often does a team win and not cover when the spread is 2 or 3, etc.) and that is the basis of your playing so many moneylines. But if we had the closing lines and final scores of all NFL games, we should be able to run an analysis that addresses my specific question. No?

[JTrout]

Question-

Your #s indicate 8.46% of the time NFL games end on 6 or 7.

Would this % be significantly different on games favored by 5 or 5.5, compared to all games?

My thinking is that games closer to even have 2 strong chances of ending on 6 or 7 (favored team and underdog).

And games with a 5+ pt. favorite have 1 strong chance of ending on 6 or 7 (favored team only).


It may or may not be a significant difference.
And I don't have such info available to me.

[ QUOTE ]
Second, the fact that its not liquid is a good reason to pay attention to it -- ie. theres a greater likelihood of finding real value.

[/ QUOTE ]

Well yes, but not in this particular case because of time issues. The buy-sell spread for -10.5 is 28-40 while the spread for -7.5 is 41-43. Obviously there's no harm in putting up a buy offer somewhere in the middle and hoping a fish jumps the hook but you may be wasting your time while the favorable prices for -7.5 escape you.

[ QUOTE ]
Third, are the tools available to determine how often a team wins by more than 10 when that team is favored by 5 or 6 or 7? I know you've done this analysis on the small lines (ie. how often does a team win and not cover when the spread is 2 or 3, etc.) and that is the basis of your playing so many moneylines. But if we had the closing lines and final scores of all NFL games, we should be able to run an analysis that addresses my specific question. No?

[/ QUOTE ]

Not really, because the problem you run into is games at spread of 14 end on 10 with a lot more frequency than games with spread of Even, for example. My analysis is based on frequency of certain end results for games with proximate spreads to that result. 3.10% and 5.36% mean very little to a game with spread 17. I believe I may be addressing your point JTrout, which you're correct about.

[YoureToast]

I'm not interested for this purpose in knowing how often a game ends on a particular number; the query I'm seeking is one that would tell me this: In games in which a team covered a 6 pt spread, how often did that team win by at least 10 pts? 11 pts? 12 pts? Etc. If this question could be answered, I could then generate an approach to determining whether there is value in higher spreads.

So, maybe I'm not reading you correctly, but I don't think your response addresses that, does it?

Ah, I see where you're going with this. I don't have direct research on this type of thing. However, if it means anything, there are wagers called "challenge" bets or pleasers in which you give 6 points for 2+ games and get increased odds. Having done a little research on these recently, there is very little value to them. 6 points as it turns out is a lot of points to give in almost any situation.

the question seems to be somewhat related to point spreads vs. moneyline. i know he's asking something different i.e. point spread that is different from the line or a zero point spread (traditional money line).

i have looked for articles about point spread vs. money line and never found it... for college football in particular, i wonder if 1) the distribution of possible scores is normal??; 2) huge standard deviation of results meaning the tails can be quite large i.e. should a 6 point spread in NFL have same moneyline as 6 point spread in NCAA. my thought is no, but not sure if this is reflected in the market. might start this as its own thread.

[ThomasJoe]

[ QUOTE ]
"Hypothetically", Team A is favored against Team B by 5. I think Team A has a good chance of beating Team B by 20 points. Therefore, I am looking to get action on Team A at a point spread higher than 5. Suppose I could bet on Team A, give 5 points at -110. How do I go about determining at what number I should get to increase the line to 7? Related, if I can get the line at -10.5, how do I figure what fair value is?

I'm looking here not for some sort of general "feeling" type answer, but more of a mathematical approach.

Syg? Anyone?

[/ QUOTE ]

In theory it is all based on probability. Another way to look at it is you are playing the pointspread plus the moneyline. In football if team A has been estimated to be the better team and the oddsmaker would go through a process of determing a number which would determine his estimate of the number which will balance the action, statistically this will be a median and not an average. To get to the number 5 he would have probably tried either 7,6 or 3 because these are key numbers and they also tend to balance the action. The "cigar smokers" then play the game at each other until it is determined that 5 is the number. From there they just will let you buy off of five, but will charge more vig. At pinnacle as I write this the line on SD at NE week 4 is NE minus 5 plus 102, minus 5.5 plus 104 minus 6 plus 112, and minus 6.5 plus 119. On the other hand SD is +5 -110, plus 5.5 -117, plus 6 -126, plus 6.5 -134. At THE numer (5) the juice is only a 4 cent line. but if you move it to 6.5 the juice is a 7.5 cent line.

The rams and giant game is currently NYG minus 3 -113 and the rams are +3 +105 which is a 4 cent line. Since 3 is such a key number the line at 3.5 is NYG - 3.5 +106 and rams +3.5 -120, which is a 7 cent line.

IMO the oddsmakers know the number better than anyone. It is their job and they are as much and expert at what they do as Warren Buffet, The math faculty at MIT, or the floor traders of the NYSE. They do this every day and every year and if they show weakness they either get smarter or are replaced.

So always assume the number is good and that you could not have better information that they do. They put out a number that is a very good estimate of the median, and please know what the median is and the difference between a median and an average. (especially for totals.

In baseball If the NYY are minus 210 on the moneyline and Tampa Bay is plus 190. Then the no vig line is NYY minus 200, which says that NY is 2:1 favorite or that the probablility that they will win 2 of 3 games is 50/50. If the line would be NYY are 180/+170 then the probability if NYY winning 2 of 3 is less than 50/50. but the probability of them winning 175 and losing 100 games is 50/50.

Back to the SD at NE game the money line is NE-210 plus 190. So the estimated probability is that NE wins this game 2 of 3 times when played in week 4 at home which I think is reasonable.

Football scores can be very volitale as we all know, that is why they have teasors, which are always sucker bets because the variance of the predicted score and the final score differ signicantly. This does not mean that the line was not a good number. All it means is that one team got the breaks or had an unusually good or bad day.

The bottom line to me is that if you buy off of points you are usually making a mistake especially buying off of a key number. To me this is wrong because the price is wrong, for the same reason that playing teasors is wrong.

Hypothetically your opinion that team A will win by 20 and not by 5 would mean that you know more about making a line than the experts and if this was so you would be having fun here on 2+2 but would be betting sports full time. All your really doing is having fun.

So theoretically you are never getting the right price unless you are betting with a friend with no vig. If you really want to know how to estimate the moneyline associated with pointspreads you can see the differentials at pinnacle if you have an account. If you don't get one ( I only play there I have no interest in them. But the main guideline is to move the moneyline in relatively small increments for non key numbers and when you move it on key numbers a reasonable amount as my example above demonstrates.

I may as well elaborate some more on this subject hoping to generate some discussion.

If I think a game is somewhat volitale (usually a total) I will look to take the plus and lay the extra juice as the sportbook such as pinnacle is stuck on putting a plus to a number that has a variance. Example SD at NE total is 47.5 under minus 102 over minus 108. I say here why not take under 47 plus 105. I think this play makes more sense. It makes me into a bookie charging them juice as I do not think that there is any difference from 47.5 and 47.

Hope this helps, gl.