Determining the true line with formulas...

[The13atman]

There are some people on here who talk about how they use formulas to calculate what the true line (or total) should be for a certain game and bet when there is a large enough disparity between "their" line and the line offered by a sportsbook.

What I'm wondering is how you came up with these formulas. Did you just take certain stat categories and assign them a certain weight/value and use trial and error to determine which of these formulas produced the best, most consistent results?

I'm not going to ask for the formulas or what the weight assigned to each category is, but what stat categories do you use when determining what a line should truly be at? Do you mainly just bet games in which there is large amount of square money (NYY, BOS, CHC) or do you often find incorrect lines in games where there is little public interest, such as CIN vs PIT? In those games especially, what makes you think your lines are better than the books? What do you see that they don't?

Also, have these formulas or systems or whatever you may call them proven to be consistent winners over many seasons?

Thanks, I find this sort of thing very interesting and impressive for those who are able to do it.

[wadea]

I've just recently started with this type of analysis and must say that I've been fairly successful. My approach has been particularly good at making good O/U lines. I use a formula to estimate an average number of runs scored by the home team and away team. I figure it twice, figuring expected runs scored and expected runs allowed for each team. Here is a list of the stats I plug in:

Runs/game in home games
Runs/game in away games
Ballpark factor (relative runs/game from ESPN.com)
Pitchers' season ERAs
Pitchers' home/road ERAs
Pitchers' last 3 starts ERAs
Teams bullpens' collective ERAs

Once I determine the number of runs I expect each team to score (offense and allowed on defense for each team), then I do the following. Using a poisson distribution function, I determine the probability of each team scoring each number of runs. It follows, then, that I can determine probability that the sum of their runs is greater than the O/U and the probability that teamA scores more runs that teamB. I make a line for teamA vs. teamB defense and then for offense. I average them (I'm still thinking about whether they should be equal weight) and that's my line for the game (or O/U).

To my mind, the poisson part of the calculation is not debatable. It is what it is, and it's correct for this purpose. The critical thing is to plug the right number into the distribution as the mean. That's where I figure to make or lose the money.

At the end of it, I analyze it subjectively to account for injuries or other oddities. I only bet the relatively big edges and I talk myself out of bets if I can. If I can't talk myself out of it, I bet it. In the short term, it's held its own for picking winners (and it's been improving with some recent tweaking). The real news is that it's been extraordinary at picking O/U. Since the most recent changes to the formula, it's like 16-8 over about 60 games (most games were passed on, of course).

Alright, I could type all night about this stuff but I've gotta start paying attention to my poker table in the other window. As always, I'd appreciate any feedback.

Bet the river,
-w.a.

[DougOzzzz]

[ QUOTE ]

To my mind, the poisson part of the calculation is not debatable. It is what it is, and it's correct for this purpose. The critical thing is to plug the right number into the distribution as the mean. That's where I figure to make or lose the money.


[/ QUOTE ]

Runs are not scored according to a Poisson distribution in baseball. A Poisson distribution should yield WAY too few 0 and 1 run games, and WAY too few 10+ run games. Runs often come in bunches. Over 70% of innings no runs score. But the ratio of 0 run to 1 run innings is much greater than the ratio of 1 run to 2 run innings, or 2 run to 3 run innings, etc.

This is why the O/U will always be set at a number that is LOWER than the mean expected runs scored. A poisson distribution would not show this. Either your means are much too low, or you're betting the over in every game.

[mrbaseball]

[ QUOTE ]
Did you just take certain stat categories and assign them a certain weight/value and use trial and error to determine which of these formulas produced the best, most consistent results?


[/ QUOTE ]

Sort of? I just started using the formula approach last season. But I had been betting/capping for a while before that so basically I took all the stuff I was already looking at and plugged it into a formula. I constantly tinker with it but the meat and bones are the same things I have been analizing for years. I started using the formulas because it gives me a great starting point and a reason to either ignore or fundamentally break down a game.

[ QUOTE ]
what stat categories do you use when determining what a line should truly be at?

[/ QUOTE ]

Hopefully the right ones [img]/images/graemlins/smile.gif[/img] The 4 main catagories of my formula are starting pitching, offense, bullpen and team performance overall (ie record). I have a few subcategories within these main categories. My lines are typically real close to what the man has so I'm guessing I'm on the right track.

[ QUOTE ]
do you often find incorrect lines in games where there is little public interest

[/ QUOTE ]

Little public interest? You mean like outside the AL East right? Most skewed lines involve the Yankees [img]/images/graemlins/smile.gif[/img] And yeah most skewed lines involve the square money and typically involve overpriced names (big name pitchers) and teams. But sometimes undervalued guys are ripe too and they are generally on the more obscure teams. The Brewers for example are one of baseballs best kept secrets. Yost and Maddux are doing a marvelous job and despite being a tad under .500 they are plus to the moneyline so far this season. St Louis dominates the National League but against the moneyline they are only up 8.2 units while the Nationals are up 15.2 units to the moneyline. Texas is 48-48 but plus 8.75 units to the moneyline and Minnesota. is 53-44 yet down a unit to the moneyline. The worst moneyline performed so far? Of course the Yankees at -15.35. There are opportunities everywhere but the most aggregious ones typically involve big name teams and big name pitching talent. That said I see a small edge on Colorado vs Pittsburgh today for a small play.

[ QUOTE ]
have these formulas or systems or whatever you may call them proven to be consistent winners over many seasons?


[/ QUOTE ]

Like I said I just really started the actual formulas last season but since they are just a mathematical representation of what I was already doing I would say yes as I have been a consistent winner over the past 7 or 8 years.