Quantifying Edge: for the Kelly fellas

[matskralc]

If this has been asked before, please excuse me. The search function here sucks.

Fairly simple question: how does one quantify his edge on a point spread or game total? If, for example, I cap tonight's play in game and I've got Mt St Mary's covering easily with an 11-point win, how do I figure my edge against the spread of 7.5 (dog -103/fav +102) for the purposes of determining a Kelly stake? Similarly, if I've got the total at 124 against a line of 126 (o+104/u-110), how do I quantify that edge?

I thought I remembered this being asked somewhere, and while the answer wasn't simply the percentage my line was off the book's there wasn't a clear answer given. If there IS no clear answer, that's OK, too.

[donjuan]

I'd say you should start by re-examining how you got a number that is 4 points off the current market number.

As for the actual question though, if you have an NCAA basketball game at -5 and you can get -4 at -110, I'd go ahead and use Ganchrow's half point calculator to determine that your edge is 3.53%, assuming no margin of error. I'd use a margin of error, though. Basically you want to look at the push percentages of the numbers you are gaining from fair value.

[matskralc]

I'd say you should start by re-examining how you got a number that is 4 points off the current market number.

On this particular game, I would guess it's because Coppin State is the "hot" team after an awful start while Mt St Mary's has been fairly solid all season. I don't normally find trends/streaks like that to be particularly useful, and would prefer to use a team's entire body of work unless I have some compelling reason not to (Coppin St didn't make a big trade and acquire a new center at the trade deadline or anything). Their terrible 4-19 start counts, too. I didn't actually take MSM to cover, though, just made a straight moneyline bet (where it's fairly easy to determine my edge )

As for the actual question though, if you have an NCAA basketball game at -5 and you can get -4 at -110, I'd go ahead and use Ganchrow's half point calculator to determine that your edge is 3.53%, assuming no margin of error. I'd use a margin of error, though. Basically you want to look at the push percentages of the numbers you are gaining from fair value.

Yeah, the HPC is useful for a point or two one way or the other, but if I really believe the line is off 3.5, I'd like to be able to figure out my edge.

Also, uh, something I can plug into my spreadsheet would make the process more efficient, too.

[Data]

This should help:

http://www.sbrforum.com/Betting+Tools/Spread+ML+Converter.aspx

[donjuan]

Yeah, the HPC is useful for a point or two one way or the other, but if I really believe the line is off 3.5, I'd like to be able to figure out my edge.

You really just aren't going to consistently find lines off by 3.5 points. But if you really want to do it, just use the same methodology Ganchrow and Stanford Wong use, but instead of finding the push percentages for 7 when the spread is between 5 and 9, find the push percentages for 8, 9, 10, 11 when the spread is between 5 and 9.