While the approach that accounts for the moneylines and the totals is certainly the right one, it seems that the Ryan's method produces results that are way off. I think that the easiest method is getting a large sample of lines and making a cheat-sheet out of it. I have my reasons to believe that the linesmakers have a good handle on that conversion.
Comment
Ganchrow
SBR Hall of Famer
08-28-05
5011
#4
Originally posted by Data
While the approach that accounts for the moneylines and the totals is certainly the right one, it seems that the Ryan's method produces results that are way off. I think that the easiest method is getting a large sample of lines and making a cheat-sheet out of it. I have my reasons to believe that the linesmakers have a good handle on that conversion.
Another method commonly used is to simply go through the historical record and come up with a conversion chart that maps ML+Total to RL. With a much larger sample size this would be great, but the real problem with this method is that in reality one runs into some nasty data artifacts and so also needs to employ some sort of smoothing mechanism.
Ryan's method, while certainly not perfect, does demonstrate one such smoothing mechanism that could be used in conjunction with the brute force above. As I said, it makes for a good place to start.
While books such as Pinnacle do have a good handle on these conversions, their mappings are still a good deal less than perfect (a decent part of which may well be the result of less than perfectly balanced order flow).