You seem to insert you personal judgment often. Have you measured if your personal expertise improves the model? If you just adhered blindly to the model would it win or lose?
I've run back tests in the past, and one could tell the program to do various things blind, or one could of course retrospectively do so since line histories are stored. But no, I'm not explicitly tracking the blind method. The sample size required to know the difference between the two, given they mostly bet the same stuff and the differences are games on the margin, would be quite large, and multiple-hypothesis testing leads to problems.
Putting here for the record that if I did do blind wagering, I would do something like: Minimum double-vig edge on MLs, triple-vig edge on totals, always bet opener (or at least a fixed time overnight). If you wanted to, you could codify some of the adjustments (e.g. if you're offered 2 wagers on the same game). Also, you'd ideally back off games with unrecognized pitchers and at least demand bigger edge, because sometimes they'll be good people coming off injury or just traded, etc.
If I was doing this seriously as a source of income, I'd be tracking it for sure anyway of course.
One other problem with simulated blind wagering is that a few bad lines or pieces of bad data that aren't real can make you think you're doing great when actually all your edge comes from things that aren't real, so you have to confirm that everything's legit. That's especially true when you're trying to bet openers.
You seem to insert you personal judgment often. Have you measured if your personal expertise improves the model? If you just adhered blindly to the model would it win or lose?
I've run back tests in the past, and one could tell the program to do various things blind, or one could of course retrospectively do so since line histories are stored. But no, I'm not explicitly tracking the blind method. The sample size required to know the difference between the two, given they mostly bet the same stuff and the differences are games on the margin, would be quite large, and multiple-hypothesis testing leads to problems.
Putting here for the record that if I did do blind wagering, I would do something like: Minimum double-vig edge on MLs, triple-vig edge on totals, always bet opener (or at least a fixed time overnight). If you wanted to, you could codify some of the adjustments (e.g. if you're offered 2 wagers on the same game). Also, you'd ideally back off games with unrecognized pitchers and at least demand bigger edge, because sometimes they'll be good people coming off injury or just traded, etc.
If I was doing this seriously as a source of income, I'd be tracking it for sure anyway of course.
One other problem with simulated blind wagering is that a few bad lines or pieces of bad data that aren't real can make you think you're doing great when actually all your edge comes from things that aren't real, so you have to confirm that everything's legit. That's especially true when you're trying to bet openers.