[Bet Shooter]

Let's say you have a documented edge of 57%, with enough samples on a theory for wagering. Does anyone adjust their wagering if the actual results start to drift too far from that 57%? If the 57% is a rock solid #, would it pay off in the long run to wager "on paper" if the real % started to approach 60%? Or at what % away from the true # would it start to make sense to do this? Or not at all, and why?

[Ganchrow]

Let's say you have a documented edge of 57%, with enough samples on a theory for wagering. Does anyone adjust their wagering if the actual results start to drift too far from that 57%? If the 57% is a rock solid #, would it pay off in the long run to wager "on paper" if the real % started to approach 60%? Or at what % away from the true # would it start to make sense to do this? Or not at all, and why?

There's no
magic number here.

When dealing with a single historically motivated model, you need to take all out-of-sample data in aggregate, possibly discounting past data as predetermined.