Handicapping Handicappers by Points

**Dark Horse** · 02-14-09, 07:34 PM

No, but you could use it for teams, to see how they perform ATS and under what conditions. For instance, strong and motivated teams will often beat their opponent by more than the spread for the entire game in just the 1Q.

**B1GER1C828** · 02-14-09, 08:25 PM

y the heck would it matter?? if a team wins by .5, its a win, if it loses by .5 its a loss, doesnt matter how much u win by

**yisman** · 02-15-09, 02:27 AM

action points grading. It's occasionally used for contests, but I'm not sure your suggestion is a good way of using it.

**gamble4heisman** · 02-15-09, 08:44 AM

Don't quit on it too soon, i actually think this may have some merit.... not sure exactly how, or how much weight to give it but easy winners are more valuable than close winners when i'm throwing my darts....

**Arilou** · 02-15-09, 11:11 AM

I think this proposal is quite good. We know it works for teams in many sports far better than their W/L records. You'll have to proceed carefully to control for different cappers taking different types of lines (since the mean and the median are not equal for certain spreads in some sports) but I have little doubt, despite never having tested the theory, that you would get a much better predictor of future capping success by taking margin of victory against the spread into account than by using simple win/loss. However, I also would caution against ignoring the wins and losses themselves as there are various distributional reasons only one of which I've mentioned that can cause the expected win % and the expected mean to not correspond properly.

**Justin7** · 02-15-09, 03:11 PM

Originally posted by TotallyTilt

Would it make more sense to track a handicapper by points instead of win/loss? In other words, assume they have the Celtics -5 and the Celtics win by 12. This would be a +7 for the handicappers. If they have the Bulls +4 and the Bulls lose by 5, this would be a minus 1. You could then keep a running total. Would this be any more accurate than a simple win/loss record?

This is called "action point" grading. This is a much better predictor long-term than plain W/L record once you get to a big sample size. The problem is that short term, your volatility is higher (i.e. +17 for that play instead of +1/-1). But with big sample sizes - like 1000, action point grading is much better.

**Dark Horse** · 02-17-09, 02:39 AM

I haven't kept action point records. What would be the point unless you could sell points to the book?

**Dark Horse** · 02-19-09, 01:35 AM

On the other hand, if action point grading (for betting systems) is more insightful than W/L record, that could be useful to determine bet size.

How would one determine bet size by this method, rather than by winning expectation? For instance, if a method beats the spread by an average of 7 pts.