So how do you rate a handicapper?
Who is better?
Handicapper 1 who goes 8-2 (80%) over a week of picks
Handicapper 2 who goes 60-40 (60%) on 100 plays
Handicapper 3 who goes 550-450 (55%) on 1000 plays?
Or the Handicapper who bets underdogs and wins 156.93 units on 129 units wagered across 89 plays? (Like I did in April 2014)
If you are rating a handicapper who often bets underdogs like I do you have to pay attention to units won, units wagered, wager size and number of plays. I will explain further in section 2 below.
Section 1 – Rating handicappers who bet against the spread – Examples 1-3
If you are ranking handicapper who bet against the spread you can use math – specifically the standard deviation and z-score.
If you think about it when you are rating handicappers who pick against the spread you are trying to figure out how rare their results are. Spread bets can be compared to flipping a coin because the results are basically binary (not taking into account pushes)
So what is more unexpected? Flipping a coin and landing on tails 8 out of 10 times or 60 out of 100 or 550 out of 1000?
The first thing we need to figure out is the expected result, I will use 3 examples to illustrate this.
For Handicapper 1 the expected result is 5 out of 10
For Handicapper 2 the expected result is 50 out of 100
For Handicapper 3 the expected result is 500 out of 1000
Next we need to determine the standard deviation how far off the expected result can be explained by random events.
For Handicapper 1 we have 10 events
Standard deviation is probability (in this case 50/50 or.5) * square root of # of plays (in this case 10 plays)
Std = 0.5 * sqrt (10)
= 0.5 * 3.16
= 1.58
Therefore we expect a variance of about 1.58 per 10 events
For Handicapper 2 we can do the same calculation
Std = 0.5 * sqrt (100)
= 0.5 * 10
= 5
Therefore we expect a variance of about 5 per 100 events
For Handicapper sports handicapper picks 3 we can do the same calculation
Std = 0.5 * sqrt (1000)
= 0.5 * 31.62
= 15.81
For handicapper 3 we expect a variance of 15.81 per 1000 events
So that’s great now what?
Well the next step is to determine the z-score. The z-score is a description of how unlikely an event is with handicappers.
The formula for z-score is observed wins less expected wins divided by the standard deviation
Z = [(observed wins) – (expected wins)] / std
For Handicapper 1
Z = (8-5)/1.58
=1.90
For Handicapper 2
Z =(60-50)/5
=2
For Handicapper 3
Z =(550-500)/15.81
=50/15.81
=3.16
The key metrics to note are the normal distribution.
68% of all handicappers will achieve a z-score of 1 or less
95% of all handicappers will be below a z-score of 2. There for a z-score of 2 puts a handicapper in the to 5% of all handicappers
A z-score of 3 is extremely unlikely. Puts a handicapper in the top 1% of all handicappers see the below image to review the normal distribution with z-score across the bottom.
As a rule of thumb when selecting a handicapper to follow you should consider both the number of plays they provide in a sample and the number of units wagered and units won – see below for a further description of how to factor in different size unit plays.