The Group Overlay Method

by Dick Mitchell

How would you like a method that combines low risk with generous returns? Unfortunately, it takes a little work, but it's worth it. If you don't take advantage of this very solid method, you might be leaving money on the table.

The first skill you'll need is the ability to convert odds to probabilities. It's actually pretty simple. There's a relationship between odds and probabilities. In fact, they're really the same things, expressed in different ways. Even-money and 50% probability are synonymous. A 2-1 horse has a 33 1/3% chance of winning the race. A 3-1 shot has a 25% chance of winning. The relationship between probability and odds can be stated as follows:

Probability = 1/("odds-1" + 1)

If you know the "odds-1" you can calculate the probability by adding 1 and taking the reciprocal. Four-to-one (4-1) can be expressed as 1/(4+1), or 1/5, which is 20%. Whether you refer to a horse as a 4-1 shot or as a horse having a win probability of 20%, you're saying the same thing. To convert any odds to "odds-1," you simply divide the second number into the first. For example, 7-2 is 3½-1. You divide the 2 into 7 to arrive at the result. Try 8-5. Did you get 1.6-1? What about 4-5? Did you get .8-1? If so, you got it.

If you know the probability and want to convert to odds, the relationship is as follows:

"Odds-1" = (1/Probability) - 1

If a horse has a 25% chance to win, its fair odds are 3-1, because 25% = .25 and 1/.25 = 4 and 4-1 = 3. If a horse has a 10% chance to win, its fair odds are 9-1.

The following odds percentage table should help in quickly converting probability to odds and vice-versa.

Odds Percentage Table
1-10 ......... 90.91    4-1 .......... 20.00
1-5 .......... 83.33    9-2 .......... 18.19
2-5 .......... 71.42    5-1 .......... 16.67
1-2 .......... 66.67    6-1 .......... 14.29
3-5 .......... 62.50    7-1 .......... 12.50
4-5 .......... 55.56    8-1 .......... 11.11
1-1 .......... 50.00    9-1 .......... 10.00
6-5 .......... 45.45    10-1 ........... 9.09
7-5 .......... 41.67    11-1 ........... 8.33
3-2 .......... 40.00    12-1 ........... 7.69
8-5 .......... 38.46    15-1 ........... 6.25
9-5 .......... 35.71    20-1 ........... 4.76
2-1 .......... 33.33    25-1 ........... 3.85
5-2 .......... 28.57    30-1 ........... 3.23
3-1 .......... 25.00    50-1 ........... 1.96
7-2 .......... 22.22    99-1 ........... 1.00

To make a "betting line," simply estimate the chances of each of your contenders. Don't worry about this now, it won't be really necessary, as you'll see in a few moments. Later, I'll give you some "group statistics" that'll blow your mind.

Here are the rules for applying the Group Overlay Method:

1. Take your betting line for each of your contenders in a race and convert it to probability. Add up the probabilities. This sum is your Group Estimate. The Group Estimate must equal .67 or more. If it doesn't, forget about this method.

2. Take the public's betting line (actual toteboard odds) for each of those same contenders and convert it to probability. Add up the probabilities. This sum is the Public's Estimate.

3. Compare the Public's Estimate to your Group Estimate. If the Public's Estimate is lower than your Group Estimate by 10% or more, you are in a positive-expectation situation and can apply the Group Overlay Method to the race. If the Public's Estimate is not lower than your Group Estimate by 10% or more, skip the Group Overlay Method for this race.

Take the following example:

Your Group Estimate of the contenders' win probabilities is .33 + .25 + .20, or .78, which is more than .67.

The Public's Estimate of the contenders' win probabilities is .36 + .17 + .14, or .67. The Public's Estimate is lower than your Group Estimate by .11, which is more than 10%. Hence you can apply the Group Overlay Method to this race. Sticking with this example, you determine your bet size on each horse using the following steps:

4. Subtract the Public's Estimate from your Group Estimate. The result is Total Discrepancy. In this case, the Total Discrepancy is .78 minus .67, or .11.

5. Divide Total Discrepancy by the following: (1 - Public's Estimate). The result is Optimum Percentage. In this case, Optimum Percentage is .11 divided by .33, or .33.

6. Divide Optimum Percentage by the following: (Public's Estimate) ´ (Public Odds + 1). Do this for each of your contenders. The result is each contender's Bet Size Percentage:

Horse A .33/((.67 ´ (1.8 + 1)) = .176
Horse B .33/((.67 ´ (5 + 1)) = .083
Horse C .33/((.67 ´ (6 + 1)) = .071

7. Divide each contender's Bet Size Percentage by Optimal Percentage. The result is the percentage of your total bet size for this race you should bet on each contender.

Horse A .176/.33 = .533
Horse B .252/.33 = .252
Horse C .071/.33 = .215

If our total bet in this race were $100, we'd bet $53 on Horse A, $25 on Horse B and $22 on Horse C.

What if you don't make a betting line on a race? Fear not. An advantage to the Group Overlay Method is that you need only estimate the group's chances and not be terribly concerned about a betting line for each of its members.

For example, say you've handicapped 1,000 races. In 720 of those races, the eventual winner can be found among your top three contenders. In other words, you have the winner among your top three contenders 72% of the time. Guess what your Group Estimate for your top three contenders in any given race is? You guessed it. It's 72%. And you don't need to divide that 72% among your top three contenders as you would in a betting line. All that remains is to compare the Public's Estimate with your Group Estimate of 72%. If the Public's Estimate of your top three contenders is lower by more than 10%, you have a race in which you can use the Group Overlay Method.

Let's try using the Group Overlay Method without a betting line. Let's suppose that the eventual winner can be found in our top three contenders (Horses A, B and C) 72% of the time. This means that our Group Estimate is 72%. Here is the public's line on the race:

Horse A 3-1
Horse B 4-1
Horse C 5-1

The Public's Estimate is .25 + .20 + .17, or .62. Our Group Estimate is .72. The difference is .10. Hence the Group Overlay Method can be applied to this race.

The Total Discrepancy is .10. The Optimum Percentage is .263. The Bet Size Percentages are:

Horse A .263/((.62 ´ (3+1)) = .106
Horse B .263/((.62 ´ (4+1)) = .085
Horse C .263/((.62 ´ (5+1)) = .071

The percentage you'd bet on each contender is:

Horse A .106/.263 = .403
Horse B .085/.263 = .323
Horse C .071/.263 = .270

If you bet $100 in this race, you'd divide it $40 on Horse A, $32 on Horse B and $27 on Horse C. Let's calculate the edge on this bet.

Let's see what happens if A wins and we bet $100. A pays 3-1, therefore we get back $160 for our $100 making a profit of 60%. If B wins we get back $160 - same thing. If C wins we get back $162 for a 62% profit. Our mathematical expectation is: E(x) = .72($60)-.28($100) = $15.20. That's an expectation of a $15.20 profit for every $100 that we bet. That's a 15% edge!

Please remember that most professional money managers on Wall Street would give an organ to consistently achieve a 15% return. In fact, in many cases, a vital organ!

Here are some stats compliments of Jim Cramer, a brilliant researcher and president of Handicapper’s Data Warehouse (HDW), that may help you apply this method:

1. 83.8% of all winners at 5 furlongs on the dirt were first, second or third at the second call!

2. 67.8% of all winners at 5 furlongs on the turf were first, second or third at the second call!

3. 77.1% of all winners at 5½ furlongs on the dirt were first, second or third at the second call!

4. 75.3% of all winners at 6 furlongs on the dirt were first, second or third at the second call!

5. 69.9% of all winners at 6½ furlongs on the dirt were first, second or third at the second call!

The above five scenarios offer a wonderful application of this powerful method. Simply project the second-call time for each horse in the race at the above distances on the above surfaces. Note the top three and check their odds against the public's estimate. If there's a 10% or more differential, bet all three in the appropriate proportions. You'll win most of your bets and make a profit that's almost obscene. For 6- and 6½-furlong sprints in Southern California, this method hasn't had a losing month in this decade. Unfortunately, for the shorter distances, many of these races are for babies. Hence, you get a bunch of first-time starters and can't project second-call times for all the contenders. Pity.

Another good use for this method is when you've boiled a race down to only three possible contenders. Rather than agonize, bet all three if the public gives you the appropriate discrepancy. In any case, it's nice to have a technique as powerful as this in your arsenal. See you on the short line.