# Creating Odds Lines – Episode 4 Power Ratings

If you haven’t had a chance to read past articles about creating odds line then you can find out…

Today we are going to be moving into episode 4 and how to use power ratings. Although this isn’t strictly odds line creation, it’s a slightly different approach, it very much belongs to the same area of your betting.

What is a power rating?

A power rating is a single rating that attempts to measure everything there is to know about a horse with a very high level of accuracy. It is essentially a combination of ratings into a single number that can then be used to make profits from your betting.

The information included in the power rating can be specific to certain race conditions. In other words, we can have multiple power ratings which we use in different situations. The benefit of this is that we only need to include the information that is relevant to the specific race conditions and we don’t end up with data that is less relevant.

Ultimately any kind of power rating or odds line is a combination of information. I have already looked at methods previously to create these single power ratings or probabilities, and we will look at more methods in the future, but today the focus is on how do we turn a power rating into an effective odds line and then use it to bet with.

I am going to assume that we’ve already created our power rating and we have a single number for each runner in the race. What do we do with it now. Do we…

- Always bet the top rated
- Always bet the top two rated
- Only bet the top rated if it is 10 points higher than the second top rated
- Look for races where there is a big difference between the ratings for different runners

These are just a few possible options. In fact, there are almost limitless ways that you can use your power rating. What we need to do is to narrow it down to just one.

This isn’t to say that the other methods aren’t effective. They absolutely are and is something you should investigate. However, today we are focusing on odds lines and so the method that we are about to use is going to give you an odds line that is far more accurate than just using your power ratings.

To do this we need to turn our power rating into a probability. I am going to use the example in Episode 1 to show you how to do this. You take your rating for each horse:

Horse 1 = 56

Horse 2 = 75

Horse 3 =23

Horse 4 = 37

If we add all these ratings together we get 191. We now divide each horses rating by this number to get…

Horse 1 = 0.29

Horse 2 = 0.39

Horse 3 = 0.12

Horse 4 = 0.19

These numbers will always be less than 1 and they are the probability of each horse winning based on your rating. This means that horse 1 has a 29% chance of winning, horse 2 has a 39% chance of winning etc… To change these into odds we divide 1 by each of the horses ratings to get…

Horse 1 = 3.45

Horse 2 = 2.56

Horse 3 = 8.33

Horse 4 = 5.26

That’s pretty easy, all you need is a calculator and you can have it done in a couple of minutes. But how do we make this more accurate?

We want to do this in a way that is effective and powerful, but also as quick as possible.

So…

We use the live market!

Betfair is the best free market for sports bettors, it shows true market prices participants in an event, in our case horses. This doesn’t mean that it always gets them right. But, what it can do is to highlight where we may have gone right and wrong.

What we are going to do is to use this source of information to adjust our own probabilities created from our power ratings. And we do it with knowledge we have already learned in previous episodes. We use a weight to determine the level of importance of the Betfair probabilities and our own probabilities.

To turn Betfair odds into a probability you simply divide 1/Odds. If the odds were 12 then you would do 1/12 = 0.083. Easy isn’t it 🙂

Let’s take our four runners again but this time let’s put Betfairs odds next them…

Our Odds | Betfair Odds | |

Horse 1 | 3.45 | 5.75 |

Horse 2 | 2.56 | 2.13 |

Horse 3 | 8.33 | 12.5 |

Horse 4 | 5.26 | 4.7 |

We can see that we disagree with the Betfair market on the runners chances in this race. But we want to make as sure as possible that the horse we bet on is the one that we are right about. So, let’s convert these odds into probabilities.

Our Odds | Betfair Odds | |

Horse 1 | 0.29 | 0.17 |

Horse 2 | 0.39 | 0.47 |

Horse 3 | 0.12 | 0.08 |

Horse 4 | 0.19 | 0.21 |

Now is the stage where we need to decide whether we believe our own probabilities are better than the Betfair market. You can assume that they aren’t. It is very difficult to be more accurate than the Betfair market on your own, but by using the Betfair market as part of your odds lines then you can be in a much better position than the average punter.

For example let’s say that we want to use the full weight of Betfair’s probabilities but we don’t want to increase their importance. However we are going to assign just 75% of the value of our probabilities to the final calculation. This means that we multiple our probabilities by 0.75. We now get…

Our Odds | Betfair Odds | Total | |

Horse 1 | 0.22 | 0.17 | 0.39 |

Horse 2 | 0.29 | 0.47 | 0.76 |

Horse 3 | 0.09 | 0.08 | 0.17 |

Horse 4 | 0.14 | 0.21 | 0.36 |

Here you can see that Our Odds probabilties have reduced slightly because we have multiplied them by 0.75 or 75%. We then add together both Our Odds and Betfair Odds probabilities to get the Total figure. We need to make one more adjustment before we can convert them back to odds however. We need to normalise the Total. If you have read my other articles then you will already be familiar with this process.

To do this we add together all the totals for all runners in this race and then divide each horses Total by this number. For example…

0.39 + 0.76 + 0.17 + 0.36 = 1.68

We divide each horses Total by 1.68 to get…

Our Final Prob. | Our Final Odds | Betfair Odds | |

Horse 1 | 0.23 | 4.35 | 5.75 |

Horse 2 | 0.45 | 2.20 | 2.13 |

Horse 3 | 0.10 | 10.00 | 12.5 |

Horse 4 | 0.21 | 4.76 | 4.7 |

You can see above that Our Final Probability has changed slightly. We then convert it back to odds using the 1/probability to get our much more accurate odds line.

We can see now that Horse 1 and Horse 3 are underbet slightly while Horse 2 and Horse 4 are slightly over bet. We only want to be betting on runners that are significantly under bet to make sure that we have an edge you may choose to only bet runners that have odds 50% higher than we think all the way down to 20% or less. In this example I am going to use 20%. This means that for us to bet on these runners they would need minimum odds of…

Horse 1 – 5.15

Horse 2 – 2.64

Horse 3 – 11.86

Horse 4 – 5.68

We can see straight away that Horse 1 and 3 would be bets for us in this race. Not only that but by using our probabilities you know that by betting on both of these runners you have a 33% chance of winning. Now you may decide that you don’t want to bet in races where there is an overbet short odds favourite because this horse, although not a profitable bet, has a very high chance of winning. But that is to do with wagering strategy and is for another article.

You have totally lost me here. Previously You have stated to normalise a race that you take a horse( highest rating – horse Minimum rating divided by maximum rating in race – minimum rating in race) When highest number is top performing and to reverse the formula for when the smallest rating is top performing.Also I cannot arrive at the same figures as you when dividing by 1.68. for example 0.39/1.68 = 0.23. Or am I just stupid and have missed something

Thank you for the questions James. There is more than one way to normalise the ratings. The method you are thinking of is when we have multiple ratings for each runner and we need to normalise each horses ratings so we can combine them, whereas above we are looking to normalise one final rating, the one after we have combined our power ratings with the Betfair odds. This needs a different approach like above.

You are absolutely right with 0.23 and you can see this above under Our Final Prob. This is the final probability we achieved for the runner after normalisation. We then turn this back into odds by dividing 1 by this number. I did the calculations in our database and this provided some rounding errors as the database works to 10 decimal places. I have corrected this now.

You have used the same formula for converting the power ratings into probabilities ie (total of ratings / individual runner ratings) and the same (total of ratings / individual ratings) for normalization why is this the same ?

This is because we are mathematically doing the same thing. After combining our probabilities what we have is in effect another rating which we need to adjust so that all figures add up to 1.

Thought! The whole rating process involves every runner, if you know which runners you have classed as no hopers would it be more accurate to only rate the runners that are left excluding the no hopers.

A very good thought. It is not necessarily more accurate but it is certainly a lot faster. You can assign 20% of the book to the no-hopers and then calculate odds for the rest based on them receiving 80% between them.

It may happen that the odds you get are to close to each other to be used efficiently:

what should be done in this case?

When this happens you want to raise each probability to the power of a number greater than one. The higher the number the more it will spread out the values. You will need to find the right value for your rating. Once you’ve raised each probability to the same power, you will need to normalise them again.

ok, thank you Michael, I’ ll give it a try.

Hi Michael,

Excellent articles on creating odds lines from ratings, i was wondering if you know of any software available that can do all the calculations after you have entered your original rating and set your parameters eg, power of no. and your probability % confidence like the 0.75 in the example above but also have a live constant update of betfair prices that constantly changes and updates your odds line as the market moves one way or another with maybe a traffic light color system that makes it easy to hit the button to bet to a predetermined stake just on the off eg, red = negative advantage , orange = slightly negative , green = positive advantage . any help much appreciated . just found your site when searching for odds line creation the only one with a clear easy step by step explanation . thanks again .

Colin.

Thanks for your comment Colin. At the moment there is not any commercially available software I know of which can do this. However it is on our list to create something similar to what you’re looking for in the future 🙂

many thanks,

colin

Figuring out the weights for Betfair and model probs is obviously the key. One approach I took is to use a Genetic Algorithm to process the rating probs and the Betfair probs however if left to its own devices the cost function of the Genetic algo will get swamped by the Betfair odds, in other words it will look at your probs and the Betfair probs and say, hey Betfair is far better I will give it a large weight which means you are heading towards agreeing with Betfair. It will do this because its designed to find winners and not profit however if you plug in your own cost function to say calculate variable stake profit it will optimize on this criteria and not allow Betfair to swamp matters

Really interesting Mark. I saw you were working with genetic algorithms. Thank you for sharing this insight.