Private Members Service - Paul Dyson

Paul Dyson works as a Chartered Statistician and has been involved for nearly 20 years in the computer modelling of customer data for business clients including several blue chip companies.

Over recent years he has turned his modelling and forecasting expertise to the construction of horse ratings.

Paul’s technical expertise and Dave’s knowledge of sprint races have been combined to produce a set of ratings for all sprint handicaps since the start of the flat season in 2003. These have been provided to Drawn2Win members as a part of Dave’s daily message.

Over the last 12 months the models have been developed and improved and we have now reached the stage where extensive research has identified a number of profitable ways in which the ratings can be used both for backing and laying horses.


Ratings Bets

I have been developing ratings for sprint handicaps (5f / 6f / 7f) since 2003 with Dave Renham's help. These ratings are based on a number of factors (including of course, draw bias) and have been developed using data covering the last 4 years.

Behind the ratings is a complicated mathematical modeling procedure exact details of which I obviously wish to keep close to my chest.

Members of Drawn2win have been receiving the raw ratings for some time now as part of the ongoing research aspect of the site.

Having monitored these ratings over the last year, and after analysing the historical data, we have identified what we believe are a number of profitable ways of selecting horses from the ratings to back and lay.

The ratings are designed so that the average is 100 and the range is roughly 60 to 140. The ratings reflect a horse's chance of winning - the higher the rating the better the chance - and they are a combination of ability, form, race suitability, trainer/jockey form and draw bias.

As an example, here is an analysis of the ratings for 5f handicap sprints (on grass) over the last 4 years:

Strike Rate

You can see that most horses have a rating in the 90-109 range, but you can also see that as the rating increases, the likelihood of winning also increases.

In this way the ratings can be used as another tool in the analysis of a race helping to eliminate those horses with a very low rating and identify whether one horse stands out or whether a number of horses have a good chance.

However we also wanted the ratings to identify bets for us that would be profitable over the season. One approach is to back the top rated horse in each race.

For 6f handicaps, backing the top rated over the last 4 years would have produced 157 winners from 773 bets (strike rate of 20%) and a profit of 96pts (12% of total staked). However this is a lot of bets with a relatively low profit rate and we felt that restricting the bets further should produce better returns.

As an example, only backing the top rated horse when 10+pts clear of second rated produced 35 winners from 90 bets in 6f handicaps over the last 4 years. This generated a profit of 54pts or 60% of total staked.

From analyzing past results, we have found 3 types of bets:

1. Win bets - horses to back with a 1pt stake
2. Reverse forecasts - pairs of horses to back in a 1/2pt reverse forecast
3. Lays - horses to lay

The first two are straight forward but the lays need a little more explaining. Initially our analyses showed that laying horses more than 10pts behind the top rated horse was highly profitable. Even if we added a 20% premium to SP (when you lay horses on exchanges such as Betfair you usually need to offer more than the SP) the profits were still high.

However, monitoring these horses showed that it was very difficult to lay long-odds horses within 20% of SP. For instance horses at SP 20/1 often have to be laid at 30/1 (+50%). Longer priced horses attract an even higher premium sometimes double the SP. This could seriously affect the profitability of laying if one or two outsiders won in a season.

We therefore decided to restrict the laying approach to horses at 10/1 or shorter since we felt it would be possible to lay these up to 20% above SP (we have subsequently monitored the exchanges and found this to be generally the case). Hopefully most can be laid well within the 20% premium.

Since we will not know the SP of horses when the ratings message is sent out in the morning, we will list all possible horses in a race and suggest that any are laid if possible at 13.0 or lower on the exchanges (this is 10/1 + 20% = 12/1 plus the extra point for decimal odds).

We will be looking at different staking plans including - 1pt lay per horse and 6pt risk per horse. For instance, laying a 9/1 horse at 10/1 on the exchanges would produce:
1pt lay per horse: 1pt profit if it loses, 10pt loss if it wins
6pt risk per horse: 0.6pt profit if it loses, 6pt loss if it wins.

The most profitable staking plan will depend on the SP of the losing lays - if most are shorter prices 1pt per lay will make more profit, otherwise the constant risk approach will be best.

Ratings Bet Update 24 Sept 2004

The following is a summary of the results since the ratings bets went public to service members in April 2004.

Win bets – top rated horses 10 or more points ahead of 2nd rated are recommended as win bets. This produces a decent profit rate but only a few bets – roughly 2 per week. Current profit level to 1pt stake is 5.5pts or 14% of total staked.

Reverse Forecasts – when the 2nd rated horse is 5 or more points ahead of 3rd rated we recommend a half-point reverse forecast. This is more of a speculative bet with the opportunity for an occasional big win. Historical analysis suggests that profits should be made, but with only an expected 10% strike rate patience is needed. There have been 9 winners from 95 bets over the 2004 flat season but we are currently showing a loss of 9.9pts.

Lays – analysis shows with the ratings that it is difficult for a horse to overcome a 10pt gap (i.e. beat a horse rated 10 or more pts higher). Lays are based around this rule and any horse 10 or more pts behind the top rated horse (and also outside the top 3 in the ratings) is selected as a possible lay. However we also apply a rule based on the SP – the horse must be 10/1 or shorter (this avoids profits being wiped out by big priced winners). Unlike the first 2 types, lays produce a high volume of “bets” with interest in most rated races.

Lays are currently 110pts in profit !

(News Date 24-9-04)

All the above bets are available to members as part of the Drawn2Win members package.

Join Now


Dave Renham   |  Paul Dyson |  Join

horse racing uk
Copyright 2005

Forgotten your password?



Horse Racing Research &

Free Horse Racing System