The purpose of this blog is to introduce the concept of a System Methodology for determining the odds at which a runner becomes a potential value bet in any horse race, which I have christened the VALUE CALCULATOR.
Before I document how the VALUE CALCULATOR works and the results from my initial investigations, which have been collated during the major part of 2016, it is probably worthwhile if I explain what is meant by the term value and why it is important for serious punters to strive to always back their selections at value odds.
A Definition Of Value
In simple terms, value is achieved when a punter backs a selection at odds that are better than the probability of the horse winning the race.
At this stage, the authors of many of the books that I have read in respect of betting on horse racing tend to explain this definition using the example of the flip of a coin.
However, 2 horse races are thankfully rare in real life, so my preference is to use the analogy of the rolling of a normal 6 sided dice.
Value Betting Odds
If you can get odds of 11/2 on a roll of the dice then place that bet!
Providing that the dice are not loaded, the probability of each of the numbers 1 to 6 being the outcome on each individual roll of the dice is 100/6, or 16.67% which equates to odds of 5/1.
Under these circumstances, any punter who is able to place a bet at 5/1 on a specific number for every roll of the dice would expect to break even or achieve a position close to a stalemate over an extended number of dice rolls.
However, if the punter was able to place these bets at 11/2 (equivalent to a probability of 15.4% of success) would expect to make a small profit over time, as they should find the winner on approximately 100 occasions from 600 rolls of the dice, which would net a return of around 650 points for a total outlay of 600 pts, resulting in a profit of circa 50 pts.
Unfortunately for punters, bookmakers are successful profit-making operations that always build in a profit margin within the odds that they quote for every event. Therefore they probably offer only 9/2 (equivalent to a probability of 18.2% of success) for each possible outcome, so whilst the punter would still find the winner on approximately 100 occasions over the same period, the net return would be around 550 pts, resulting in an overall loss of 50 pts.
However, in real life, bookmakers are not infallible as their teams of risk managers and odds compilers are unable to wave a magic wand to produce accurate prices for every race and have to make educated guesses as to the probability of each horse winning the race.
Whilst in the long term these estimates are unerringly accurate when considered from an overall perspective, they regularly make mistakes which can be exploited by punters, who are able to identify when the odds for a runner are over-generous.
The Importance of Backing Value Selections
The above scenario clearly shows that for a punter to achieve a long term profit it will be necessary for them to frequently place wagers on horses at odds that are greater than the underlying probability of the runner winning the race.
In theory, this is an easy comment to make but is a difficult position to achieve in practice, based on the fact that only a small proportion (probably between 2% and 5% if recent anecdotal evidence can be believed) of regular punters are able to make a sustainable long term profit from betting on horse racing.
This situation has been blurred further by the prominence of media tipsters such as Pricewise, who have made the term ‘value wagers' synonymous with big priced selections in extremely competitive, bookmaker sponsored, big field handicaps. When in reality even short-priced favourites in uncompetitive maidens and stakes races can offer value when the odds on offer understate the probability of the horse winning the race.
Many of the books that I have read that attempt to provide punters with the tools required to join the exclusive long term winners club extol the virtues of calculating the probability for each runner winning the race.
From this, the punter can produce a personal odds list (tissue prices) that can be compared to the prices that are available from the bookmakers and exchanges, which in theory will highlight horses that can be backed at value odds.
Would the creation of a betting tissue help in finding winners?
However, I have yet to read a book that provides an adequate explanation of how the calculation can be achieved in an objective manner.
Most of the explanations recommend that the punter will be required to consider a multitude of factors, but fail to list the most likely criteria that should be included, whilst a few publications have provided a few suggestions in respect of the criteria, but have failed to explain a suitable method for turning this information into a probability figure for each runner in a race.
This in my opinion is about as useful as a cookbook that tells the reader the list of ingredients to produce a dish but fails to explain how the ingredients are to be mixed and most importantly both the oven setting and time required to cook the dish.
The stock advice tends to be that the Punter should experiment producing probability figures and tissue prices for each horse in a race for a period of time during which they should become more proficient with experience.
However this trial and error process must surely be riddled with subjectivity, as there are many opportunities for the punter to allot a higher probability figure for a runner, including the horse is strongly fancied by the media tipsters, or has won for the punter before, whilst the opposite problem may arise when the horse represents a trainer that is not rated by the punter, or is to be ridden by a jockey that the punter is wary of, to name but a few possible pitfalls.
In order to address these issues, I have developed the VALUE CALCULATOR which I will explain in the following section.
How The Value Calculator Works
The VALUE CALCULATOR allocates a separate numeric value for each horse in a race, in relation to a set of criteria which includes historic form and the suitability of the likely race conditions, which can then be used to calculate the probability of each runner winning the race.
In order to keep things simple at this stage of my investigations, I have used the criteria which are produced and assessed by the Postdata service in the Racing Post, as follows:
|1.Trainer Form (TF)|
|2. Suitability Of Going (G)|
|3. Race Distance (RD)|
|4. Course Form (CF)|
|5. Draw Bias Where Appropriate (DB)|
|6. Ability Of Horse (A)|
|7. Recent Form (RF)|
Postdata assigns a number of symbols to indicate whether each horse, or trainer is considered to be in form and the likely suitability of each of the above race conditions. Positive results for each criteria are shown as ticks, whilst likely negatives are indicated by a cross, with blanks, question marks, or dashes assigned where the criteria is not applicable, or there is insufficient data to assess whether the horse should be assigned a positive, or negative mark.
The number of ticks allocated for each Postdata criteria is an indication as to the strength of form/ability, or suitability of the race conditions.At this stage of my investigation, I have chosen to ignore all symbols other than ticks, which I have equated to points as follows:
Ticks in each Postdata box:1 tick equals 2 pts2 ticks equals 3pts3 ticks equals 4 pts. Once the ticks have been turned into a numeric value, I then multiply the pts figures for each separate Postdata box together to provide the final figure for each runner in the race, as follows:
Runner A has 1 tick in 5 separate boxes, would score 2x2x2x2x2=32 pts
Runner B has 2 ticks in 2 boxes and 1 tick in a further 2 boxes would score 3x3x2x2=36 pts
This calculation is then repeated for each horse that is due to run in the race.
The final figures for each horse are then added together to provide a grand total for the race in question.
To calculate the probability of each horse winning the race, I then divide the final figure (FFig) for each horse by the grand total for the race and then multiply the result by 100, which produces the VALUE CALCULATOR percentage chance (VCpc) for each horse.
At this stage, I log on to an odds comparison site and list the best odds (BBOdds) that are currently on offer for each runner in the race, which I then convert into the percentage chance that the bookmakers believe that the horse has of winning the race (BOpc) Once the bookmaker odds have been converted to a percentage chance of winning, it is easy to compare the VALUE CALCULATOR derived probability figure with that calculated using the bookmaker best odds, by dividing the first figure by the second figure to produce the Value Comparison Figure (VCF).
Any runner that achieves a VCF of greater than 1.0 is technically a potential value bet at bookmaker best odds, based on the results of the VALUE CALCULATOR. The following example may assist readers to better understand the process:
|Name Of Horse||TF||G||RD||CF||DB||A||RF||FFig||VCpc||BBOdds||BOpc||VCF|
The VALUE CALCULATOR analysis suggests that Frankel would be the most likely winner of this fictitious race, but his VCpc figure of 50.7% means that he offers no value at bookmaker best odds of 4/6.
Likewise, the 3 outsiders are all underpriced, which leaves Mill Reef as the only value wager at 4/1 based on his VCpc figure of 33.8%.
Obviously we will never know the outcome of this event, however, I have detailed the results of my initial investigations in the following section.
Results To Date
The following data table summarises the results for the VALUE CALCULATOR since I commenced this analysis early last year.
|Value Calculator Qualifiers||908|
|VCQs % of Total Runners||40.0%|
|Ave Probability of VCQ Win||57.6%|
|Ave Probability of VCQ Win based on BBOdds||36.3%|
|Actual VCQ Wins||117|
|Actual % of VCQ Wins||57.1%|
|LSP if backing all VCQs||234.28 pts|
|VCQ Return On Investment||125.80p/£|
The above results clearly show that backing all of the runners that had achieved a Value Comparison Figure of greater than 1.00 to a level stake would have produced a very reasonable profit and ROI.
However, at this stage I need to qualify these results with the following explanatory notes:
Although the races to be analysed were to a certain extent chosen at random throughout last year and covered flat races run on both turf and all-weather surfaces, plus National Hunt events over both hurdles and fences, at various race distances, the sample did not include any races run in Ireland.
In addition, the sample consisted purely of handicaps at Class 3 level and above, which were contested by mature runners that generally had plenty of historic form for Postdata to assess.
Nursery and handicaps restricted to 3-year-olds on the flat, plus novice hurdle and chase handicaps over jumps were therefore excluded from the sample.
I should also point out that wherever possible, I tried to analyse races on turf where there was little likelihood of the underfoot conditions becoming markedly different to the going description on which Postdata had assessed the race.
To ensure a reasonable level of consistency throughout the sample period, I attempted to collate the bookmaker best odds data by 9.00 pm on the evening prior to the day of the race, which seemed reasonable bearing in mind that I was only able to access the latest Postdata assessment of the race by downloading the Racing Post iPad edition at around 8.00 pm that evening.
Finally, I would stress that where appropriate, all the above figures account for the impact of non-runners that were notified from the time I completed the VALUE CALCULATOR analysis up to the start of the race the following day, including the calculation of Rule 4 deductions on the level stake wagers.
Based on Industry Starting Prices, the average probability of the VCQs winning was 47.4%.
In addition to the above explanatory notes, there are 3 further points that are worth mentioning at this stage:
In total 68% of the VCQs were sent off a shorter odds than they were offered at around 9.00 pm on the previous evening, after accounting for Rule 4 deductions due to subsequent non-runners.
Based on Industry Starting Prices, the average probability of the VCQs winning was 47.4%, which suggests that many of them still remained value bets as the races started, albeit I haven't undertaken the analysis to determine the profit that would have been achieved.
I have also identified 2 additional criteria for the VCQs which when applied as filters significantly reduces the volume of wagers but massively improve both the Return on Investment that can be achieved.
However, I hope that readers will not begrudge me not sharing these filters, as they are potentially the equivalent of the ‘crown jewels' and represent my reward for having spent so much time analysing the VALUE CALCULATOR, so far.
Based on the results of the above proof of concept testing, it would appear that the initial Postdata sourced parameters are appropriate for assessing the probability of horses winning better class, all aged handicaps in the UK, across all types of racing.
Indeed, on the basis that I had only anticipated that the VALUE CALCULATOR would at best provide a useful means of assessing whether selections derived from general form study offered value at the odds available from the bookmakers and exchanges, it came as a very pleasant surprise to stumble upon a potentially lucrative betting system based on the VCQs.
However, I have already identified a number of additional parameters which need to be considered and tested in an attempt to identify the most reliable process for assessing value, including:
1. Trainer stats in respect of course and race form.
2. Jockey stats in respect of recent form, plus a historic performance at the track.
3. A method of assessing horses that have not raced for some time and therefore have no rating in respect of the Postdata recent form parameter.
4. Should negative Postdata assessments (crosses) be included in the VALUE CALCULATOR analysis?
It may also be worth reviewing the weightings that are applied to each of the Postdata parameters that are currently used within the VALUE CALCULATOR methodology, as at present each of the criteria is considered to be of equal merit.
By completing the VALUE CALCULATOR analysis for each race using the best bookmaker odds at around 9.00 pm on the evening prior to the races, it would appear that the resultant profit is significantly higher than would be achieved at Industry SP
However, at this stage, I have no idea whether the profit margin could be improved by taking the bookmaker best odds at an alternative time between 9.00 pm and the time of that the runners are placed under starter's orders, or whether placing bets at Betfair SP would produce a further uplift to profits.
Although my analysis of the VALUE CALCULATOR concept is in its infancy, I strongly believe that this methodology will prove to be a better predictor of value than the 2 suggestions that I put forward in the following article which was posted back in December 2015.
Future Projects For The Value Calculator
Where do I start?
In addition to the enhancements to the current parameters and methodology that I mentioned in the Conclusions section above, there are a multitude of additional projects that I would wish to investigate, including (in no particular order):
1. Parameters for open Graded/Group races in the UK and Ireland.
2. Parameters for lower-class Open Handicaps.
3. Parameters for races restricted to 2 and 3-year-olds on the Flat.
4. Parameters for National Hunt Novice Hurdle and Chase races.
5. Development of betting strategies based purely on VALUE CALCULATOR analysis.
6. Use of VALUE CALCULATOR analysis as a filter on existing betting strategies.
Based on the fact that the VALUE CALCULATOR analysis takes at least 20 minutes to complete manually, I suspect that the above projects will take me the rest of this decade to complete unless the process can be automated.
Unfortunately, I am not convinced that the process can be automated, as the necessary data will need to be ‘scraped' from various sources, and even if it is possible the IT skills required are certainly beyond my capabilities (I am regularly outwitted by our television and connected devices).
The Way Forward
I am not sure whether a similar suggestion has ever been attempted before, but I was wondering whether a few OLBG members would like to join me in setting up a small project group (maximum 10 individuals, plus myself) to work on the various initiatives that are mentioned in the previous 2 sections of this article.
Clearly, I cannot guarantee that any of the above betting system initiatives will prove effective.
In many ways, we will be starting with a blank sheet of paper, but based on the results so far, I suspect that between us we will find some profitable applications for the VALUE CALCULATOR, over time.
Ideally, members of the project team will be enthusiastic and sufficiently numerate to be able to analyse races using the VALUE CALCULATOR methodology and be able to commit to working on the project for an average of at least 2, or 3 hours per week.
Although it will not be possible to offer any direct rewards in return for this ‘work', I suspect that focussing on value will have a positive impact on the project team member's betting practices and whilst I intend to share the high level results with all OLBG members and readers of my future Blogs, any findings that considered by the project team to be worthy of classification as a ‘crown jewel' will be withheld for the benefit of just the project team members.
Perhaps this will prove to be a very effective example of the OLBG ethos of ‘working together to beat the bookmakers'.
If any members are interested in joining the project team, please notify me via either the comments section below or via a private message.
I hope that readers have found this analysis of interest and as usual, I would welcome any comments and feedback, especially if readers have any similar ideas/methodologies for assessing value that they are happy to share.
Also, if you have enjoyed this blog, please feel free to share the link on social media, with your friends and contacts who are interested in horse racing. Further articles relating to the VALUE CALCULATOR are easily accessible via the following > VALUE CALCULATOR INDEX
Finally, even if readers do not wish to undertake the VALUE CALCULATOR analysis on a regular basis, it may be worth the effort when considering maximum stakes bets, or for a bit of fun occasionally applying the process to the selections of OLBG members.
Click here to view other free horse racing tips.