Fast Track Simulcapping

THE C-FACTOR: Handicapping the Crowd

Introduction

Whatever game we play (assuming it is not one of pure chance like playing a slot machine or cutting cards), whether gambling is involved or not, it is necessary to have some knowledge of your opponent. In chess you have to consider what moves your opponent may make in deciding your own, or defeat will be quick. In tennis it is useful to know whether your opponent's backhand is his strength or a weakness before you hit the ball there. And in cards besides knowing the percentages and remembering what cards have been played it is useful to take into account the skill level of your opponent.

In handicapping we also need to know and understand our opponent, but that opponent is not other individual handicappers. Our opponent is the crowd as a wagering market that we all participate in establishing the odds for each race. Anyone who practices the pastime of handicapping horse races has some knowledge of the importance of the crowd in the outcome of his or her endeavors. The favorite tends to win a significant number of races, but often at uninteresting odds. In fact, numerous texts are usually references for the common wisdom that we can expect the favorite to win close to one third or 33.3% of all races, regardless of class, distance or surface, and result in a net loss of about 10%. This would translate into an average return per $2 win wager of about $1.80 or so.

So the crowd is a pretty good handicapper. And however we want to measure our results as handicappers we end up measuring up against the crowd because the pari-mutuel wagering system that has been adopted in horse racing pits us all not against the track or the house, but against each other collectively. The track takes its 17% or more off the top, and then divides up the remainder amongst those handicappers with the winning tickets. The odds offered on the tote board are thus determined by the crowd's collective wagering preferences much like a commodities or futures market.

But not much time is spent by most of us considering and understanding the workings of the crowd in arriving at its choice of favorite and the odds offered on the various horses. In this day and age with the impact of computers and improved speed ratings are the anticipated results of favorites still as earlier studies would suggest? Or does the crowd find more winners in more predictable races now, and is there variation between the crowds at different tracks?

The Wagering Pool As A Market

Considering the pari-mutuel wagering system at the track as a market we can look for efficiencies and inefficiencies within that market to assist us in determining when a particular horse in a race may be an overlay (under bet and offered and higher odds than its probability of winning would predict) or an underlay (over bet and offered a lower than fair odds). The efficiency of a market can be described as measuring how accurately the prices offered equal the value of a commodity. The efficiency of the wagering market then can be measured by its ability to arrive at fair odds for all horses. With a 17% track take such a market would result in the handicapper losing 17% over time regardless of which horses were back. An efficient wagering market would be expected over the long run to return close to $1.66 for each $2 wagered, regardless of the odds offered for each ticket.

The final wagering decision of the crowd in determining the wagering market odds can be described as being arrived at by combining and weighting of all the various handicapping systems of the individuals that comprise the betting public. A massive amount of information is available contained in the daily racing form, online and in various texts. Some of this of course will contain errors or misinformation or can lead us astray. All of this goes into the mix, and the crowd collectively does the best it can in setting the odds through the wagers purchased by the various handicappers.

The Study

The following study is an attempt to begin understanding the crowd as our opponent. Data was collected from a random sample of 932 races at numerous North American racetracks in the first part of July 2006. Two year old races were excluded on the theory they were too unpredictable. Races with fields of fewer than five horses, and races moved off the turf and run on dirt, were also excluded.

Since it is the crowd factor we are studying, for this discussion it will be referred to as the "C-Factor". The favorite will be referred to below in the tables and analysis as the C1 horse or C2 group of horses, the second favorite will be the C2 horse ... and so on.

To begin let's consider Table 1. This table lists the results depending on whether the horse was C1, C2, C3 etc. The results for C8 and higher were also grouped together at the end. This is intended to balance out the slightly high $2ROI for the C9 and C12 categories which were inflated by the high price of a few long shots. Those should be expected to rarely occur.

Table 1 -- Win Percentage and $2ROI grouped by Crowd Choice

These results are quite interesting. First of all the C1 favorites performed a little better in terms of win percentage than we would traditionally expect, winning 35.6% of the time. In fact, either the C1 and C2 horses won a remarkable 59.8% of the time … almost 3 in every 5 races. This may be a result of more sophisticated computer handicapping and speed ratings that we all have access to these days.

However, on the down side, the average net return (the $2ROI) for the C1 group was only $1.71 per $2 win wager. This is less than the traditional expectation of losing 10%, and not much better than breaking even against the 17% track and government take.

Collectively the horses ranked 8^th or higher won only about 2% of their races, returning a meager $1.27 $2ROI. Long shots as a group should clearly be avoided in the absence of clear reasons to support a particular horse.

Focusing on the C2 Group

There were surprisingly positive results when we focus on the C2 group. The C2 group won almost 25%, or one in four of the races, and performed significantly better in the $2ROI than expected at $1.92, close to the break-even level. A loss of just 4% without any handicapping exceeds the expected 10% loss for favorites, and is a good place to start. It would appear that possible overlays resulting from inefficiencies in the win wagering market may lie in the C2 (second favorite) horses rather than the C1 group.

A closer look appears justified to see if we can isolate where these market inefficiencies may lie.

Inefficiencies could be expected to arise for a number of reasons. These could include, but not necessarily be limited to, the following:

1. Size of field.

2. Odds of the C1 horse, reflecting the degree of crowd confidence in the favorite.

3. Distance.

4. Class.

5. Surface

6. The particular crowd wagering at the track.

A Comment on Sample Size

At the risk of stating the obvious, it should be noted that the following observations often relate to smaller sample sizes than the initial 932 race sample. Naturally this will make the findings less conclusive the smaller the particular sample is.

1. The size of the field

The size of the field in the race might effect whether there are inefficiencies in the wagering market as handicappers tend either to over bet the C1 group incorrectly assuming the competition unworthy or under betting it on the assumption that the odds are too low. Tables 2 and 3 compare field sizes and the effect on win percentage and $2ROI for the C1 and C2 horses.

Table 2 -- Comparing different field sizes for C1 horses

What is noteworthy in Table 2 is that the C1 horses were apparently over bet for the small fields (NH5 and NH6) and also for larger fields of NH10 or more. It was only with the NH7, NH8 and NH9 results that there was a $2ROI higher than $1.66, as would be expected adjusting for the track take. As noted above the sample size for NH5 is quite small, but it would seem natural to combine it with the NH6 group.

Table 3 -- Comparing different field sizes for C2 horses

Table 3 shows some interesting results comparing the C2 group of different field sizes. These results further suggest that the key to finding inefficiencies in the wagering market might lie with focusing on the C2 horses. Although generally winning fewer races than the C1 horses, the C2 horses tended to have higher $2ROI results at most levels. For several field sizes average $2ROI overcame not only the track take but also was over $2.00.

Table 4 compares results for 227 races where the number of horses is at least 10. Where NH exceeds 10 the C2 horses often performed as well or better than the C1 horses in terms of win percentage. This may indicate that in many races of larger field sizes there is often really not much to choose between the top two choices, with the crowd tending to artificially follow one or the other to establish a crowd choice. As a result the C2 horses appear to have been frequent overlays for races with NH more than 10.

Table 4 - Comparison of crowd choices in races where NH is 10 or more

Although the sample size of 227 races is not large, the results of Table 4 support the possibility of finding overlays among the C2 group of horses with a positive $2ROI, and even a few more winners than among the C1 group.

2. Odds of the C1 horses

The relationship between the odds offered on the horse preferred by the crowd is considered in Table 5.

Table 5 - C1 Odds and results for C1, C2, and C3 compared.

The results of Table 5 suggest that when the odds on the C1 group of horses increase they perform less efficiently in terms of $2ROI, and as expected they tend to win less frequently. Beyond about 3/2 the C1 horses tend not to be able to overcome the track take, and the win percentage for the C1 group was no longer higher than for the C2 group. Once the C1 odds were more than 5.2 the C3 group actually outperformed both the C1 and C2 groups, but this was a smaller sample size. Again the best possibility for finding overlays appears to be in the group of C2 horses.

Combining Field Size and Odds

When an effort was made to combine the effect of odds on the C1 horse with field size the result did not appear to improve much, if at all, on the results in Table 4.

Table 6 - Results for C1 and C2 where NH is 10+ and C1 Odds considered

3. Distance

As shown in Table 7 the tendency for the inefficiencies in the wagering market appeared more in sprints and races of a mile rather than for route races. However the C2 group still outperformed the C1 and C3 groups in terms of the $2ROI at all distances.

Table 7 - Results grouped by distance

4. Class

The findings regarding class in Table 8 are also interesting, and they do coincide with what one might intuitively expect.

Table 8 - Results grouped by type of race class

That the C1 horses perform better in races with a higher class of horses is not surprising, and in this smallish sample they actually just past the break-even point. The inefficiencies favoring C2 horses seem to present themselves in the races with more inconsistent horses, as the C2 horses did well especially in the maiden claiming, maiden special weight, and optional claiming races. However, both the C1 group and C2 group did relatively equally well in the claiming races in terms of $2ROI and the C1 group won a solid 37.5% of those races. Neither the C1 nor C2 horses did well in Allowance races, with the C3 group performing best for that class.

5. Surface

The results for the C2 group were stronger for dirt races than for turf races, with both the C1 and C3 groups producing a better $2ROI on turf. These results are set out in Table 8.

Table 8 - Results grouped by track surface

6. The track

When we break down the results by track the sample sizes are small, some very small, and it would be unwise to read too much into them without further investigation. However, it was noted that there were differences between the various tracks in terms of both win percentages and $2ROI results. This suggests that there may be some significant differences between the bettors following different tracks, and it may be productive for you to compile statistics for your preferred tracks to determine the profile of the particular betting public that you are competing against.

CONCLUSIONS

This study suggests that for dirt races there may be inefficiencies in the win wagering market that tend to create more overlays within the C2 group of horses than for the other groups. The results were more conclusive in terms of a positive $2ROI for the following categories:

• Races with more than 10 horses in the field

• Races in which the odds on the C1 horse were 3/2 or higher

• Sprints and mile races

• Claiming, maiden, maiden claiming and optional claiming races.

These findings should not be relied upon in the absence of other solid handicapping criteria as the sole basis for making a wager. However, in instances where the C2 horse appears to be a solid contender based upon your handicapping criteria, they do suggest that it may tend to be an overlay more often than the crowd favorite C1 horse.

Future Research

Further research certainly appears justified to confirm the above findings and to see if we can learn anything more. For example, is there a relationship between positive findings for C2 and how close its odds are to C1 or to C3? And can we narrow in on when C3 may be a good choice? It will also be useful to look into what variation there may be from track to track, or perhaps that should be from crowd to crowd. Do the crowds following the New York races follow different patterns from those in California, Kentucky or Florida?