Consistency (part 2)

Wizard Daily Report and Research - Monday, 18 December 2023.

Consistency (part 2)


Consistency Versus Win Probability

It is worth considering the relationship between the overall win percentage of a horse going into a race and its actual result in the race.

As noted in the earlier article on this topic, winning percentages are not necessarily comparable because a bush horse whose recent racing has been against lesser quality opposition may have a win percentage of 40% while another horse with a win percentage of only 16% may have achieved it against stronger metropolitan fields.

Also, whilst one might assume that a horse with a better win strike rate than its opposition should be favoured over that opposition, the fact is that the horse might not be well placed under the race conditions, not yet in race winning condition, etc.

So, merely using the raw win strike rate as a qualifier or a filter is not advisable.

The experience of most form analysts would confirm it is not possible to predict likely win probability purely from the consistency statistics in the formguide.

The table below compares the actual percentage of wins recorded by horses relative to the win strike rate they took into the race.

There are two sets of data.

First, there is the record of the 2022-2023 seasons Top 100 trainers in the metropolitan premierships, over the past five years. (Given the expertise of this Top 100, I would expect the figures for the Australian trainer population as a whole to be somewhat lower.)

The second set of stats shows the record of the Wizards Wrat 100 pointers. The Wrat stats cover the 2022-2023 season only. Obviously, there is no limitation with respect to trainers in this set.

Source all distances  sprints <=1600  distance 1601+ 
ageodds0-19% iv20-39% iv40%+  iv0-19% iv20-39% iv40%+  iv0-19% iv20-39% iv40%+  iv
Top 100          
all agesall odds14.6  1.014.0  0.920.2  1.414.5  1.013.7  0.920.3  1.414.5  1.014.8  1.019.7  1.3
 <=10/121.1  1.019.9  0.924.6  1.221.2 1.019.7  0.924.6  1.220.5  1.021.1  1.024.5  1.2
2yoall odds13.7  1.014.2  1.019.7  1.413.7  1.014.1  1.019.7  1.4nanana
 <=10/120.6  1.020.3  1.025.0  1.220.7  1.020.2  1.025.1  1.2nanana
3yoall odds16.2  1.015.1  0.920.1  1.216.4  1.015.3  0.920.2  1.214.8  1.013.4  0.918.9  1.3
 <=10/122.7  1.020.9  0.924.6  1.122.9  1.020.9  0.924.5  1.121.6  1.020.5  0.925.8  1.2
4yo+all odds13.5  1.013.5  1.020.5  1.513.1  1.013.1  1.020.5  1.514.5  1.015.2  1.020.4  1.4
 <=10/119.8  1.019.5  1.024.6  1.219.7  1.019.1  1.024.7  1.220.2  1.021.2  1.023.7  1.1
all agesall odds26.0  1.024.6  0.926.0  1.0      
 <=10/127.6  1.026.2  0.930.3  1.0      



The above analysis examines three distance ranges - all distances - sprint distances (up to and including 1600m) - longer distances (more than 1601m).

Also, the analysis records the results of all runners and, following our normal practice, only runners starting at 10/1 or less,

The win percentage for each of the distance groups is recorded, along with the impact value.

Despite the conventional view that sprinters should always show a higher level of consistency than stayers, the above figures tell a different tale with respect to the older (4yo+) gallopers.

In the 0-19% all-age group they are equal on 14.5%;  with the sprinters recording a 0.7% advantage when 10/1 or less. However, in the 20-30% group the longer distance runners outperform the sprinters by 1.1% at all odds, and by 1.4% when in the market, starting at 10/1 or less. (Note: I would not put great store on the 40% stats in this table as the number of observations was relatively small.)

Another finding is that the 3yo gallopers deliver a better result than both the 2yo and 4yo+ horses in the sprint category; in the longer distance all-odds category the older gallopers outperformed when 20% plus. (Again, ignoring 40% plus.)

We can see from the figures that the level of performance rises when we limit the survey group to those in the market, when 10/1 or less. No surprise there.

The same pattern emerges when we compare the results of those rated 100 by Wrat to the overall population (which, of course, includes these 100 pointers) in the all-distances category. There is a significant improvement, with around 26% being the norm, compared to around 14% in the overall population.

In summary, there is some pretty useful information to be derived from the above analysis.

One noteworthy conclusion is that a higher lead-in win percentage going into the race does not automatically translate into an actual improved race result.

Also, as well as the performance of the 3yos, another takeaway was the relatively strong performance of the 0-19% category, across all groups, and across all distance ranges.


Warren - Wiz-Ed

If you have any comment or suggestions about the Wizard Daily articles please feel free to drop me a line at: