In this, the final of a three-part series, Heale Yardley looks at how to grade races according to their predictability so as to offer a means for determining relative bet sizes from one race to another.

In the last two parts of this series I have looked at various aspects of staking.

One of my most important conclusions was that selection methods and staking systems should be considered together. Indeed, the strength of your selections should have a bearing on the amounts you decide to back them with. For example, more money should be wagered on the horses with a higher probability of winning and less on the lower chances.

What I am going to suggest to you in this article is that you should take this approach one step further and concentrate your money not just on the horses with higher chances but only on those with higher chances in the better quality races. The plan of attack I have in mind is to grade each race on a day's programme according to its predictability.

The concept of predictability is important because in theory there will be more upsets in the less predictable races. Clearly, you should be betting more cautiously or not at all on these less predictable races.

Let's look at how we can grade races according to these and other factors affecting predictability by considering the Melbourne Cox Plate day meeting (October 22, 1988).

Predictability usually goes hand in hand with quality as the better quality races are usually contested by more well performed horses with clearly disclosed form. By 'quality' I am really referring to the class of a race-from maidens, improvers and restricted handicaps at the lower end of the scale through to welters, highweights and WFA races at the high end.

The class of a race can be determined once you are familiar with race classes as outlined by Don Scott in his recent books The Winning Way and Winning More, or as tabulated by Brian Blackwell in his Invader Ratings. The class of a horse can be determined by looking at the class of race it usually races in and by considering its win percentage and prizemoney earnings.

Notice how I have separated the concept of a race's class from that of the class of the horses racing in it. This is because you could have a welter in which many of the horses entered are restricted gallopers having won no more than one start over the past 12 months.

Let's look at the Moonee Valley races on Cox Plate day to see what I mean. We will try to grade them from grade A for the most predictable through to grade C for the least predictable races worth considering and grade D for those that should be ignored. Please note that these gradings do not officially exist but have been devised simply as a means for us to determine in our own minds how predictable races are.

Obviously the Cox Plate is the highest class race of the programme. You don't need to look at its $1.5 million in prizemoney to tell that. It would have to be a grade A race.

The Moir Stakes ($200,000), Herald Vase ($100,000), Famous Grouse Moonee Valley Cup ($90,000) and Waterford Crystal Mile ($65,000) were the other feature events of the card. Reasonably good quality, but not quite up to the Cox Plate standard, they would probably all classify as grade B races.

Should any of the other races be considered? Race 1 was a 2yo. race in which six horses had only had one start in their career and the others no prior race experience at all. Not a race I'd put any money on-a grade D race. Note though that particular good quality 2yo. races could be given a C or even B grading.

Race 2 was for 3yo. fillies which, as any class conscious punter will tell you, is not all that far up the class scale. Is it worth considering? Well look at the horses in it. Not many of them were in the money at the recent starts and only the two topweights had any real class about them-look at their win percentages (both over 40%) and prizemoney earnings (both around $100,000). Most of the other horses in the race had prizemoney earnings around the $10,000 mark and win percentages well below 20%. You might pass this race unless you were keen about one of the top two horses in it (which were Memphis Blues, came lst, and Grey Fille, 4th). A grade C race.

For future reference I suggest you consider horses with prizemoney earnings above $100,000 and win percentages above 25% as good horses. If there are lots of these horses in a race then the race is probably worth considering.

The remaining race was race 8, the BMW Australia Stakes ($45,000) for mares four years old and upward. You might wish to leave this race alone because the horses in it don't quite live up to the prizemoney and win percentage standards mentioned above and because it is not an open class race. Let's make it a grade C race.

We will now look at some other factors to see if we need to further rank the remaining feature races.

The number of starts a horse has had in its current preparation is important as horses that are first up or just one or two runs into a preparation have not usually displayed enough form for reliable predictions to be made.

Let's look back at one of the feature races, the Moir Stakes. Four of its seven runners were only two starts back from a spell. Makes it a bit risky as far as I'm concerned. Take away this risk and I might have considered giving the race an A grading. As it turns out the risk was such that I would have graded it a C if it weren't for the presence of top horses like Redelva, Special and Rancho Ruler.

Distance affects predictability in more ways than one.

Firstly, races where a large number of horses are jumping up in distance are not very predictable as there is no guarantee that the horses will accommodate the extra ground unless they have done so before on a number of occasions. The same argument applies to horses coming back in distance.

Secondly, races run over long distances, say 2000 metres and longer, have more room for uncertainty because of the more numerous combinations of tactics that can be employed not to mention the greater amount of time available for things to go wrong. For this reason I would rate the Moonee Valley Cup a grade B race rather than a grade A race.

For example, another reason why I classified the Moir Stakes as a grade B rather than grade A race is that the favourite, Rancho Ruler, was dropping from 1600 metres back to 1000 metres. This (together with the fact that it hadn't raced for nearly a month) makes anyone's assessment of its chances less reliable.

Rather than size your bets evenly across all races, you should put more aside for the higher graded races. Let's say one unit on the grade C races, two units on the grade B races and three units on the grade A races.

Following the example of Cox Plate day, suppose you have $50 to split up amongst the one grade A race, four grade B races and two grade C races. Applying three, two and one units for these grades respectively gives a total of 13 units which, when equated to $50, is approximately $4 per unit. Hence you
should bet $12 for the grade A ($4 times three units), $8 for the grade B and $4 for the grade C races.

The following steps should be taken when trying to grade races according to predictability.

  1. Races should first be graded as A, B, C or D according to their class. All open handicaps, welters, highweights, WFA races and feature/quality handicaps should be given a B grading and those of an exceptionally high quality an A grading. The lowest class races, like those restricted to very poor class gallopers and 2yo. races should be given a D grading while those in between a C grading.
  2. Gradings should be adjusted upwards or downwards by one grade to reflect the fact that the field may be of much higher or lower class than is usual for such a race. Prizemoney earnings and win percentages should be closely looked at to identify the class of individual horses.
  3. Gradings may be adjusted downwards by one grade to reflect the fact that a large number of horses are resuming or are only one or two runs back from a spell.
  4. Gradings may also be adjusted downwards by one grade to reflect the fact that a large number of horses are stepping up in distance or if the race is of average quality and distance over 2000 metres.
  5. Bets should be allocated in the proportion three units to grade A races, two units to grade B races, one unit to grade C races and none to grade D races.

Remember that these rules are not hard and fast but rather a means to encourage you to consciously think about the relative strengths of races upon which you may be about to invest your hard earned money.

Until next time, good punting.

Regular readers will recall that Neale Yardley began writing for P.P.M. last year with a series of articles called 'Winning With Computers'. Since the conclusion of that series Neale has received many letters from readers interested in more punting programs for their computers. To assist these readers, and those who have more recently acquired computers, Neale has released a fortnightly newsletter specially for computer conscious punters.

Click here to read Part 1.
Click here to read Part 2.

By Neale Yardley