Some years ago I was introduced to the fascinating world of horse-racing statistics by a chap who had made quite a success of punting. Statistics can open our eyes to so many interesting things, greatly increasing our understanding of the racing game.

Once you start collecting your own sets of figures for different aspects of racing and then analyse what they mean for your punting, you'll be amazed by what you discover.

It really can be an exciting hobby, especially when your statistics provide you with more profits from your betting activities.

While computers are a powerful tool for processing statistics, if you don't own one there's no need for despair. There are many punters out there in the same boat, even in this computer age. Some res.,, them because they find it a bit of a nightmare learning how to drive the darn things!

Getting to grips with the Internet, databases and spreadsheets can be a traumatic experience for some, and not everyone can afford to buy an elaborate system with all the latest bells and whistles.

Not that I'm against computers. In fact, I own one, but I find it much quicker to use pen and paper when I start looking into some aspect of racing that I haven't investigated before. Later, if something begins to look promising, I may use my computer to test and confirm what I've found.

However, I've discovered that if there's a useful trend in my data, it begins to show up pretty quickly on paper, even from a fairly small sample.

Provided you have kept your daily newspapers with the tipsters' polls and pre-post betting, your formguides and race results, you will have a wealth of information to tap.

Statistics are absolutely essential for building winning systems. By way of illustration, I'm going to show you two sets of statistics gathered some time ago using only a pen and paper. We will then see how these figures can be used to develop a winning system.

Firstly, we will look at the figures relating to the beaten margin last start (BMLS) for winners on metropolitan tracks, Saturdays only, in Queensland, New South Wales, Victoria and South Australia. This information is readily available in all formguides.

The following table gives the distribution Of winners by their BMLS in lengths:

0.06127.4 27.4
0.1-0.52410.8 38.2
0.6-1.09 4.0 42.2
1.1-1.517 7.6 49.8
1.6-2.014 6.3 56.1
2.1-2.513 5.861.9
2.6-3.07 3.1 65.0
3.1-3.54 1.8 66.8
3.6-4.011 4.9 71.7
4.1-4.59 4.0 75.7
4.6-5.07 3.1 78.8
5.1-5.54 1.8 80.6
5.6-6.04 1.8 82.4
6.1+ 14 6.3 88.7
First-up 13 5.0 94.5
1st start 12 5.4 99.9

TOTAL: 223  100 per cent

These figures take nothing else into account other than the beaten margin of each winner's last start. They include all classes of Saturday metropolitan racing in the four States. The table immediately highlights the importance of good recent form in our search for winners.

The column headed 'BMLS' gives ranges for the beaten margin in lengths. The 'WINNERS' column gives the number of winners in our sample having a particular beaten margin last start. This is expressed as a percentage in the next column.

The column headed 'CUM. %' gives a cumulative percentage total as we add in each successive figure. It shows the proportion of winners for a range of beaten margins and it allows us to check that our figures are correct at the end of the process.

It's worth noting that over 25 per cent of all winners in this sample had won their previous start. Little wonder, then, that quite a few punters only back last-start winners.

Our figures show there's very good reason to do so. Horses just pipped at the post by no more than half a length at their last start tend to be good betting propositions also, on the basis of these figures.

You will notice in the table that over half of the winners (56.1 per cent) had a BMLS no greater than two lengths. For this reason, as a general rule, I don't like backing horses with a BMLS greater than two lengths. This is one of the basic rules underpinning my punting.

There are exceptions, but I use this as my benchmark most of the time.

Of some concern is the fairly high figure of 6.3 per cent for horses with a BMLS of 6.1 lengths or more. This highlights those infamous form reversals that are the bane of every punter. Perhaps this is where those very attractive longshots are lurking!

Next, we will look at figures for the win rate (some refer to it as "consistency") of all winners, once again on metropolitan tracks, on Saturdays only, in Queensland, New South Wales, Victoria and South Australia.

This information can be found in most formguides these days. Some, like the Sportsman and Wizard, provide the win rate for each horse in a very readable table.

We assign a rank to each horse based on its win percentage. The horse (or horses) with the highest percentage is given rank 1, the second highest percentage, rank 2, and so on down the list of runners in each field.

A slight problem arises when two or more horses have the same win percentage. In this case they share the same ranking. Even though horses with very close win rates have virtually the same consistency, we separate them for statistical analysis.

Hence, if we give a horse with 29 per cent a rank of 3, a horse with 28 per cent would receive a rank of 4, and so on. Although this way of doing things is not perfect, it's about the best and simplest way around the problem.

The following table gives the distribution of all winners by their win-rate ranking:

CUM. %
1 41 20.520.5
2 30 15.0 35.5
3 35 17.5 53.0
4 25 12.5 65.5
5 18 9.0 74.5
6 13 6.5 81.0
6+ 38 19.0 100.0
TOTAL: 200 = 100 per cent

Clearly, the more consistent horses win a higher proportion of races than those with a lower win percentage. Notice that slightly over half of the winners (53 per cent) come from the top three ranks, and the top five ranks account for almost three-quarters of the winners in this sample.

Of some concern is the relatively large proportion of winners outside the top six rankings. Once again, this might be the place to search for longshots.

I readily admit that both sets of figures are based on small samples. However, a clear trend exists in both sets of statistics and, more to the point, the trend is in line with our commonsense expectations, which is reassuring. Ideally, before we use these figures for betting purposes, they should be confirmed by using much larger samples.

Preferably, one or two thousand races should be tested before we can confidently use our figures for punting purposes. Often, after testing larger samples, I have found that the percentages do not change significantly from those provided by the initial smaller sample. A variation of two or three per cent is not uncommon.

However, although the sample is small, let's see what would happen if we used these figures to develop our own selection system. The figures are telling us that, other things being equal, it makes good punting sense to bet only on horses with very good recent form.

They also tell us that it makes sense to bet on the horse with the highest win rate in the field.

With these two factors in mind, suppose we adopt the following rules:

  1. Bet on metropolitan tracks in Queensland, NSW, Victoria and South Australia only on Saturdays.
  2. Consider only the top-ranking horse in each race in terms of win percentage.
  3. Bet on the top-ranking horse only if its BMLS was no more than two lengths.
  4. No horses first-up from a spell of eight weeks or more.
  5. No bet if more than one qualifier.

For my first test on paper, I used these rules on Saturday, January 17, 1998, and obtained the following results:

1Lost$6.80No betNo bet
2No betNo bet No bet$1.50
4No bet$4.50Lost Lost
5LostLostNo bet$14.40
6$2.70$2.10 No bet $2.05
7LostLostNo betNo bet
The TAB dividends are those for Queensland. For an outlay of $22, the total return was $36.10, for a profit of $14.10 which is 64.1 per cent profit on turnover, a most worthwhile figure. The strike rate was 8 wins from 22 bets, a healthy 36.4 per cent.

I'll admit the excellent dividend of $14.40 in race 5 in Adelaide made all the difference between winning and losing overall, but the $6.80 dividend in the first race in Sydney was a respectable return as well, so this system, as it stands, has the potential to land winners at very nice odds, perhaps regularly.

What you need to remember is that only two main selection criteria have been applied to the system. Using a bit of imagination and your own experience, I feel sure it would be possible to improve on this. It may require only one or two refinements using these two main rules as the basis. This is something you could develop yourself.

For example, you may wish to set a minimum win rate for your selections of, say, 20 per cent. You may wish to tighten up the BMLS requirement. These measures would probably give you a better result. I say 'probably' because the dividends may be smaller as a result of being more selective. You may wish to research this yourself.

Other fairly basic considerations would be to ensure your selection has placed previously over today's race distance and to ensure it can handle today's track conditions. Having a top jockey on board always helps.

There is much scope here to refine and develop a system like this until it becomes a formidable weapon in your punting arsenal. As you become more selective, the number of suitable bets will drop off and you will need to be patient as you wait for the right ones to come along. But isn't it worth the wait?

Profits require patience and being very selective is one of the most effective strategies a punter can have. Remember to always test your system on paper before you use money.

The ideal time span for confirming your theories is at least one year to see how they perform during all the ups and downs of a typical racing season. If you have newspapers and formguides for the previous 12 months, you will not have to endure waiting one year before you begin betting.

A good system should produce profits more or less continuously all through the year. Also, it should produce good results at all major racing venues. For instance, I would be a bit worried if a system delivered the goods only in Melbourne during the spring carnival. This would suggest there was some hidden factor at work that might fail without warning, causing considerable losses.

Don't be discouraged if your first attempts at developing your own system are not a runaway success. Draw on your own experience and refine it as necessary using only facts derived from statistics in a way similar to what has been done here.

I have no doubt that some of the very best systems have been developed by individual punters. (They probably guard them jealously, too!)

I have tried to show you what can be done with only two sets of fairly basic statistics. However, there are many other things you can research in a similar way and I will deal with some of these in future articles. Until then, I wish you every success with your statistical research and your system building. Good luck and good punting.

By Sid Marks