Readers of last month's article will recall our conclusion that last-start winners were good candidates for quinella betting. Such horses won 14.1% of all races and came second in 12.3% of all races during the racing season August 1, 1989 to July 31, 1990. Furthermore we found that last-start city winners won 17.8% of races and came second in 13.5% of races.

Compared to our average benchmark figure of 10%, last-start winners and last start city winners were better than average win and second place candidates but still placed third no better than average. What's more, horses that came third at their last start only came third next time out in 9.9% of races.

The question posed at the end of last month's article was whether we could find a set of selection criteria that produced a better than average third place strike rate? After all we have managed to find a means of generating a better than average second place strike rate for quinella bettors so what about a higher than average strike rate for trifecta bettors?

In much of what follows we will attempt to answer this question. Although we will be able to generate a higher than average third place strike rate, the way in which it is achieved will probably be of more interest to win only bettors than trifecta bettors! The reason for this is that there is no way you can produce a good third place strike rate at the expense of win and second place strike rates.

Accordingly we will now try to generate a higher than average third place strike rate by improving our win strike rate beyond 20%. We already know that last-start city winners that last raced within 0-14 days win 19% of races. Now we need to consider other form factors that will further improve this strike rate.

So far we have considered factors like last-start wins and days since last start. Both these would ensure the horses in question are likely to be fit and ready to win. What they don't really tell us is whether such horses are racing above or below their class. (We do however know that they are at least Saturday class horses as evidenced by their last start city wins.)

Without trying to assess the class of every race and in turn of each horse in the race, we can usually gain an indication as to whether a horse is suited to a race by it's price. By this I mean SP or pre-post price quoted for the race the horse is about to race in.

Readers of my Annual article will recall how I used a horse's last-start SP (by restricting it to no more than 4/1) to help improve both strike rate and profitability. What I plan to do here is to take a look at last-start SP as well as next-start SP and compare the effect of the two factors.

First we will consider last-start city winners that last raced within 0-14 days and had a last-start SP of 4/1 or less.

My computer tells me that such horses won 24% of races during the racing season August 1, 1989 to July 31, 1990. During the same period the strike rate for these horses coming second was 13% and for coming third was 12%. Finally we have achieved a third place strike rate better than the 10% average!

Now we will forget about last-start SP and look at next start price. We will use starting prices for the purpose of this test although in practise you will have to use pre-post prices either from the betting ring or morning newspaper.

So we will look at last-start city winners that last raced within 0 -14 days and have an SP at their next start of 4/1 or less.

My computer tells me that such horses won 30% of races during the racing season August 1, 1989 to July 31, 1990. During the same period the strike rate for these horses coming second was 20% and for coming third was 14%. Yes, we have finally got a significantly better than average third place strike rate. As suggested earlier we have done it not at the expense of the win and second place strike rates but by actually increasing the win and second place strike rates.

The fact that our strike rates vary with SP should come as no surprise as the shorter priced a horse, the more likely it should win. Next month we will examine the relationship between SP and winning chances more closely. For example do even-money chances really win 50% of the time, or do 2/1 chances really win 33% of the time. Their odds suggest they should but our statistics will reveal the true answer. Of course the results of this research will have tremendous benefits for the punter wishing to know whether a particular priced horse is likely to be at value odds or not.

First up I must thank all of you who have written to me with your system ideas. The response to the What If article in the Annual and this current P.P.M. series in particular has told us two things. First there is a lot of interest in serious betting statistics and second, a lot of readers have some very sound ideas about how to make money on the punt.

Since I have not had time yet to personally respond to all letters I will take the opportunity to respond to some of them each month in these articles. While most answers published here will give you useful ideas. about how to develop profitable systems, many will be brief due to the wishes of some readers to keep their ideas confidential.

Mr J. B. of Ararat, Vic., suggests that my Top Notch system from this year's Annual could be further improved by only considering horses at their third or fourth run back from a spell. He correctly points out that this is when, at least on average, most horses reach their peak. I'll have to do a bit more programming before I can test this but in theory it should improve the strike rate. Whether the profit improves as well depends on how many other punters follow the same idea.

Mr J. B. also suggests confining ones selections to TAB Numbers 1, 2 and 3 only. This is an interesting idea that a number of readers seem keen on. Mr G. J. of Gawler, S.A., considers TAB numbers 1-8 while Mr M. A. of West End, Qld., considers only TAB number one. All these readers argue that more winners come from the top TAB numbers (e.g. 1, 2, 3). Tests I have carried out support this-not just because there are more horses racing with the top TAB numbers but because the top-weights are generally the better class horses.

Another factor mentioned by a number of readers is the win percentage of a horse. A number of readers suggest considering only horses with high win percentages. For example Mr L. C. of Newmarket, Qld., suggests 20% and above while Mr K. B. of Grovely, Qld., suggests 25% and above. Tests I have carried out show these horses to be not just more consistent but also more profitable. Like the top-weights, they are generally the better class horses. This topic will be examined more closely in a future article.

TELETEXT BREAKTHROUGH
In case you don't know, Teletext is a text information service transmitted by Channel 7 T.V. stations and many of its affiliates. The information includes TAB pre-race approximate dividends and until now could only be decoded using special Teletext television sets. I say until now because recently an Australian company commenced manufacturing Teletext decoder cards that slot inside IBM compatible computers to provide the computer with Teletext capabilities.

The advantages of having Teletext on a PC are numerous. For a start the 'Teletext PC' can save in its memory nearly 500 Teletext pages and continually update them. This means you no longer have to wait for a particular page to come around as you do with a Teletext T.V. You can instantly change from one page to another with no waiting time, e.g. from Vic. TAB pre-race approximates to N.S.W. TAB pre-race approximates.

Another advantage of the storage availability is that the 'Teletext PC' will keep earlier versions of pages so if you are looking at the latest Sydney pre-race approximates you can quickly flip back through earlier versions of the same page to see what the pre-race approximates were earlier in the betting.

Two more advantages of the 'Teletext PC' are that you can print Teletext information on a printer or save it to disk for later processing. Great if you want to keep a hard copy of TAB results or save dividend fluctuations for later analysis.

To be able to do all of this you will need an IBM compatible computer capable of accepting the decorder card and a VCR (or T.V. with a video out socket). Printer and hard disk are not essential. Cost is about a quarter what you might pay for a Teletext T.V. so even if you have to pay an extra \$800 for a computer it is still well worth considering.