Last issue, Russell Clarke, one of Britain's most prominent professional punters, revealed aspects of form analysis. Drawing on his research, he concluded that weight does not have the effect universally supposed and that the effect of Class is greater than we think. He now concludes his article.

With top-rated horses, the positive effect of a recent evident in research I carried out but less pronounced than with last-start winners. Those running within three days still achieved a 25 per cent strike rate and only a marginal 2 per cent loss on stakes invested, but the fall-off in strike rate is much less pronounced, than with runners whose break from racing was greater.

Top-rated horses running after a break of 16+ days still won at a rate of around 20 per cent. The reason for this is possibly because form ratings do not reflect last-time-out form as clearly as last-start winners. The statistics are interesting enough at an academic level but, as punters, we all want research that can provide us with profits. We require statistical trends with a long-term record that is unknown to the general betting public and likely to remain unidentified for some time.

We can then use these trends to bet profitably with confidence. I therefore examined the most likely source of profits for us: collateral form ratings in non-handicap races.

Using the research I had previously completed, the obvious area to concentrate on was that of recency. Logically, this should turn up a profitable trend and it does. Any top-rated horse in a non-handicap that has run within the past three days provided the following results from the seven years 1990-96: 190 qualifiers, 74 winners (33 per cent) with a levelstake profit of 43 units (+23 per cent on turnover).

Such a horse is four times more likely to win than a horse selected at random. A trend to follow even if there is little action with only about 30 bets a year.

In a search for more action, I turned to a favourite filter of mine. It improves almost any trend you identify. This is backing colts and geldings against fillies. This simple rule produced the following results from the 1990~96 top-rated, non-handicap races: 8606 qualifiers, 2840 winners, a 33 per cent strike rate, with a level-stakes loss of 216 pts (less than 3 per cent on turnover). No profits, but an excellent basis on to which a further filter could be added.

At this juncture it should be noted that too many caveats should not be applied to any trend. The temptation is to place the filter to produce the best results from the past. You end up with a small number of qualifiers that is not a strong enough statistical sample. So future results can prove disappointing.

Proven statistical trends only should be added as filters. These trends, such as colts/geldings running against fillies, stand up in their own right and are legitimate rather than artificial filters.

The best final filter of all is your own perception of value. The statistical trend above gives you approximately 1200 bets per year, about four per day to study All you need do is overturn the 3 per cent loss on turnover.

It is clear that form is vital to the assessment of races but, equally, it is difficult to profit from form alone since those that influence the market have an excellent grasp of form and their knowledge is readily accessible through newspapers and publications bought by the mass of punters in the betting market.

A few on-course professionals can make small profits on large turnovers by utilising form assessment, though it is generally their own personal slant on the form that gives them the edge.

*Russell Clarke is the author of a number of best-selling books on punting.

Click here to read Part 1.

By Russell Clarke