Last month I stated that laststart winners repeat at about an 18 per cent strike rate, roughly twice the strike rate of horses selected at random. Backing all these last-start winners, I pointed out, won't make you any money but will still provide twice as good a resulting loss (18c in the dollar) than you would get from totally random backing.

I have done quite a lot of research on this topic. To statistically measure the impact of current form I used two basic measures. Firstly, I looked at horses who won last time out and then I looked at horses who were top-rated on collateral form ratings (such as those in newspaper formguides; Zipform in Sportsman for example).

Last-time-out winners may be a crude measure of current form since there is no allowance made for class or weight but it is, at least, unambiguous. Collateral form ratings are a more accurate measurement of current form because they do take into account class and weight but are the subjective opinion of a private handicapper.

Last-start winners, then, repeat at about 18 per cent. My conclusion is that recent winning form has important predictive powers that are reflected by the odds on offer. The analysis was taken a stage further by eliminating the effect of weight, by excluding all handicap races. This increased the strike rate to 22 per cent but the loss on turnover remained almost the same at 17 per cent.

Interestingly, when you examine the record of last-start winners in handicaps you find that topweights have a 21 per cent strike rate, whereas those outside of the top two in the weights win less than 14 per cent of their races.

The conclusions I draw are that weight does NOT have the effect universally supposed and/ or the effect of class is greater than we think. Even more interesting, from a punter's viewpoint, is that laststart winners carrying top weight or second-top weight in handicaps lose only 11 per cent on turnover.

Next I examined the effect of days elapsed since the last win. The conclusion is 'the more current the current form is, the better'. The strike rate of last-start winners falls in an almost uniform pattern as the time since the last win increases.

Horses that won within the previous three days boast a strike rate of 25 per cent and a break-even situation with level-stake betting. It is when you get to 16+ days since the last win that the strike rate falls below 17 per cent and the levelstake losses are 20 per cent and more.

The betting market does not recognise the advantage possessed by last-time-out winners when that victory was 'very recent'.

The 'collateral form horse' (the top-rated horse by your favourite tipster, etc.) wins at a strike rate of around 22 per cent, recording a level-stakes loss of around 12 per cent on turnover. Immediately, this looks a better indication of form than a last-start win.

All collateral form ratings take into account the accepted effect of weight but, nonetheless, the discoveries I made when analysing last-start winners were mirrored by the form-rating top-rater when handicaps were excluded.

The strike rate leaps to 31 per cent and the loss on turnover is only 5 per cent. Once again, when we looked at handicaps, it was the top-rated horses with the two top weights that performed best with a 21 per cent strike rate and a loss on turnover of about 9 per cent. Once again, the effect of weight is either overestimated and/or class is underestimated.

• Russell Clarke is a UK professional. His statistics relate to UK racing.

NEXT MONTH: Russell Clarke concludes his series with more details of his research into various aspects of form analysis.

Click here to read Part 2.

By Russell Clarke

PRACTICAL PUNTING - JANUARY 1998