I have been involved in horseracing, in one form or another, for many years. In my early twenties, while working as a newspaper reporter based in London, England, I had a friend who was a dyed-in-the-wool gambler.

It wasn't long before I was learning the art of reading form in the Sporting Life and placing bets in one of the numerous betting shops in that country. Like all newcomers to the sport, the obvious form was what I followed, and of course it was not too long before I realised that I needed to learn a lot more and bet, at least in the meantime, a lot less.

I visited a bookstore in London that specialises in sporting books, and after browsing around for a little while happened across Picking Winners, written by an American named Andrew Beyer. In America, of course, most of the racing is done on dirt tracks.

Beyer himself would be the first to tell you that his methods do not work particularly well with turf horses, and there is very little dirt racing in Britain. I was so impressed with the first chapter of the book, however - which I read right in the bookstore - that I decided to purchase the book anyway. That turned out to be a very wise investment.

If you are a serious handicapper and have not read the book, make an effort to do so. You may not agree with all that Beyer has to say, but one fact I can vouch for: Andrew Beyer has made a lot of money betting on horses using his speed rating system. Of course, I read many other horse-racing authors too.

I later moved to the United States, so it was natural that I would find titles by people like Tom Ainslie, Tom Brohamer, Steve Davidowitz and Howard Sartin, to name but a few. I noticed, over time, that some of these writers tended to discount the importance of weight, while others thought it was a major factor.

I became intrigued with weight as a handicapping factor, and eventually developed a speed rating system (similar to that taught by Andrew Beyer) that took weight into account.

In a nutshell, horse A may be rated at 60 and may be carrying 57kg in a certain race; horse B may be rated at 56 and may be carrying 50kg. With my system, I subtracted the weight from the rating, and the horse that potentially had the best chance was the horse with the highest positive figure.

To illustrate my example, the figure for horse A would be 60 - 57 = 3; and for horse B the figure would be 56 - 50 = 6. Horse B thus had a 3kg advantage over horse A. Using Timeform figures of the era, if the race was over 1400m, the 3kg translated to about two lengths.

I had moderate success with this system, after taking into account a host of other factors, such as current form, distance suitability, the projected pace of the race, and so forth. In many instances, however, I noticed that the theoretical weight advantage did not actually happen when the race took place.

Horses that had an advantage of three or four kilograms to other animals on paper were being beaten. I wondered why.

I began noticing that if horse A beat horse B by (for example) two lengths, and then these two horses met again, with horse A now carrying 3kg more, horse A still tended to beat horse B by one or two lengths. The question now became, was weight a highly overrated handicapping factor?

That question is, of course, easy to ask but difficult to answer, and there is no shortage of punters and horsemen who have very firm - and polarised - beliefs about the subject. After digging around a bit, I found that there were trainers who regularly scratched horses when they were allotted 55.5kg instead of the 54kg the trainer thought was fair; and there were other trainers who did not give a hoot about the weights that were set.

I knew that there were roughly 7000 turf races run in the United States each year and I figured that ten years worth of data, covering some 70,000 races, would be a large enough sample to give reliable results. (I prefer turf to dirt racing, so wanted to study that surface.)

I am no whiz with the computer so my husband Gabriel suggested that we approach a friend, Brent Carlson, to write some programs for collecting and massaging the huge amount of data. Brent is also an avid horse-racing fan, and it did not take a lot to persuade him.

We purchased the result of every race run in America, including dirt races, for the years 1991 to 2000. The Daily Racing Form publishes these results on compact discs, and it was easier - and cheaper - to simply buy all the results rather than selected ones.

The results are very comprehensive, by the way, listing the running position for each horse at three, four or five points in the race (depending on the distance), times for various stages of the race, post position, weights, jockeys, betting odds, track conditions, any pertinent comments about the running of the race, and so on.

Once we had the data safely stored on the computer, we had to decide how best to use it. The three of us spent many an evening arguing the merits of various points of view. Eventually, we decided that we would attack a single area at a time, and then move on only after thoroughly analysing that area.

We started out, as practice because it was not too difficult, by looking at the times recorded at various distances, from 1000m sprints to 3200m routes, with everything else in between.

Perhaps I should point out, before we go much further, a few facts regarding turf racing in the United States. Firstly, turf tracks are generally built as a separate oval inside of the dirt track (referred to as the main track here), and as the majority of main tracks are one mile (1600m) ovals, most turf courses are not much more than 1400m in circumference.

In most cases the courses are more or less flat, that is to say they lack the hills which are encountered on European courses. Epsom Downs, home of the English Derby, is a good example of a hilly course.

The length of the stretch on most turf courses is about 925 feet, and the average width of the courses is right on 70 feet. Some courses have banked turns, and the inclination in such cases is about 6 per cent.

A typical turf course here has a four-inch base of sand and clay, likely a fifty-fifty mix, and Bermuda grass, blue grass or fescue is often used for the sod (a mixture of these grasses is quite common). The grass itself is seldom cut very short - four inches, perhaps even five, is about the standard - and this ensures a certain amount of "spring" in the surf ace.

The official descriptions of turf course conditions are "hard", "firm", "good", "yielding", "soft" and "heavy", but the great majority of turf races are run on courses that are more or less firm. Unless a major race is involved, track officials quickly cancel turf racing when it is raining, with turf races being switched to the main track.

Except for Florida and California, turf racing disappears during the winter months. In New York, for instance, the temperature may never rise above freezing point in January and February, and grass goes into hibernation.

One final note. In most cases, a day of racing at any given bigger track may include six or seven dirt races and two or three turf races. It is very unusual for all the races on the day's card to be scheduled on turf.

We collected times, at each distance, for a number of classes of race - Graded Stakes, Major Stakes, Minor Stakes, Handicaps, Allowance, and three levels of claiming: high ($75,001 or more), mid-range ($40,001 to $75,000) and low ($40,000 or less).

We also collected times for maiden races, both claiming and maiden specials. (In maiden special weight races, horses may not be claimed and all runners carry the same weight, except for sex and apprentice allowances.)

The conventional wisdom is that, for a variety of reasons, different tracks produce different times. This tenet is an important part of the Beyer philosophy, and Beyer thus keeps averages for all tracks in North America.

However, the first results that we produced shocked us so badly that we ran the exercise again, with Brent checking the program to make sure there were no errors. The same results came back a second time.

What we found was that a Graded sprint of 1000m produced virtually identical average times no matter where it was run - Gulfstream, Belmont, Churchill, Laurel, even the supposed sun-baked, lightning-fast tracks of southern California (Hollywood, Santa Anita and Del Mar).

The same was true for 1600m races, for 2000m races and for 3200m races, and all the other common distances.

In ungraded stakes races, i.e. Major Stakes, the results were quite similar. The times were slightly slower than for Graded races, of course, but the times were the same across the country. Minor Stakes, Handicaps and better-class Allowances all produced the same result.

The only slight variations to show up were in low-level Allowance races, and in claiming races. One mile claimers that were run in California, for example, were slightly faster than those run in Maryland. After some debate, we agreed that the time variance in the claiming races was almost certainly due to the differences in the intrinsic ability of the horses competing in those races.

In spite of the commonly held belief that a $20,000 claimer in Texas is equal to a $20,000 claimer in California, the statistics tell us this is not so. If the better-quality animals can all run the same times no matter where they are, the only logical reason that the times vary for cheaper horses is that the quality of those horses varies depending upon where they race - and that the claiming price is not necessarily indicative of ability.

If the purse for a $20,000 claimer is $18,000 in California and $10,000 in Indiana, it is natural that the better animals will tend to be in California and the inferior ones in Indiana.

We tested this theory by running the program again, this time ignoring the claiming prices and instead sorting the data by purse value.

As we expected, the average times were now much more uniform for claiming races. For example, when the purse was in the $18,000 to $22,000 range, the average time for aclaiming 1600m was 1:36.3 (1 minute, 36.3 seconds) but when the purse dropped to the $9000 to $11,000 range the average was a full second slower.

More importantly, the averages were again true nationwide. Thus, the first important conclusion that we reached from the database was that any given horse should be able to reproduce the same time on any United States course.

Put another way, if a horse ran in a 1600m race at Hollywood Park last Friday and clocked a time of 1:34.0, it would probably have clocked the same time anywhere else provided that the going was the same and the pace was similar.

Next, we wanted to know if there is any relationship at all between, in very broad terms, weight carried and time recorded. If the lower weighted horses consistently clocked the faster times, it would indicate that carrying less weight was a definite advantage. Should the faster times be spread more or less evenly over the entire weight range, it would indicate that a high or low weight made no significant difference to an animal's ability to run at its best.

On the other hand, if the faster times were clocked by the horses carrying the most weight, it would mean that weight - certainly in the American context - was hardly any factor at all.

Roughly speaking, of the 70,000 or so races that we studied, about 33,000 were above average in terms of time, and 34,000 were below average. The remaining 3000 races were right around the average.

Four very common distances run on turf in North America are 1000m, 1600m, 2000m and 2400m. Brent sifted through the database and isolated only those races, then selected only the faster than average races, the weights carried by the winner of those races, and the highest weight in the race.

We looked at the 1000m figures first, and found that 72 per cent of the above-average times were recorded by horses carrying within 2.5kg of the high weight. We were rather surprised by this result - we had expected the faster times to be evenly spread over the entire weight range - but our initial feeling was that in such short races weight may well be less of a factor than in longer races.

At the 1600m distance, however, the result was almost identical. A little over 73 per cent of the above average times were clocked by horses that were carrying weights within 2.5kg of the high weight.

For 2000m, it was 71 per cent of the fast times recorded by horses carrying within 2.5kg of the high weight; and for 2400m races, the figure was almost 72 per cent.

We considered these results for a while and tried to come up with reasonable explanations for the unexpected results.

Ynez Ybarbo runs a daily jackpot based on four American races with an entry fee of 25c per line. You can check out her website at the following address: www.ynez.com/racing/menu.htm

NEXT MONTH: Ynez Ybarbo completes her special report, and comes up with some very surprising conclusions. Don't miss this great article AND in our September issue don't miss a review of Ynez's articles by our senior contributor E.J. Minnis.

Ynez Ybarbo, a native of Las Palmas, has a BA degree in Communication Arts and Sciences from the University of Barcelona. She was a staff reporter for the English-language Costa Blanca News for three years, and was then a business reporter for Diario Information (Daily Information) in Alicante for about four years. Since moving to the US, Ynez has been a freelance writer, with assignments at various large companies such as Ford, Boeing and Du Pont. She has also completed several investigative reports for North American newspapers located in Canada, Mexico and the United States.

Click here to read Part 2.

Special report by Ynez Ybarbo

PRACTICAL PUNTING - JULY 2003