Hucksters have been peddling tips, systems and gadgets to beat the races for as long as the sport has existed. Because of the widespread human desire to make a fast, easy buck, this has often been a profitable business.

The purveyors of these systems traditionally have been small-time hustlers trying to make a fast buck for themselves, but now a major corporation has moved into their territory. Mattel Electronics is mounting an unprecedented advertising and promotional campaign on behalf of its new $100 computer, the Horse Race Analyzer.

Mattel proclaims in newspaper advertisements across the country that "thanks to the marvels of modern electronic computer technology there really is a better way to handicap a horse race." The company has teamed with major retailers to market its product. Woodward and Lothrop will invite members of the local media to a luncheon at Bowie next month so they can see the Analyzer in action.

Racing systems have come a long way from the days when their inventors would place little ads in pulpy magazines offering to sell for $20 "the wonder-working formula that provides the surest thing in turfdom." But the Analyzer is still based on the same premise that those simplistic fraudulent systems were: that the raw data in the Daily Racing Form can somehow be marshaled and weighted to produce a winning formula.

A bettor using the Mattel device feeds 30 pieces of information on each horse from the Form into a snazzy hand-held computer, which then analyzes the data and gives the animal a numerical rating. This may sound reasonable to a neophyte but, in fact, nobody beats the races with formulas. Computers may be able to master blackjack, bridge and backgammon, but these games are governed by immutable probabilities. Horse racing is not. Statistics and probabilities have only minor importance in handicapping.

Successful horseplayers win principally because they have superior information. They may have insights based on watching horses run in their previous starts. They may calculate speed figures that give them precise measurements of horses' ability. They may be astute judges of horses' appearance. Above all, they beat the game by using their creative intelligence.

But the notion that a formula might unravel the mysteries of racing remains a seductive one, and it is especially seductive for people who appreciate the power of computers. One man seized by this vision was William Quirin, a mathematics professor at Adelphi University. He fed data on thousands of races into a computer and published the results in a book, Winning at the Races: Computer Discoveries in Thoroughbred Handicapping. Mattel approached Quirin to develop the Horse Race Analyzer, and he employed the findings in his book as the basis for a computer program.

As a user of the Analyzer starts pressing buttons to feed the computer the number of each horse's first-, second- and third-place finishes this year and last, the number of lengths he was behind the leaders at four stages of his last two races, and various other figures, he will quickly become aware of the device's drawbacks.

It is time-consuming and time-wasting. A horseplayer using the Analyzer will spend at least 2o minutes feeding all the information for a single race; he will need three hours to do a nine-race program.

I suspect that if a horseplayer is going to devote three hours to the Daily Racing Form he would be better served by applying his own intelligence to it. He would be able to make observations about horses' trainers, jockeys, workouts, weight, claiming prices, ability to handle different racing surfaces -- all factors which the Analyzer does not acknowledge. He would certainly get more entertainment using his rational powers than by pressing buttoms. And he would probably pick more winners.

The Analyzer's performance suggests that strongly.

When Mattel officials first contemplated their Analyzer, they thought they had made a great breakthrough. For them, this was not going to be a cynical promotion of a worthless product. They truly believed in it. And they showed the extent of their confidence by asking the Nationwide Consumer Testing Institute -- a reputable independent laboratory -- to give the Analyzer a rigorous objective test.

Mattel expected to get results that would be the basis for a powerful advertising campaign. But when the Institute delivered its results, informed sources say the Mattel people were surprised and chagrined. They had to be. In 1,000 races, the Analyzer had selected 221 winners.

Considering that public favorites win 33 percent of the time, picking 22 percent isn't much of an achievement. It would be a tolerable result only if the Analyzer were selecting a lot of long-priced winners. But people who have used the device say it gives them mostly short-priced horses.

The most significant statistic from the 1,000-race test would be the net profit or loss from a $2 win bet on all the Analyzer's selections. The fact that Mattel never cites such a figure presumably is significant. Instead, advertisements trumpet such meaningless statistics as the fact that one out of three highest-rated horses finishes first, second or third 90 percent of the time. Probably every threadbare tout-sheet seller outside the gates of Timonium could make the same claim.

The Horse Race Analyzer occupies a unique place in the colorful history of race track hustles. Every system peddler in history has promised that he will, for a minimal price, sell an easy way to make a fortune quickly. Mattel is asking $100 for a laborious method that is documented failure. But given the racing public's insatiable appetite for well-packaged consumer frauds, the company probably will make a bundle anyway.