11/17/2006 1:00AM

Technology woes in the news


NEW YORK - Two very different types of robotic betting led the parimutuel news this past week, each illustrating the fragility and inadequacy of the wagering machinery that fuels the $15 billion annual wagering handle that drives Thoroughbred racing.

In Florida, an investigation into a bizarrely large pool and low payoff in a pick three at Calder Oct. 28 revealed that a software glitch led to a computer's mistakenly betting nearly $40,000 on behalf of a high-rolling customer. The rogue computer made the wager in $31.74 rather than $1 units, turning an intended $1,260 play into a $39,992 investment into a pool that usually handles about $20,000. Oops. The pick three, which should have paid about $250, came back $54.40 for linking winners that paid $7, $29.20 and $4.60.

The arbitrageur had enough of the winning combo to break around even for the race, while the few normal bettors with winning tickets got a horrendously underlaid payoff. The bettor was probably not only relieved but also pleased with the result, because people who wager through these computer programs are not handicapping wizards who profit by picking winners. They are playing to break even or even lose a little parimutuelly, their profit coming from the rebates they receive on their handle.

The most disturbing thing about the incident is not that someone bet a lot of money and got lucky at the expense of regular bettors. This happens all the time when a high roller swings and hits, and it can work the other way when a big punch misses and an exotic pays more than it should. The question that lingers is what kind of arbitrage system, the foundation of which is to exploit discrepancies between odds and actual probabilities, would be robotically triggered to play the pick three, a multirace pool where possible payoffs are supposed to be kept secret until the first two legs have been run?

Officials at the tracks and betting companies involved say that no such information is made available to computer traders, and that pick three possibles can not even be calculated until money from various hubs is commingled after the close of betting. Even if that is the case here, it has been well established in the past that some high-tech bettors do receive mutuel feeds containing information not available to the general public.

In the mid-1990's, New York tracks regularly posted a mutuel feed showing the total amount of money bet in trifectas with each horse on top. It is impossible to believe that even more sophisticated analyses are not available to the best and savviest customers of account-wagering outfits that offer batch-betting and rebate services. Making sophisticated wagering data available is not an inherently bad thing, but it is incumbent on tracks to make sure it is being offered to anyone who wants it.

The week's other case of Robots Gone Bad involved an individual who has been blindly betting on all of Calvin Borel's mounts at Churchill Downs, which was working out pretty well for him, since Borel is the leading rider at the meet with a positive return on flat bets. On Thursday, however, the bettor put at least $2,000 on each of Borel's seven scheduled mounts without realizing the jockey was taking the day off to nurse a sore knee. All seven horses, most of whom could not possibly have won even with both Borel and a lengthy head start, ran up the track, providing $14,000 in free money for more rational investors.

This was good news, as is the fact that there are horseplayers out there who will bet $14,000 on a jockey's name alone. The bothersome part was the statement by Rick Smith, Churchill's director of mutuels, that the track did not have the technology to trace the wagers because "only at the exact time the wagers are made are we able to pinpoint where the bets are made, and for how much."

If true, this is a scathing indictment of parimutuel technology. Any bank or chain store has as many outlets as parimutuel betting and can trace an ATM withdrawal or store purchase to an exact machine and time. You would think the parimutuel system would be similarly equipped, and that this would be a boon to investigating suspicious wagers and payoffs.

It instead sometimes seems that racing's investigatory apparatus is hopelessly behind the times, dusting for fingerprints and staking out suspects when there should be a perfect evidence trail of any wrongdoing in the largely unexamined records of parimutuel betting. In a world where sophisticated software can analyze betting pools for high rollers' benefit, the same software should be able to uncover even the tiniest irregularities in wagering patterns. It shouldn't take major incidents like the blatantly impossible Fix Six tickets of 2002, or a robot's $40,000 mistake, to trigger investigations. Racing may well need more testing, surveillance and security, but a quicker path to uncovering wrongdoing could be simply to follow the money - starting with a better system to track it.