The Retro-Graded Stakes Formula

A considerable amount of racing chatter recently has been about the quality of certain graded stakes races and how their winners have been little more than (well-)paid workouts for the horses and their connections. I’m inclined to agree – top class races should attract many horses of a certain caliber but the graded stakes field size is, on average, one of the smallest in the sport. (Allowance races are not far behind – claiming and lower-level turf races attract the largest fields).

What’s at issue here is “black-type”: when a horse (or his/her progeny) go to sale, having placed in a graded-stakes race can mean a considerable premium to their auction price. This makes total sense – thousands of horses of all ages are sold each year and the bold, sometimes ALL-CAPS,  font in the sales catalog allows buyers to assess the potential class of the [yearling/two-year-old/mare/stud prospect] they are buying at a glance without reviewing a lifetime of past performances. It’s an elegant solution to a problem that existed before the Internet and electronic data was a thing, and retains some value to this day.

The American Graded Stakes Committee (AGSC) is the “be-all-and-end-all” determiner of what races get the vaunted Graded Stakes designation, those that can get the BOLD CAP font in a sales catalog. The Thoroughbred Owners and Breeders Association (TOBA) controls this designation, of which their policies can be found here. To their credit, the AGSC has been quite responsive to upgrading the designation of races that have shown considerable improvement in the quality of horses running in them over the years. The best example, given my familiarity with them, is the upgrade of the Arkansas Derby for 3 year olds to Grade 1 status and its preps (the Rebel and Southwest) to G2 and G3 status, after the likes of Smarty Jones, Afleet Alex and Curlin used the Oaklawn route to prep for later classic wins.

My main criticism of the AGSC is that, while they have been responsive to upgrades, they have been much less so to downgrading races that haven’t been as good. It’s a natural response for well-meaning decision-makers: demonstrable class deserves and upgrade, suspect class deserves just one more chance. That bias has produced what I would call class “creep” whose end result is too many graded stakes with too small fields and, frankly, too many horses earning graded black-type. The AGSC uses “gut feel” more than data to determine the top quality races, which has contributed to the bias.

The main trend driving this is the declining North American foal crop, which has shrunk from a high of 40,000 in 1990 to 25,000 this past year (Source). The number of Graded Stakes has remained steadily above 450 for the last seven years despite the falling foal crop and the number of races run in North America. This means its roughly 40% easier today to earn black-type than it was just a few years ago. The AGSC has not been responsive enough to these trends; the impact is that black-type means less and less.

The fix I propose leverage the unique power of the age in which we live – use the vast information collected about races, and the past and future performance of the horses that run in them, to determine black-type. More importantly, tie the total number of graded stakes to a reasonable estimate of the foal crops eligible for those races. Lastly, tie the earning of black-type from placing in a graded stakes not to the horse’s placing, but the number of contenders the placing horse beat to earn it. What results is what I call the Retro-Graded Stakes Formula. These are the guidelines I’d suggest:

  • In 2006, roughly 100,000 thoroughbreds (3 years of foal crops) would have been eligible for graded stakes eligibility, or roughly 1 GS for every 210 born (100K/475) . Let’s be generous and say that a GS win should be available for every 200 foals.
  • Black-type is especially valuable for fillies and mares, but their graded stakes representation is outsized compared to the open races for which they are eligible. If fillies and mares are eligible for all graded stakes, but colts, geldings, and horses are eligible for all, then gender-restricted graded stakes should represent just one third (33%) of all graded stakes – currently, 41% of Graded Stakes are gender-restricted.
  • Many graded stakes are age-restricted, so tying them to the eligible foal crop makes sense. For 2- and 3-yos in a foal crop of 24,000, that means just 120 graded races to split between the 2- and 3-year-old races, and only 40 for fillies and mares. Currently, there are 184.
  • Open company races, having a larger eligible foal crop, would get a majority of the graded stakes races. This aligns with the industry desire (supposedly) for keeping horses running at a later age.
  • Field size matters in a stakes race – it is easier to place in a 5-horse field than an 8-horse field, naturally. To earn black-type, require that a horse beat at least 60% of the field they are in. For a 4-horse race, only the winner earns black type. In a 5- to 7-horse race, top 2. Only in a field of 8 or more can an ITM guarantee black-type.
  • The total number of graded stakes would shrink to the foal crop of 3-4-5 year olds/200 (roughly 360, based on 2011-2013)  but distributing those more to open company races versus any kind of restriction. If there are 120 age-restricted races, there would be 240 without.

The above proposals are conditions that the AGSC could implement today. The biggest change, however, would be to use the past and future performances of race horses to determine the true class of a race. This would take some doing. The RGSA would assign a provisional class to a race before its run based on its current historical standing, determined by prior class of the horses in it. For example:

  • A race could be graded a PG[1,2,3], meaning a Provisional Grade 1 (2 or 3) based on the level of horses who have run in it, and their subsequent performance. Minimum purse sizes would be required – the AGSC gets this right.
  • After a suitable period of time, probably 3-6 months, the race would be graded RG[1,2,3] , again based on both the past and subsequent performance of its entries. The total number of RG races will be tied to the eligible foal crop for that race.
  • One revision to a races grade would be allowed should multiple horses from the race go on to greater things.
  • Ungraded stakes could get bumped to RG status (and future PG status) if multiple performers win subsequent RG contests.
  • Allowance level races with multiple past and future RG performers could get special “Key Race” status that could be noted in a catalog page.

I am not suggesting that the AGSC adopt these changes; though that might be ideal, it would be too radical a change. I’m saying that any group with data and sufficiently publicity could use the RGSF to challenge the status quo with regard to the class of sales horses. The AGSC has no competition – it’s time they had some.

Towards a Better Handicapping Contest – Part 2

If you asked most professional poker players what World Series of Poker (WSOP) event they would be proudest about winning, I think few would say the $10K buy-in No-Limit Hold-Em (NLHE) main event so popularized by ESPN. The large crowds and the nature of NLHE means that luck plays a much larger role in a successful outcome for those seasoned veterans. (They might say any schmo can suck out a straight on the river to beat my set, in the parlance).

I think most would say they would prefer to win the 50K HORSE event. HORSE combines, in a rotating tournament, five different types of poker that test different skills and play-styles. (That’s Hold-Em, Omaha, Razz, Stud, Stud-Eights or Better). I’m a fair Hold-Em player (positive lifetime ROI) and I can honestly say I’d have a better shot at winning a 10,000 person hold-em tournament than a 100 person HORSE tourney. I simply haven’t studies the other four games – I would very much be the dead money at the table.

Given the stakes and the multi-game skill involved, the WSOP HORSE tournament winner very rightly deserves the prize earned. I think the HORSE model is one that could be implemented for a skill-determining horse racing handicapping contest. The idea would be to combine the structures of different contests while introducing novel scoring mechanisms. Below, I have some suggestions that might get racing contests closer to that outcome.

  1. Combine the two types of Win-Place contests today, live scoring and upfront picks – if 50% of your score is determined by who you really think will win 2-4 hours in advance and the other half determined by assessing conditions, picking logical longshots, or reaching for a score, you have diminished (but certainly not eliminated) the element of luck inherent in either scenario.
  2. Utilize the win parlay – A frequent argument from contest players is that they want to be rewarded for picking winners. A parlay component would aid that. Say, for example, 20% of a hypothetical contest bankroll was bet on each race. With live scoring, contestants who picked winners early would be able to wager additional dollars on their next pick. (If you have $100 to start, bet $20 on a 2-1 horse that wins, the next round (with $160 in bank, you may bet $32). First-race losers only have 20% of $80 to bet ($16).
  3. Bowl for picks – Inspiration can be found in the most unlikely of places. Bowling (that of the ten pins and an alley) has one of the very best mechanisms for rewarding streaks and consistency: a strike adds the next two rolls to your total; a spare one. Apply the same to your handicapping contest – a win gets you 50% payout of your next two races, a place your next one (or some other percentage)
  4. Devise synthetic pick-Xs –  Allow a portion of the contest bankroll to be dedicated to picking potential winners of a set of races. If in an 8-race contest, have 2 potential Pick-4s that the contestant try to hit with a hypothetical (say $200) bankroll. This would identify skill in ticket-making (very valuable in handicapping, generally) but would lead to an interesting set of choices. Should you heavily lean on favorites for more than the minimum (say $1) or spread? The payout would be determined by a simple parlay of the four winners.
  5. Show Parlays – Again, portion the bankroll for a show parlay over all races. Rewards consistency for identifying competitive horses, but not producing huge multiples.
  6. Rolling Doubles, Pick 3s – Like options 3 and 4, rewards streaks and consistency.

Without a doubt, these scoring ideas trade simplicity for rewarding skill. To implement properly, the holder of the contest must be very thoughtful about how to weight the various elements. A lot of trial and error will probably be involved. That said, I have been tossing around a handicapping contest design around for some time. This is a first iteration, but I believe it would be a fun contest to play.

  1. 40% Weighted Upfront Picks, Win-Place-Show, Uncapped – This is designed to reward handicapping in advance of the event. The show payoffs reward identifying longshots that may figure into payouts, but may not win. Uncapped winnings reward identifying horses whose morning line do not reflect its eventual payout – a handicapping skill in and of itself.
  2. 40% Weighted Live Scoring, Win-Place, Capped (20-1, 10-1) – The traditional handicapping design. Allows players to change picks to changing conditions and longshot players to get back in, but with a lower chance of catching up to those who handicapped correctly in the first place.
  3. 20% Parlay, Live Scoring – Win, Capped (20-1) – Rewards consistency of picks in order to maintain a bankroll. Picks can be changed to reflect conditions. Multi-winners should have larger bankrolls into final races. Cap evens out impact of large longshots.

Given the ubiquity of free tools, like Google Sheets, to track this information, such a contest would not be too difficult to coordinate. It may be somewhat difficult for a 10th place contestant to figure out how exactly to bet the final race to make the top 5. You know what, that’s okay – the simple designs make it too easy.

If my work and life allows, I may inaugurate a contest like the above. Do stay tuned.

Towards a Better Handicapping Contest – Part 1

The handicapping contest in horse racing is a special animal: it offers the promise of a better return on investment (ROI) for the skilled handicapper because the parimutuel takeout does not apply or is effectively lower. A $100 entry into a contest with 100 competitors may pay out $9000 in total prize, meaning the average return for all players is 90%. $10,000 of bets into parimutuel pools may pay $7500-8500 depending on what pools are bet, exotics having higher average takeout than straight pools.

 Of course, a handicapping contest doesn’t pay out unless you finish in the top 5-10% of all competitors, and often your winning is an entry into an even larger (but more lucrative) handicapping contest. This is essentially what the National Handicapping Contest (NHC) tour is about – win a qualifying event to play for a million-dollar-plus prize. Other contests, like Derby Wars, offer the more standard contest: pool the money, winners take all.

 There are more casual events that may be played among friends or, say, a large e-mail list following a prominent racing meet. Serious handicappers should devote a portion of their handle to playing in these contests for several reasons:

  1. They may improve your ROI
  2. They encourage an in-depth look at races that can improve your parimutuel wagering
  3. They can be a tremendous amount of fun

The last is important. I understand that people, generally, gamble because they may win money but the activity of gambling itself is a form of entertainment. I will not bemoan a trip to Vegas losing a few hundred dollars if that time was spent, with friends, at a blackjack or craps table. I know the risk, I know the outcome is mainly luck; if I’ve passed the time in the company of friends, it’s money well spent.

Racing is the same – betting on the ponies is meant to be fun. I think handicapping contests capture that spirit better than most. It is true, however, that the contests themselves leave something to be desired in how the winners are ultimately decided. Let me explain.

The most popular contest today is the win-place contest. Eight to ten races across a single or multiple racing cards are selected for contestants to pick horses. For each race, a contest picks a horse and, if it finishes first or second, the contestant wins the win, place, or win+place payout on that horse toward their contest score. Trouble is, sometimes a huge longshot comes in; those that had it are almost impossible to beat in the contest. So, most contests place a cap on how much can be won – typically 19-1 to win, and 9-1 to place.

Those same contests, however, usually have scoring that is updated live, such that each contestant knows where they stand. Those in the back of the pack, trying to get into contention, will only pick longshots in the hopes that a placing will vault them back up the standings. Those in front, who might have picked legitimate chances, may find themselves losing to luck instead of skill, when the contest is mainly designed to reward skill.

Thus, the skilled handicapper says, all picks should be made in advance. A different skilled handicapper may rightly say, however, that her picks are based on odds nearer to post-time, she is picking between two horses, and besides the track for this race has shown a bias that I want to play (or it’s come off the turf, or turned sloppy). An upfront contest couldn’t anticipate such an outcome and would be as much determined by luck as the live scoring option.

You might notice that both the above options sacrifice the impact of skill (vs luck) for simplicity. It’s very easy to comprehend the rules for a win-place(-show) contest, capped or uncapped, and their simplicity I have no doubt makes them popular. Common variations like the NHC introduce more races, mandatory and optional races, and best bets (2x payoff). The Breeders Cup contest is a “live money” contest where the contestants can play anything on the BC races. If I recall correctly, however, the winners of recent ones simply made a large win bet on the BC Classic winner – while recognizing their skill, ultimately the contest came down to picking the winner of one race.

For a handicapping contest to truly identify skill (consistency in picking winners) over luck, the scoring system must ultimately increase in complexity. In my next post, I will make some suggestions to improve contests that do that very thing.