Defending the Indefensible 2 (NYRA Edition)

I get it – I really do. I work in pricing and I know the complexity that contracts can frequently take on. In my line of work, we had a fairly complex client contract that involved a 3rd party, wherein we would have to pay the 3rd-party in instances where their install software loaded our software onto a customer’s device. We used the 3rd-party software perhaps 30% of the time; I discovered that our accounting group was paying the 3rd party in 100% of instances. I figured I saved our company$100K/year by pointing out the error. (In a fit of gratitude, my boss told me to take the rest of the year off, having paid for my salary and then some for the whole year. Not really.)

But the truth of the matter is I actually saved our company about 1/100th of 1 cent per subscriber/year. The accounting group in charge of knowing this particular thing is counting sums so much in excess of this, with variances in the millions of dollars, that I think it would have cost our company money to train this particular handful of people to know this. I came at a bargain.

I look at NYRA overcharging some exotic bettors with a too-high take rate (26%-25%) as an example  of administrative oversight as opposed to a willful money grab. There are multiple prices (takeout splits) between the racing association and its vendors, namely the totes and the ADW services, that I can see how it happened, that the legislated agreement to lower takeout just got overlooked. Not changing is easier than changing, and it never made the top 10 list of NYRA prioritized activities.

So, now that we’ve caught it, I see the lower takeout on NYRA superexotic bets (24% vs 25/26%) as an opportunity for horseplayers. This is the perfect type of natural takeout experiment we have been looking for – an unwanted price decrease. I assume that NYRA won’t be promoting the lower takeout, but if a handle increase occurs for these pools, the association will have powerful evidence to take to the legislature to justify lower takeouts in the future.

It’s human nature to be mad about this situation, perhaps boycotting NYRA as a result. But I’m willing to support the corrective action, and in it take advantage of the opportunity to show I support lower takeouts. In 2012, I will concentrate my NYRA betting in the S-E pools, the Superfecta, Grand Slam, and Pick-Ns. I encourage you to do the same.

My Handicapping Process

Inspired by this post at Equinometry, I am writing down my handicapping process in an effort to be more consistent at betting the races. Lenny’s post – and his entire blog – is solid throughout, and touches on a lot of the same topics I want to.

My Handicapping Process

  1. Glance over whole card, find races I like to bet – features, stakes, large (8+ fields), maidens – and ones to lay off (short fields, short fields, short fields)
  2. Look at the conditions of a given race
  3. Look at the morning line odds, identify favorites, noting one or two reasons for favoritism
  4. I use Brisnet Ultimate PPs, so I’ll look at the upper right of the past performance, noting Speed Figs at Track, Surface, Distance and Recent. Looking for outliers among non-favorites can be useful
  5. Handicap the Pace Scenario. BrisPPs pre-note running style, making it pretty easy to sketch out pace scenario. Again, I like to look for outliers, esp. lone Es among Ps and Ss or lone Ss among a lot of Es
  6. Check trainer/jockey angles after this
  7. Look at the running lines, noting class of previous races especially, and look for similarities to current conditions. Also, try to figure out if the trainer has targeted this race in a meaningful fashion.
  8. Look at recent works
  9. Look at pedigree stats in maiden races; need to develop something similar for turf races
  10. Plot out betting strategy

Of all these things, I feel like I’m worst at 10, but I’m open to the idea that it’s actually 1-9 I’m bad at.


Improving the Horse Racing Data Product

Studying the Daily Racing Form the night before a day at the track was one of those customs that appealed to me as a novice horseplayer. As a self-proclaimed “stat geek”, this made sense – the numbers in the form presented a puzzle to be solved, and I ate it up. Studying the Form has actually lost some appeal to me as I’ve become a more serious player. I’ve realized two things: I’m the exception among new and potential racing fans, and I don’t get enough data.

I’ll post in the future about what I think can be done to make racing data more accessible to casual fans and bettors, those who want to improve their chance at cashing a ticket but who are not going to be handling hundreds of dollars a day at the races. These suggestions are geared to the serious bettor and can be summed up by telling the racing services (DRF, Equibase, BRISnet) to do one thing: REDUCE SEARCH COSTS.

Note: I use BRISnet in examples below since that is the service I currently use the most, and am familiar with their pricing.

Let me start off by saying I’m not afraid of a lot of data. I’ve got about 12 years experience with database development and data mining and there’s nothing I’d like more than to have access to “the database” that all PPs and the accompanying stats are generated from. I don’t foresee ever getting that kind of access at a low price, so I’m going to concentrate on making suggestions that are nearer to the current business model.

It occurs to me that a typical Past Performance Card is, in marketing parlance, a “bundle” – the single unit is the PP information for a single race. Historically, bundling PPs for all races at a single track makes perfect sense for how the races used to be bet – on the track. The majority of handle today comes from simulcasts or online thru ADWs – a racetrack PP card is not a natural bundle for many bettors.

Think of it this way – a typical day at the races bundles stakes races with claimers, maidens and allowance runners at different distances on turf or dirt or AW and with fields of 4-12 runners. A racetrack card primarily helps those bettors targeting the multi-race pools, the P3s-P6s. Many bettors – and I daresay most big bettors – have a niche. Handle numbers would lead you to believe that quality – graded stakes races, HQ meets like Saratoga, Del Mar, and Keeneland, Derby/Belmont/BC – attracts the most action. In the absence of quality, larger fields always produce good betting opportunities, which ultimately drives handle. I personally like betting maiden races, since a little pedigree knowledge goes a long way to finding a price on a horse. Some bettors like horizontals; others, vertical bets.

A bettor could, with a half hour’s work, find the set of races that he’s most attracted to do on any given day. An information service (say BRISnet) can help that bettor reduce his search costs (his time, basically) by creating bundles targeted at simulcast and online bettors.

PP Bundling Options
– Class: 10 most expensive stakes races on a given day; all stakes/allowance races on East Coast; 10 best MSWs, etc.
– Field Size: Based on current entries, the 10-12 races with the most horses entered to run on a given day
– Bet Type: 10-12 Races offering $0.50 tris, $0.10 supers – sorted by Class/Field Size; Late P4s for east coast tracks
– Surface/Distance: 10-12 Turf Races, Sprints, Two-turns, etc.
– Post Time: Races within a given simulcast window, say 2-4 EST

I think you see that the real value here is in combining these options to attract bettors to the races they like the most. Have a bundle that has all allowance-level races and above with 9 or more entries – these formful races with many opportunities to find price horses should attract big bettors, but they’re not always going to find them if they limit themselves to playing just 3 tracks.

How would one price bundles like these? Right now, the effective price for a BRIS Track PP is effectively $0.20; 10% TwinSpires share of takeout of a $2 win bet to get Ultimate PPs free. I can envision pricing these bundles for $0.10/race. For a 10-race card, that’s $1.00. Using the same math as above, a $10 daily handle would offset the price of a single bundle. Handle $20/30 and Ultimate PP “Better Bet Bundles” are free for the day. During its Breeders’ Cup promotion, TwinSpires/Brisnet showed the ability to associate higher handles with information discount and, apparently, had considerable success with that promotion.

I believe the feedback loop of driving bettors to good races and producing handle gains will lead tracks to card better races. California would have done fine this spring if field size jumped up to 9 a race like Tampa Bay, takeout hike be damned. If bundling drives more bettors to a 12-horse claimer at Turfway, then we’ll have supported the practices the sport needs. But for me, I want to identify more overlays, win more money, and this would help me do that.