Emd and Philosphy 101

[MV McKee]

I threw together the posting below for an Emerald Downs forum at the bequest of a trainer/friend. It may not be of interest to many (does anybody else here play Emerald?).
I kind of use it to demonstate a decision I am always trying to make with the expansion of available tracks and wagering opportunities.
It is a quality quantity type if thing. Do I limit myself to one or two tracks I know like the back of my hand, or do I expand to ?? number of tracks and apply more generalized principles to them all. Probably possible to profit both ways, but I know I would never have come up with the edge I have gained in Seattle if I were playing more than 1 or 2 circuits at a time.

Any thoughts?

************************************************** **
I have not posted here before, but I wanted to try an elicit some input from serious handicappers who have had consistent success at Emerald Downs.
I have played the races on a full-time basis for nearly 20 years now. My primary venues are both California circuits (North and South), Arizona (TuP and Yav), for the past 2 years the West Virginia tracks, and, believe it or not, PM.
I play these tracks because I have been able to consistently profit over the long term from them. Although there are some subtleties that vary from track to track, I use the same basic handicapping principles at most of them.
Now, when EmD opened I very much wanted to use it as my primary venue. Before simulcasting, Longacres was my sole track. There were (and still are) a number of factors specific to the wagering opportunities on Washington racing that make it, at least to my way of thinking, a very attractive investment. With simulcasting and the influx of "outside" money, I was certain that Emerald would be perhaps the best wagering opportunity track I had ever played. This was in 1996. After the 1999 season, I abandoned Emerald. I had lost money there for 3 consecutive seasons (1st time this had happened). I very much wanted to play the track, and beat myself silly reviewing wagering records, and every conceivable factor that I felt I had perhaps overlooked, but in retrospect, it was probably ego that kept me playing that extra year.
I should probably explain a bit about my background before I launch into the next section. My education and career has been in the area of quantitative analysis, data mining, etc. So I am decidedly left-brained in my approach to endeavors such as horseracing. While I may risk being flamed for defining myself as a "numbers" handicapper, I really only do quantify the aspects of racing that can be reasonably represented numerically. It’s an accommodation for my thought patterns, and kind of acts as a "level" against which I can weigh all the subjective judgments that are necessary in handicapping.
Anyway, as luck would have it, last February a consulting job landed me in Bellevue for 6 months (the town, not the hospital). So, armed with 4 years of charts and PP data, I decided to do some mining. It took me 3 years to formulate a methodology that allowed me to have success at the Northern California tracks, so I tried to crack the Emerald nut once again. I suspected that any answers might lie in something that was not in-line with conventionally held racing wisdom, so I undertook the project determined to look at the data and the resultant correlations and regression analysis as a neophyte, and take them at face value.
Well, after a year and a half of success, I am beginning to think perhaps I am not dealing with an anomaly. I play Emerald with a very different (from most tracks) set of rules and numbers, so for a long time I had concerns that I was successful just through coincidence.

When waxing philosophical about the racing surface and it's impact on track times in a book titled "Handicapping Speed" author Charles Carroll writes "The vast majority of handicappers either ascribe some mysterious and mystical powers to the dirt or, like good scientists, argue that the untestable does not exist."
I was more in line with the latter school of thought. When calculating a track variant, I generally went with the larger sample size of the entire days race card, as opposed to splitting the variant into 2 or more separate numbers. But years of data were telling me this was the wrong approach in Auburn.
So, after this long-winded background piece, I wanted to ask a question of someone who is at the track regularly, is familiar with the maintenance routines, etc.

WHAT IS WRONG WITH THAT SURFACE?

I don't mean that in a negative way, so let me explain why I am asking. My database has the PPs and charts for every race run in North America since mid 1999. I construct a serviceable track variant (based on par times for multiple finish positions) for all tracks that run at least 30 days a year. When I say serviceable, I wouldn't use them to play a track with, but I can use them as a foundation to start projecting my own figures when I decide to play a track. I can state unequivocally that the Emerald racing surface is substantially more volatile than any other dirt surface in North America. This does not mean it is bad or dangerous, just that it varies in speed far more dramatically than most other tracks. Everyone here probably knows this, but what stunned me is how dramatically the speed changes over the course of a day's racing. Again, even though the raw data indicated this was the case, I would never have actually made statements like this unless I had somewhat validated the theory through practice.
But there is another theory that I have been successful with that I would like some input on. It seems very odd, but very often there seems to be a migrating lane bias. Again, for years (1996-1999) I was caught in a trap, early in the card I would suspect a strong inside or outside lane bias, only to see it wane just as I recognized, then, on occasion, the opposite bias would appear later in the card. Because a single bias rarely persisted over the course of an entire race card, I would consistently notate no bias, even though I had suspicions otherwise. I had no success.
The past 2 years, I have on 62 separate race cards noted an inside or outside bias for portions of a race card, sometimes "flip-flopping" the two, and handicapped the races as if the bias definitely existed, and I have had tremendous success with it. Even though I am a bit uncomfortable, the way the horses "run back" has done nothing but validate what I think are some very oddball theories. For those of you that are statistically minded, here are the raw numbers:

For track variants:
Standard deviation is an indicator of spread, simply put; on average how many 1/5ths of a second does the track speed vary on a daily and/or race-to-race basis. It's a big gap from 1st to second. This is for fast tracks only, some winter tracks have would larger numbers due to inclement weather.

Standard deviation of track variant:
1)EmD: 5.98
2)Lrl: 3.71
Avg of all other North American tracks:
2.44

Lane Bias:
This is kind of an arcane calculation. It applies to both the PostPos and the running lane of a given horse. It is then related to the field size (for Post) and the # of horses on each given lane (for running lane). It relates the Post Pos or running lane of the winner to the 2nd place runner and (depending on field size)
the 3rd place runner. Normally, and as is the case for all but a handful of tracks in North America, one would expect to see a particular spread between the winners PostPos and Running Lane that was related to field size. In other words, a track with larger fields would have a larger average spread between the PostPos or Running Lane. Again, EmD has one of the lowest spreads in the nation, and the LOWEST of any mile circumference track. Because I did not explain this very well, what this means is that at EmD in a race that is won by a horse that leaves the #1 Post position or runs on the rail, the 2nd place horse also runs on the inside lane or exits an inside post far more often that is the norm. By the way, the same exact thing applies to horses running wide, etc. Far more often than at other tracks, the exacta and trifecta horses all ran on the same part of the track.

Sorry to go-on for so long, but I am hoping someone else out here perhaps has noticed or hypothesized the same thing. I am a bit stubborn, and didn't really believe what my instincts told me was happening until I could quantify it.

Not certain why this happens, but it is probably because Emerald has an ill-advised practice of changing the track surface at mid season. I understand why they do it (don't agree with the logic), but the material never gets a chance to break down and become a true, forgiving racing surface. Most other tracks have learned this lesson the hard way, perhaps EmD will too.
************************************************** **

[Bobby]

GOOD post. unforunately, i can't help you though.

[kenwoodall]

I do not play ED much, but fascinated by your idea of radically changing biases! If true, has to be water table or jockeys who know the biases. I am interested if you play certain jockeys more than others.
I use works for variant, not a fan of final times. Do you have pars on only place finishers?

[MV McKee]

Actually I have pars for 1st through 4th place finishers (depending on field size), but do not use them to calculate variants unless I have never played a track before.
The other thing that is somewhat unique about Emerald is that it is what I categorize as an "encapsulated" race meet. in other words, I define a race meet as "encapsulated" if >1/3rd of the horses making their 1st start of the meet made their most recent (prior) start during the previous years race meet. I categorize all the major California meets as having a perpetual racing calendar as the above condition is not met.
The primary reason I make this distinction is because the effects of the N2L, N2X and other restrictions on a claiming race are radically different at a bookended (encapsulated) racemeet than they are on perpetual circuits. Lots of dropping and popping near the end of these race meets makes the low-end restricted claimers very strong near the end of a race meet and (by way of comparison) weak at the beginning. The high end claiming races tend to do exactly the opposite. The final 5 weeks of the Emerald and Hastings Park meetings are among the most profitable on my calendar. In California, there really is no beginning and end to the season. I do see some fluctuation to a lesser degree near the onset of the fair season in NoCal, and the well documented emerging trend in SoCal that Hollypark is indeed a notch below the other two major tracks. But at EmD and Hst, the fluctuation is very dramatic. No owner or trainer wants to lose potential in April, and no trainer wants to send an owner into the 6 month off season without at least one bank deposit during the previous season, be it purse money or claim money. There are a lot of tired, overraced horses at the bottom level during the late season, but the ones that win and finish in the money are often a level or two better than the standard for that claiming price.
A standardized, calendar based universal par adjustment for 3yo races, or for an N2L or N2X simply doesn't work at these race meets. Yet another dynamic that makes this game so damn great!

I also heard someone in Seattle mention the water table and tides as affecting the track surface (Emerald sits on a tidal wetlands). Would be interesting to see if there is a correlation. Feel odd adding a tide table to my handicapping arsenal, but if it works, what the hey.

[Dave Schwartz]

Michael,

As you probably know, I make pars for all tracks as well.

I do not delve as deeply into a track as you do EMD... I simply cannot put that kind of time into it. ABout all I can spend is 30 minutes per track.

Some tracks are a breeze and are done in 15-20 minutes, while others (GP and EMD immediately come to mind) take 5 times that long.

Your explanation of the par situation at EMD is very plausible.

What I have found is that the distances relate "funny." This seems to be traceable to the huge difference between pace times for short sprints versus the longer sprints, and short routs versus longer routes. It is, very unnatural.

As an aside, my good friend, Ken Klein, uses my par times for his pencil and paper approach and plays nothing but EMD these days. He has commented that my pars are dead on and he is just killing them.

However, after giving me the great compliment, he followed up by saying that the 1 1/16 par is consistently slow, causing the horses to get exceptionally high ratings. It is on my agenda to look at this week. (So, I guess my pars are great except when they aren't. <G>)

Anyway, thanks for the insight into this situation. Perhaps I need to consider a dual par for EMD.


Regards,
Dave Schwartz

[MV McKee]

Dave,

I am certain I can give you some insights into this "funny" distance relationship situation, it had me stumped for a long time.
When the track variant at EmD changes, it does so very much within the stamina part of it.
I have mentioned before that a variant (IMO) reflects a racing surfaces effect on a horse's stamina, as well as the actual "speed" of the surface (as if a car were racing on it). At Emerald, the chute (6f and 6.5) races is not maintained during the race card in the same manner as the main track. The <6 furlong races are almost always run in the early part of the cards. Very very often the track will dramatically speed up, or less often slow down during the course of a nightime card. Because Emerald tends to run particular types of races on weekday evenings, and other (generally better) races on weekend day cards, certain classes and distances are almost always run on (relatively) faster or slower tracks. Additionally, EmD runs such a huge variety of race condition/ distance combinations that it is hard to get a sample size large enough to overcome the surface volatility and come up with a reasonable par for any given class level.
The other problem I alluded to earlier has to do with when the track surface gets extremely fast. When this happens, you can commit pace suicide and not tire. This usually happens every Longacres Mile day when they pave the track. On these days, the time difference between distances shrinks dramatically. 1:10 for 6 furlongs will equal 1:15.8 for 6 1/2. 1:37 for a mile will be 1:43 for 1 1/16. There is almost no deceleration on these days. In the early season, when almost all of the 5 and 5 1/2 furlong races are run, the racing surface is entirely different (they literally revamp the surface every year in late june). So if you are trying to compare 5 or 5 1/2 furlong races to 6 and 6 1/2 races based on raw times it may be misleading...they are essentially run over 2 completely different tracks. Also, the older upper claiming levels (25k and above) are the same horses. The exact same cast will be assembled for various levels. In fact in a 6 week period, a group of the same 4 horses (plus a few new faces) competed in a 25K-N2X, a 40K open and a 50K open sprint. The same 2 horses completed the exacta in the 50K open and the 25K-N2X sprints. May not want to put too much stock in any par over the 10k level.
Quite honestly, it was impossible for me to construct a par set before applying a variant at EmD. Kind of a cart before the horse thing.
Hope this helps.

Thanks,

Michael

[Dave Schwartz]

Mike,

Actually, it makes a lot of sense.

Thanks for taking the time.

Dave

[delayjf]

Mike,

I may offer a theory that I mentioned the other day that might apply to EMD, again not original to me but Davidowitz. He concluded that on overcast days, the track at Delmar would not dry out properly which would result in the infamous outside closers bias that would seem to randomly appear from time to time at Delmar. Could it be that given the nature of EMD's track and the fact that you get so much more WX than SoCal that this same situation not only exist at EMD but in a more volitile manner.

Another plausable reason for the dramatic lane bias changes might have to do with the maintense trucks and the lanes that they tend to pass over as they harrow/water the tracks. Some on this board believe that when the maint crews change the direction that they work the track (clockwise to counter-clockwise and vise versa) will have the effect of reversing a bias.

Again the nature of EMD due to all the WX it must endure may render it more suseptable to any number of influences, so that the effects may be more accute.

p.s I sent you a couple of e-mails a while back, not sure if you got them or if you'er just too busy to respond, if so, no problem perhaps some other time.

[VetScratch]

MV MCKee,

Standard deviation of track variant:
1)EmD: 5.98
2)Lrl: 3.71
Avg of all other North American tracks: 2.44

I also heard someone in Seattle mention the water table and tides as affecting the track surface (Emerald sits on a tidal wetlands). Would be interesting to see if there is a correlation. Feel odd adding a tide table to my handicapping arsenal, but if it works, what the hey.

Great postings!

FWIW, some years ago in a newspaper column, Beyer remarked that LRL was a difficult challenge for his methods because of tidal effects.

This was before they built EMD. You now have LRL ranked second behind EMD with respect to standard deviation.

Tides don't address or explain all of your findings and observations, but I bet they play a role over the span of a full racecard.

[MV McKee]

Vet Scratch,

Thank you much for that insight. I had no idea Laurel was on a tidal plain. That fact alone may make this tide thing worth exploring. Hard to imagine.

[Amazin]

MV

I would think that if the variant was as wild at EMD as you say,the average mutuel at that track should also be among the highest in racetracks accross the nation.I don't think this is the case.So if what you say about EMD is true(surface variant changes on the widest spectrum in U.S.) then I don't understand how the perdictability is normal.

What I really like about EMD is that I find it easier to handicap than most tracks especially eastern tracks.Reason is most of the horses PP's are from EMD and not invaders from another track.I don't have to do much track to track adjusting there.Maybe that's why it's predictability rate is in the normal range with an overoscilatting variant.

I think the approach of 1 track verses many is the more profitable method.The focusing of attention on details turns to profits.Right now I am totally concentrated on DelMar.And after one week I was able to find a pattern in their turf races that has led to a breakthrough for predictability in those races.This is the power of concentration you can't achieve when you spread yourself too thin.

[Amazin]

Just checked the average mutuel with Bris.Interestingly Laurel is the lowest paying track @ 3.38-1 average winning mutuel.Then comes Saratoga @ 4.34-1.In third lowest is EMD @ 4.44-1.so apparently no one is fooled by the variant.

DMR is 6.30-1 average winning odds.Personally If I am going to concentrate on a track I would rather concentrate on a track that pays.Dmr variants are probably more uniform,but rewards are higher.

So no disrespect but why study a wildly fluctuating track variant involving twice as much work and get paid less than a track where you do less work but get paid more?

[VetScratch]

Amazin,

With respect to looking for good risk/reward/workload opportunities, I have always wondered why BRIS, which offers so many other adjusted (value-added) factors, does not offer an adjusted figure for average mutuel, taking into account takeout variations and field sizes.

Does anyone offer such statistics?