Baseline / Class Pars for Speed Figures

[Maverick58034]

Hey all,

I've been tinkering with creating my own set of speed figures. I'm trying to get all my ducks in a row, and because I know embarassingly little about the creation of speed figures, I've consulted some handicapping books.

The method I imagined was to first create a set of class par times by averaging the results of past races for each class and distance at the track. This was going to be my baseline. I was then going to set the baseline at an arbitrary number (e.g. 100), to which I could add or subtract points based on each races varient.

I've been going through Charles Carroll's Handicapping Speed, and from what I understand, he is very much opposed to using class pars as a baseline. On the contrary, it seems that he prefers using the fastest time set by the fastest horse (e.g. track record) for each distance.

My problem is - how does this help to create a race varient? Knowing that a horse ran a particular distance in a particular time does little to help me compare the running times of two races of identical conditions two weeks apart. How will I be able to tell if the track was running "faster" on either of the times by comparing them to the record? Why wouldn't it be better to compare both of them to the common class par? By comparing them to the track record, aren't I basically just comparing them to each other (only using the track record as an intermediary)?

While I'm here, I have a few more questions I'll throw out, and any help would be greatly appreciated:

For those that create or use speed figures, do you employ a "daily track varient" or a "race varient"? In effect, can Race 2 on a card be run "faster" than "normal," while Race 8 could be "slower"? Or, do you add up all the "fasters" and "slowers" in a card, average them, and then apply that average to each horse's time?

How many historical races are necessary to calculate an accurate class par?

While on that note, I only have a few weeks worth of results charts. Is there anywhere to get free, accurate, information-filled results charts from further back?


Thanks a million in advance

[thaskalos]

Your questions are many, and the answers are too detailed to be posted here.

I suggest you get a copy of Andy Beyer's book "Picking Winners". It will answer all your questions, and you can pick up a brand new copy from Amazon.com for $4.

[Maverick58034]

Your questions are many, and the answers are too detailed to be posted here.

I suggest you get a copy of Andy Beyer's book "Picking Winners". It will answer all your questions, and you can pick up a brand new copy from Amazon.com for $4.

Oh I've got my handy copy along side me. I guess my main issue was the use of class pars as a baseline - in the sense that (or so it seems to me) Beyer and many/most others do indeed use these, while Mr. Carroll (at least as I interpreted it) thought them to be too unreliable to adequately serve as a baseline.

[Robert Goren]

What is the point of a speed figure if you don't use a "daily track variant" ? The good thing about doing them yourself, is you know when a number might be a fluke. That said, with all the pretty good ones out there, it is probably not worth the effort, although there is something to be said for doing one for just one circuit. JMO

[Maverick58034]

What is the point of a speed figure if you don't use a "daily track variant" ? The good thing about doing them yourself, is you know when a number might be a fluke. That said, with all the pretty good ones out there, it is probably not worth the effort, although there is something to be said for doing one for just one circuit. JMO

Totally agree with you. Stemming from that, though: Is it possible that one race has a higher varient than another race on the same day, even absent extraordinary, intervening weather conditions? Put another way, could a race earlier in the day be on a "faster track" than a race later in the day? The reason I ask is that the answer to that question impacts whether to apply an entire daily varient, as an average from each race's varient, equally to all races, or instead to apply each races varient to each individual race and no others (e.g. no averaging or general application to the entire day's races).

And going one step back, in order to calculate the varient, must you use class and distance pars? While Mr. Carroll seems skeptical of their accuracy, I can't see any other way to actually calculate a variable. I can't see how the track record time (or average of 3 fastest times over X years, etc) could help determine if today's races were run on a "faster track" than yetserdays races (or, more specifically, if race 1 today was run on a "faster track" than race 3 today).

[macdiarmida]

By the questions you ask, it's clear you haven't gone through the process of calculating pars and variants. And the answers are in the Beyer books for the most part. Maybe "Beyer on Speed" might fill in a few more details about the process.

Either you believe in that daily variant, or you don't. Wrapped up in that is you have to create the best daily variant you possibly can, based on the best par charts you can make. You have to rigorous. 2yos are separate, young 3yos are out of the 3yo and up races, F&M are separate until you know what the adjustment difference is against males. 7F is not a mile is not a mile40yds. Forget the "about" distances; figs for those can be had by using the projection method.

BTW, you do grass races separately. Don't use that "magic" 7F(?) beaten lengths chart for all turf races. The finishes on grass are often very close in reality, and you screw up the ability to assess a horse going from grass to dirt and vice versa by using that bad fix. There will be a lot of pace situations that will throw off the par calculations, so throw out more extreme races when calculating pars, and use as many turf races as you can to calculate pars.

Is it possible that one race has a higher varient than another race on the same day, even absent extraordinary, intervening weather conditions?You mean the rough variant before you calculate the daily variant, I hope? It is surprising the number of days where all races line up almost perfectly. But oddball races should be rather uncommon. If calculated vs. pars charts, how does your variant correlate to (say) the top four finishers (the projection method)? If really different, check the actual race time against video; mistimed races happen often enough. If "higher" means "faster" (some people do the variant in negative numbers), then someone ran a freakishly fast race (and perhaps zero to a couple others may have freaked, but not everyone across the board). And as you might expect, 2yos and young 3yos can throw in a freakishly fast race sometimes. Plus extreme pace situations will affect variants for individual races.

Put another way, could a race earlier in the day be on a "faster track" than a race later in the day?That's actually a different question. You can sometimes have a split variant for different distances or over the day. Beyer gives examples of that. If you think you need to do that often, either the track is going to hell and Beyers will be useless for a time or you're having extreme weather changes that are so powerful that the track base is being affect - bad news in either case.

And going one step back, in order to calculate the varient, must you use class and distance pars?You can use the projection method, but both ways should be done as a check IMO. One failure that humans make is not compensating for improvement; your final figs start shrinking over time. The projection method used alone will cause this effect unless you're really good at assessing improvement in individual races. IIRC the DRF guys and Beyer add a point to all figs for each month to compensate for this. All I ever did was when calculating the daily variant, I would round daily variants like 17.5 down to 17 (therefore higher figs, reflecting slight improvement) - whole numbers is as close as you can get with confidence. It worked.

Carroll's method is based on one data point for each distance. Very common distances will be well supported by many other races. Not necessarily so for other distances. And track records can be for all time. But I think Carroll requires the track record to be fairly recent. Still, it's not necessarily the greatest horses that hold the track records, either. I thought Carroll was also more about using projection rather than par charts anyway. Tracks will do major teardowns and rebuildings of the surfaces every few years, and when you do Beyer par charts, you should only go back as far as the current surface has existed. And if your favorite track gets winterized with sand, you might think about not using those times in par chart calculations, even perhaps seeing if the winterized tracks have a different profile.

[markgoldie]

IMHO, save yourself some time and agg. by relying on the published number of reliable sources, such as Beyer, et. al.

Reason: Track variants are far too complex to be accurate anyway. They rely partially on the momentary condition of the track which is affected by weather, wind and track-maintenence procedures, all of which can change during a card. But more importantly, raw times are affected by the pace, speed matchups, inside-outside biases, etc. This is why Beyer winds up using projection numbers so often- the raw times just don't make sense a lot of the time.

And so a speed fig is and will always be an imperfect tool and should be viewed as such. Moreover, there was a time some years back when the blind faith in fig superiority allowed for some advantage in betting against the spurious high-fig horse. But the crowd these days is hip. They've learned not to fall into this trap.

In short, rather than try to "out-fig" the fig makers, you might be better off spending you valuable time and great energy in seeking other means and areas of profit opportunity.

[therussmeister]

I make speed figures without using class pars. I do this as a deliberate attempt to come up with a different answer to those speed figures based on class pars. My method is crude, clumsy, laughably amateurish, and very profitable.

[SPEEDHORSE]

In my case, I established Class Pars for Daily Track Variant computations. For Speed Figure computations I adopted Charles Carroll's method using a modified Baseline Time for every regularly ran distances in a particular track.

Instead of using track records as baseline times, I used mythical baseline times for every regularly ran distances. I started with a track record for one distance, ex. 6 furlongs = 1:06.0. Let's say the next regularly ran distance is 7 furlongs. I computed the mythical baseline time for 7 furlongs by computing the elapsed time between 6 furlongs and 7 furlongs using my Class Par Times for 6 and 7 furlongs as reference, which in this case is 12 seconds. I added 12 seconds to the Baseline Time of 1:06.0 for 6 furlongs and the mythical baseline time for 7 furlongs will be 1:18.0. Do these computations for the other distances. I guess this is the best way to make Mr. Charles Carroll's methodology to be synchronized with reality.

I hope this will help you with your Speed Figure problem....GOD BLESS !!!

[KingChas]

.
Why wouldn't it be better to compare both of them to the common par?


Take notice I took out class.
That's a start.
It's called a universal track par.Many here believe not possible.
You will be suprised how close most classes are.
But if you set up universal pars for your main betting track.
Watch those maiden & off track times though,some are throwouts.
Then -/+ shippers you are on your way.Takes a little home work.
Need universal pars for those tracks also.
Track variants must be factored in.Read some William L.Scott.,on this but tinker a little.After awhile the variant is easy to see.
Good Luck.

[Dave Schwartz]

Need universal pars for those tracks also.

Aren't "universal" pars, uh, ... universal?

[KingChas]

Aren't "universal" pars, uh, ... universal?

Dave,I knew this was coming............... :lol:
I was that cheapy offering you $20............. :D

[KingChas]

Bought pars for $8 one time.(not yours of course)
They factored in the gosh dam arabian races at Delaware Park.
Got what I paid for....never again.
Give me a 1 yr free sample an we will go from there.... :D

Universal=per track/per distance/per sex..... :kiss: