Making Speed Figs: charts & the value of a beaten length

[PhantomOnTour]

Begin with the given that our speed charts and class pars are accurate.

A) how do you break down your charts for pace and speed?
1/5 sec? 1/10 sec? 1/100 sec? other?

in other words: does it go from 46.00 to 46.20...46.10...46.01?
same with final time: 111.00...111.20...111.10...111.01?

is it a different breakdown for pace vs speed? sprints vs routes?
(some do pace in 1/10 sec and final time in 1/5 sec)

B) what is your value of a beaten length, and does that vary by distance?

C) if you make figs for Trakus tracks are you just ignoring beaten lengths and going by the horses' "true time" from their charts ?

At some point in ones figure making life one inevitably tools around and tweaks A) and B) in an effort to refine or simplify or whatever.
I apologize for not posting BRIS and Beyer beaten length data, as it's readily available, and would be good for comparison with anyone willing to contribute....i will try to find and post that info soon.

[PhantomOnTour]

I started this thread mainly to begin a discussion on beaten lengths (BL).
All of this may be mute to you if you trust Trackus and ignore BL, or if you just don't trust the BL margins anywhere but the finish.

If that's the case, read no further :D

I'm in that tweaking mode and considering a bit of a new concept beginning with the Bel spring meet.
I am trying to design my charts so that each increment represents the amount of time it takes to make up one BL at each specific distance. Obviously each distance (from 2f to 6f to etc...) will have a different value because horses are going faster early and slower late in just about every dirt race in America.
Some distances would be grouped together:
5.5 and 6f
6.5 and 7f
1m - 1m40 - 1m70 - 1m16

It takes less actual time to make up 2 BL thru a 1/4m in 22.10 than it does during a 1/4m in 24.80 (but not much less)-nonetheless, i want my charts to reflect that.
Why?
So that 1BL at every distance will be worth 1pt.
If the charts is accurately put together can't i adopt the 1BL=1pt policy?

Bunch of sharp fig guys here and i ask these questions to perhaps stop myself from a large project of futility :D
I understand if some don't want to reveal how they design their charts and what they assign to 1BL.

[Magister Ludi]

Begin with the given that our speed charts and class pars are accurate.

A) how do you break down your charts for pace and speed?
1/5 sec? 1/10 sec? 1/100 sec? other? I don't use pace and speed charts.

in other words: does it go from 46.00 to 46.20...46.10...46.01?
same with final time: 111.00...111.20...111.10...111.01?

is it a different breakdown for pace vs speed? sprints vs routes?
(some do pace in 1/10 sec and final time in 1/5 sec)

B) what is your value of a beaten length, and does that vary by distance? by time and distance - see below

C) if you make figs for Trakus tracks are you just ignoring beaten lengths and going by the horses' "true time" from their charts ? Trakus is the only method that could be accurate.

B) t + nlt/D
where n = number of beaten lengths
l = length of beaten length (feet)
t = time (seconds)
D = distance (feet)

example:
n = 5
l = 8
t = 24
D = 1320

24 + (5*8*24)/1320 = 24.73

[Robert Goren]

A beat length at finish is product of the photo finish camera. Tell me how fast they pull the film strip through the camera and I will tell you what a length is term of time. Somebody here should have that answer for you.

[PhantomOnTour]

B) t + nlt/D
where n = number of beaten lengths
l = length of beaten length (feet)
t = time (seconds)
D = distance (feet)

example:
n = 5
l = 8
t = 24
D = 1320

24 + (5*8*24)/1320 = 24.73


so, 5BL = 0.73 which means it took that horse ~0.146 secs to make up each length

this is what i'm getting at...
i make a 2f pace fig in sprints
let's say the 2f par for Clm10 going 6f is 22.60, which is ~0.137 secs per length...why not build a 2f chart using 0.137 as your increments?

Time - Rating
22.05 - 104
22.19 - 103
22.33 - 102
22.46 - 101
22.60 - 100
22.73 - 99
22.84 - 98
23.01 - 97
23.15 - 96
23.29 - 95
23.42 - 94

etc etc etc...now 1BL = 1pt, right?

[Capper Al]

so, 5BL = 0.73 which means it took that horse ~0.146 secs to make up each length

this is what i'm getting at...
i make a 2f pace fig in sprints
let's say the 2f par for Clm10 going 6f is 22.60, which is ~0.137 secs per length...why not build a 2f chart using 0.137 as your increments?

Time - Rating
22.05 - 104
22.19 - 103
22.33 - 102
22.46 - 101
22.60 - 100
22.73 - 99
22.84 - 98
23.01 - 97
23.15 - 96
23.29 - 95
23.42 - 94

etc etc etc...now 1BL = 1pt, right?

One sixth or one seventh of a second per beaten length is more accurate and it does vary with the distance of the race, but don't sweat it. It won't effect your win rate over the long run. You could just as well use a flat fifth of second per beaten length like the old paper and pencil handicappers did. It works just as well and you can do the math in your head.

[RaceBookJoe]

One sixth or one seventh of a second per beaten length is more accurate and it does vary with the distance of the race, but don't sweat it. It won't effect your win rate over the long run. You could just as well use a flat fifth of second per beaten length like the old paper and pencil handicappers did. It works just as well and you can do the math in your head.

Probably not as technical or exact, but thats what i do. I have used 1/5th second per length roughly for almost 40yrs. Even though its not perfect, it fits me fine but i do know that it will cause me losses from time to time and i accept it. I will say that I usually look for a wider spread in speed/pace advantage to hopefully compensate for my caveman approach.

[Tom]

etc etc etc...now 1BL = 1pt, right?

I do something like that.....makes good sense.
I'll post my method tomorrow if I get time. I come up with a chart that pits the three call times and third fraction POR on the same scale.

Doing my taxes right now.....then I plan to get very, very drunk as soon as hit "SEND." :mad: :(

[Dahoss2002]

so, 5BL = 0.73 which means it took that horse ~0.146 secs to make up each length

this is what i'm getting at...
i make a 2f pace fig in sprints
let's say the 2f par for Clm10 going 6f is 22.60, which is ~0.137 secs per length...why not build a 2f chart using 0.137 as your increments?

Time - Rating
22.05 - 104
22.19 - 103
22.33 - 102
22.46 - 101
22.60 - 100
22.73 - 99
22.84 - 98
23.01 - 97
23.15 - 96
23.29 - 95
23.42 - 94

etc etc etc...now 1BL = 1pt, right?
When I used to make my own I used 1 pt per length. Not perfect but when I see A 3 PT difference, I want it to be just that.

[Augenj]

One sixth or one seventh of a second per beaten length is more accurate and it does vary with the distance of the race, but don't sweat it. It won't effect your win rate over the long run. You could just as well use a flat fifth of second per beaten length like the old paper and pencil handicappers did. It works just as well and you can do the math in your head.
That's pretty close to what I found in a Bell curve by computer with thousands of races. That value is 0.134 of a second for lengths when adding speed and variant together to get a past performance speed figure for calculating races.

[Valuist]

I find it interesting they always round down. :46.19 is considered :46, when it should be 46 1/5. I'll round 46.14 to 46.10 but 46.15 to 46 1/5 for my pace figures. Still use 1/5ths for speed figures.

[PhantomOnTour]

So what about final time beaten lengths?

the crucial measurement in determining (or best estimating, without Trakus) the final time of non winners is the time of the final fraction (FF) and not the final time of the race itself.

Let's use a 6f race and a 1m1/8 race.
On average, the FF of a 6f race will be run in over 25.00...I have settled on 25.30 secs, and that equals ~ 0.153 lengths per second.
On average, the FF of a 1m1/8 race will be around 38.00 secs...I have settled on 38.60 secs, and that equals ~ 0.156 lengths per second.
They're not going much slower at the end of a 9f race than they are at the end of a 6f race.

Using the 0.137 from earlier in the thread for 2f we can fill in what we need.
2f = 0.137
4f = 0.143
6f = 0.153
7f = 0.153
1m = 0.155
9f = 0.156

As of now i have 3 distance groupings for final time:
5.5f - 7f
7.5f - 9f
over 9f

**for the 6f pace call in routes i would use the 4f BL number of 0.143 instead of the 6f final time BL number of 0.153

Some of y'all (like Raybo i believe) may use the actual FF of each race to get accurate times for non winners instead of the averages i am considering above. Maybe you do it for fractional calls too.
It's the optimum way imo but it's time consuming....and yes, every fig maker will readily admit that they must use an average or an estimate somewhere along the line.
Are these differences so miniscule that it's not worth the effort to build charts reflecting that, and instead go with a standard sprint & route BL multiplier?

[raybo]

B) t + nlt/D
where n = number of beaten lengths
l = length of beaten length (feet)
t = time (seconds)
D = distance (feet)

example:
n = 5
l = 8
t = 24
D = 1320

24 + (5*8*24)/1320 = 24.73



That's an example of a "variable time per beaten length", by race distance in feet. I do something very similar for my fractional adjusted times. The only thing you are missing that I use, is the track variant. I use a fractional variant, again by fractional distance in feet, for my adjusted fractional times.

[raybo]

So what about final time beaten lengths?

the crucial measurement in determining (or best estimating, without Trakus) the final time of non winners is the time of the final fraction (FF) and not the final time of the race itself.

Let's use a 6f race and a 1m1/8 race.
On average, the FF of a 6f race will be run in over 25.00...I have settled on 25.30 secs, and that equals ~ 0.153 lengths per second.
On average, the FF of a 1m1/8 race will be around 38.00 secs...I have settled on 38.60 secs, and that equals ~ 0.156 lengths per second.
They're not going much slower at the end of a 9f race than they are at the end of a 6f race.

Using the 0.137 from earlier in the thread for 2f we can fill in what we need.
2f = 0.137
4f = 0.143
6f = 0.153
7f = 0.153
1m = 0.155
9f = 0.156

As of now i have 3 distance groupings for final time:
5.5f - 7f
7.5f - 9f
over 9f

**for the 6f pace call in routes i would use the 4f BL number of 0.143 instead of the 6f final time BL number of 0.153

Some of y'all (like Raybo i believe) may use the actual FF of each race to get accurate times for non winners instead of the averages i am considering above. Maybe you do it for fractional calls too.
It's the optimum way imo but it's time consuming....and yes, every fig maker will readily admit that they must use an average or an estimate somewhere along the line.
Are these differences so miniscule that it's not worth the effort to build charts reflecting that, and instead go with a standard sprint & route BL multiplier?

Actually, I calculate my adjusted fractional times first (using variable beaten lengths multipliers, and variable fractional variant multipliers), then add them up to get the adjusted race time.

[HUSKER55]

Have you guys noticed that when the time is posted the time will be (for example), 46 3/5. and the decimal time is 46.71. that is half a length using the 5 lengths per second.

I don't think it matters UNLESS you mix time standards. Use decimal or fifths or what ever you like just make sure you don't mix them up.

I don't think it is wise to mix BRIS final time ratings with Beyer. same reason, how do you know if the math is the same.


JMHO

[raybo]

Have you guys noticed that when the time is posted the time will be (for example), 46 3/5. and the decimal time is 46.71. that is half a length using the 5 lengths per second.

I don't think it matters UNLESS you mix time standards. Use decimal or fifths or what ever you like just make sure you don't mix them up.

I don't think it is wise to mix BRIS final time ratings with Beyer. same reason, how do you know if the math is the same.


JMHO

I agree, there is quite a difference between "fifths" and decimal. Of course, many here will say it doesn't matter because the lengths are inaccurate anyway, but, I use decimal because they are more accurate than fifths, because of the rounding that occurs with fifths being more dramatic than using hundredths of a second. Again, I try to err on the more accurate side, not just go along with what has been used for decades as the standard. I suppose there are individual scenarios where the standard might actually be better, for the results of an individual event, but individual events mean very little, in the long term. I truly believe that, long term, being more accurate in your calculations will be more predictive long term.

And, of course, whichever method you use, do not try to mix them, that will just compound the inherent, existing inaccuracies of our publicly available data, in the long term results we achieve.

[Capper Al]

I'm assuming anyone working with fifths knows enough to use .20 for their decimal.

[raybo]

I'm assuming anyone working with fifths knows enough to use .20 for their decimal.

Sure, but that's not the problem we were talking about, in mixing the 2 different time formats. A time of 46.60 (converting fifths to decimal) is still not the same as 46.71 (which is the actual time), as was the example in the earlier post (46 3/5 versus 46.71).

[Robert Goren]

If you going to picky over the value of a length be sure to use the actual time of the race, not the time in fifths.

[raybo]

I think my "perfectionist" problem is a good reason for my not getting too involved in these kinds of threads. I guess the old saying that my dad drilled into me while coaching me in little league baseball, "If it's worth doing, it's worth doing right", doesn't lend itself to discussions like this.

Obviously everyone is going to do what they are comfortable doing anyway.

[Cratos]

Begin with the given that our speed charts and class pars are accurate.

A) how do you break down your charts for pace and speed?
1/5 sec? 1/10 sec? 1/100 sec? other?

in other words: does it go from 46.00 to 46.20...46.10...46.01?
same with final time: 111.00...111.20...111.10...111.01?

is it a different breakdown for pace vs speed? sprints vs routes?
(some do pace in 1/10 sec and final time in 1/5 sec)

B) what is your value of a beaten length, and does that vary by distance?

C) if you make figs for Trakus tracks are you just ignoring beaten lengths and going by the horses' "true time" from their charts ?

At some point in ones figure making life one inevitably tools around and tweaks A) and B) in an effort to refine or simplify or whatever.
I apologize for not posting BRIS and Beyer beaten length data, as it's readily available, and would be good for comparison with anyone willing to contribute....i will try to find and post that info soon.


The length behind calculation is an anomaly with many handicappers and sometimes poses a conundrum because its metric definition is typically somnolent.

However let’s bring the metric definition alive before attempting to calculate the lengths behind in terms of time.

A length equals 9 feet or at least that what was posted on the DRF’s website and printed in the Racing Form for many years. Joe Hirsch who was a columnist at DRF once published the following chart for neck, head, and nose margins:

1 length = 4 necks
1 length = 8 heads
1 length = 16 noses

How true or reliable were those metrics? I don’t know, but they provided a reasonable guideline when converted into distance using 1 length equals 9 feet.

1 neck = 2.25 feet
1 head = 1.13 feet
1 nose = 6.75 inches

How accurate are those metrics? Accuracy should not the gating item here; it should be statistical significance. Accuracy is predetermined by the chart caller. If he/she sees the margin as 4 ½ lengths as oppose to the actual 5 lengths then that is the printed metric and that is okay with me because I see the chart callers as experienced professionals. Yes, sometimes they do make a mistake; but don’t we all?

In all calculations the length distance metric is invariable; what is changing is time.

Calculation

(Distance in feet) minus (Lengths behind in feet) divided by speed (feet/one-fifth second) divided by 5.

Let’s look at an example:

Horse D ran the quarter-mile in 22.34 seconds

At the quarter-mile these are the horses behind Horse D:

Horse C behind by ½ lengths
Horse A behind by 3 lengths
Horse E behind by 3½ lengths
Horse B behind by 6 lengths
Horse F behind by 8 lengths

What are the times of the horses behind?

From the formula we get

Distance = 1320 feet
Speed = 1320/111.70 = 11.81737 Ft/one-fifth second; rounded to 11.82

Horse C
1320+4.5/11.82 = 112.056/5 = 22.41

Horse A
1320+27/11.82 = 113.959/5 = 22.79

Horse E
1320+31.5/11.82 = 114.340/5 = 22.87

Horse B
1320+54/11.82 = 116.244/5 = 23.25

Horse F
1320+72/11.82 = 117.767/5 = 23.55

I disregarded accuracy because what is not known is the actual number of feet travelled by each horse. Therefore the assumption is that each horse will travel the quarter in 1,320 feet, but we know that is not true because if the distance is on a turn some horses will be running on a larger radius; hence more distance.

But for calculation sake, accuracy is not considered, but statistical significance is.

The question becomes not how accurate was the 22.34 and 22.41 metrics between Horse D and Horse C? Was the difference of .07 seconds statistical significant?

What statistical significance means is that the .07 difference is reliable; it doesn’t mean that it is important. To go into significance testing you can do what is called a t-test significance, but that is for another post.

[HUSKER55]

d/t=rate (velocity)

660'/12 seconds is 55 feet per second

if you use 5 then a length is 55/5 = 11
if you use 5.5 then a length is 55/5.5 = 10
if you use 6 then each length is 55/6 = 9.17

like I said, I don't think the number of lengths is as big a deal as how fast. I was told in another thread, [and they were correct] 2 lengths going 24 is not the same as 2 lengths at 22.

it is real easy to get numbers mixed up.

I otta know....I do it every day!:D

[Cratos]

d/t=rate (velocity)

660'/12 seconds is 55 feet per second

if you use 5 then a length is 55/5 = 11
if you use 5.5 then a length is 55/5.5 = 10
if you use 6 then each length is 55/6 = 9.17

like I said, I don't think the number of lengths is as big a deal as how fast. I was told in another thread, [and they were correct] 2 lengths going 24 is not the same as 2 lengths at 22.

it is real easy to get numbers mixed up.

I otta know....I do it every day!:D

You are correct with this statement: “I was told in another thread, [and they were correct] 2 lengths going 24 is not the same as 2 lengths at 22.”
However you are incorrect in your varying of length metrics. If the length metric is varying and the time is varying, what is the unique answer? The answer is, it doesn’t exist.

A length metric should be interpreted similar to a mile metric. A mile is 5,280 feet and it doesn’t matter how fast a horse runs; a length metric is not any different. It is a fixed distance and it too doesn’t matter how fast a horse runs.

As I stated in my earlier post, the length metric is determined by the chart caller. For example, if the distance between two horses running in a race is 10 feet and the accepted length metric is 2 feet that is okay or if the accepted length metric is 5 feet that would be okay too because the time calculated would not be any difference; just the time increments would be.

[Tom]

It's 2013.....when is this industry going to come out of the 18th century?
There is no reason to be talking lengths at all.

Did you ever see a tack and field result where one guys won by 3 lengths?
The actual times the horses run and the actual distance they cover is what is, IMHO, the only thing acceptable.

But alas, in the world of horse racing, I thinking they still burn witches and
believe the world is flat.

We really play a game run by extreme underachievers and dolts.:mad:

[Cratos]

It's 2013.....when is this industry going to come out of the 18th century?
There is no reason to be talking lengths at all.

Did you ever see a tack and field result where one guys won by 3 lengths?
The actual times the horses run and the actual distance they cover is what is, IMHO, the only thing acceptable.

But alas, in the world of horse racing, I thinking they still burn witches and
believe the world is flat.

We really play a game run by extreme underachievers and dolts.:mad:


Finally, a voice of reason and sensibility; Tom, you are absolutely right; the length metric should be banished from horseracing data going forward forever.

Trakus is on the right track and it is puzzling to me why DRF, Equibase, and all of the racetracks haven’t installed that system or a similar system for their data collecting not unless it is cost prohibitive at this point.

Sensor and GPS technology have been around for a long time.

[raybo]

d/t=rate (velocity)

660'/12 seconds is 55 feet per second

if you use 5 then a length is 55/5 = 11
if you use 5.5 then a length is 55/5.5 = 10
if you use 6 then each length is 55/6 = 9.17

like I said, I don't think the number of lengths is as big a deal as how fast. I was told in another thread, [and they were correct] 2 lengths going 24 is not the same as 2 lengths at 22.

it is real easy to get numbers mixed up.

I otta know....I do it every day!:D

Exactly! You must consider the speed the horses are traveling.

[raybo]

You are correct with this statement: “I was told in another thread, [and they were correct] 2 lengths going 24 is not the same as 2 lengths at 22.”
However you are incorrect in your varying of length metrics. If the length metric is varying and the time is varying, what is the unique answer? The answer is, it doesn’t exist.

A length metric should be interpreted similar to a mile metric. A mile is 5,280 feet and it doesn’t matter how fast a horse runs; a length metric is not any different. It is a fixed distance and it too doesn’t matter how fast a horse runs.

As I stated in my earlier post, the length metric is determined by the chart caller. For example, if the distance between two horses running in a race is 10 feet and the accepted length metric is 2 feet that is okay or if the accepted length metric is 5 feet that would be okay too because the time calculated would not be any difference; just the time increments would be.

Yes, we are interested in times, not lengths, and if you don't consider the speed of the horses then your times will be more inaccurate than is already included in the chart callers lengths.

[PhantomOnTour]

Finally, a voice of reason and sensibility; Tom, you are absolutely right; the length metric should be banished from horseracing data going forward forever.

Trakus is on the right track and it is puzzling to me why DRF, Equibase, and all of the racetracks haven’t installed that system or a similar system for their data collecting not unless it is cost prohibitive at this point.

Sensor and GPS technology have been around for a long time.
Exactly - Trakus (or similar technology) at every track would end this whole discussion & give us actual times for each horse instead of BL and leaving the math to us.

[Tom]

Don't hold your breath.....
It took 10 years to get tracks to agree on saddle cloth colors.
And still Saratoga doesn't do it all the time. :rolleyes:

[raybo]

While it would be nice to have exact fractional times for every horse, in every race, at every track, but that ain't never gonna happen. Is it? Not in any of our lifetimes, even the younger members here. There is absolutely no way, all the tracks, are going to do this, and we would need all of them to do it, or it wouldn't change much. Some would others wouldn't, that would open up another can of worms we'd have to deal with, wouldn't it? Yeah, I think so, and my plate is plenty full, of stuff I have to deal with, already.

Plus, I know I already have an advantage over almost every other player in the game, because of the way I do these things. So, it's ok with me if it doesn't change, and that's a pretty good chance. Would I like for the beaten lengths to be more accurate, absolutely, because my advantage would be even higher.

That's my opinion, and I'm ok with it.

[PhantomOnTour]

While it would be nice to have exact fractional times for every horse, in every race, at every track, but that ain't never gonna happen. Is it? Not in any of our lifetimes, even the younger members here. There is absolutely no way, all the tracks, are going to do this, and we would need all of them to do it, or it wouldn't change much. Some would others wouldn't, that would open up another can of worms we'd have to deal with, wouldn't it? Yeah, I think so, and my plate is plenty full, of stuff I have to deal with, already.

Plus, I know I already have an advantage over almost every other player in the game, because of the way I do these things. So, it's ok with me if it doesn't change, and that's a pretty good chance. Would I like for the beaten lengths to be more accurate, absolutely, because my advantage would be even higher.

That's my opinion, and I'm ok with it.
You got that right - i just can't imagine Evangeline Downs using Trakus

Once you have all that stuff (BL multipliers, or whatever you call it) in place it isn't really tedious at all. Each 3/16m, 1/4m, 5/16m and 3/8m time will have a BL multiplier assigned to it.
And, imo (apparently Raybo's as well) it's as accurate as we can hope to be with what we've got.

[raybo]

You got that right - i just can't imagine Evangeline Downs using Trakus

Once you have all that stuff (BL multipliers, or whatever you call it) in place it isn't really tedious at all. Each 3/16m, 1/4m, 5/16m and 3/8m time will have a BL multiplier assigned to it.
And, imo (apparently Raybo's as well) it's as accurate as we can hope to be with what we've got.

Yeah, plus what about all the other tracks that are still running only because they are part of a casino multiplex. Do you think those shareholders are going to like their money being spent on a new timing system, for horse racing, the part of the multiplex that produces the least profit? Not likely.

By the way, it's "variable" beaten lengths multipliers. No specific fractional distance gets assigned anything, it varies with the speed of the leader, in each fraction, in each race, and for each race distance. That's a lot of variance.