New GPS Charts

[MJC922]

The point I eventually reached with time is to be rigorous, outwork the competition, leave no stone unturned with the research, that means you adjust for wind followed by pace. Unfortunately I didn't have ground loss info at my disposal then (though I wanted to) because that probably should be the first adjustment IMO.

In theory ground loss should be simple, that's a first-pass distance adjustment (in feet) that will change the interior fractions -- however, track geometry and conditions probably do play a significant role making the impact course-specific. When pace is adjusted for in the second pass ground loss should already have been applied IMO.

Leaving no stone unturned with pace would mean you measure the impact of pace faster / slower than normal of that segment for that specific final time and reduce the final time a variable amount.

Track condition of course can change everything when it comes to the ground loss, such as in the example race, the runner up was hung wide early and was not phased by it, strongly suggests the outside wasn't as bad of a place as it typically would've been when dry.

These are reasons why you want to retain many years of data for studies of this type. Few people are willing to leave no stone unturned. There are real facts if you settle for nothing less than that. A few folks will do some studies and some seat of the pants adjustments. Many more will throw something together for tomorrows card. 1 in 1000 will not bet a dime until they have the real answers.

[classhandicapper]

Originally Posted by MJC922

The point I eventually reached with time is to be rigorous, outwork the competition, leave no stone unturned with the research, that means you adjust for wind followed by pace. Unfortunately I didn't have ground loss info at my disposal then (though I wanted to) because that probably should be the first adjustment IMO.

In theory ground loss should be simple, that's a first-pass distance adjustment (in feet) that will change the interior fractions -- however, track geometry and conditions probably do play a significant role making the impact course-specific. When pace is adjusted for in the second pass ground loss should already have been applied IMO.

Leaving no stone unturned with pace would mean you measure the impact of pace faster / slower than normal of that segment for that specific final time and reduce the final time a variable amount.

Track condition of course can change everything when it comes to the ground loss, such as in the example race, the runner up was hung wide early and was not phased by it, strongly suggests the outside wasn't as bad of a place as it typically would've been when dry.

These are reasons why you want to retain many years of data for studies of this type. Few people are willing to leave no stone unturned. There are real facts if you settle for nothing less than that. A few folks will do some studies and some seat of the pants adjustments. Many more will throw something together for tomorrows card. 1 in 1000 will not bet a dime until they have the real answers.

The complexities of making adjustments for pace, ground loss, weight, how the track was playing (path biases and more or less than tiring than usual), wind, changing track speed, run up etc.. can be so overwhelmingly difficult to quantify accurately, I sometimes wonder how much is gained by trying over just looking at the horse's overall record and noticing the extreme trip races that clearly lead to the horse looking way better or worse than he actually was on a specific day.

There's some evidence that the more factors you add into a formula that could have small errors, the greater the probability that the final result will be very wrong.

It might be simpler to say, horse "A" typically runs figures in the high 90s or low 100s. On such and such a date he ran an 89. Then you look at the trip note and see he was very wide on a day when saving ground was a clear advantage. Then instead of doing a ground loss adjustment for that 89, you simply say "that was enough of excuse for the subpar race. He should get back to the high 90s today".

[MJC922]

It's certainly simpler to say. Part of what I take issue with is people have been simply saying things for far too long in this game. They'll write whole books about things that haven't been rigorously tested, it's all too anecdotal going from way back.

If there is data available however and you're not into cutting corners then you test without the new factor and then test with the new factor, whether it's wind, pace, ground loss etc. If win rate improves then you're looking at a net benefit to accuracy which typically you'll want.

I do see the point however that not everyone wants an all-encompassing measure of performance though, some people want just final time and nothing else, they want to evaluate the impact of the trip based upon their own subjective assessment of it.

[classhandicapper]

Originally Posted by MJC922

It's certainly simpler to say. Part of what I take issue with is people have been simply saying things for far too long in this game. They'll write whole books about things that haven't been rigorously tested, it's all too anecdotal going from way back.

If there is data available however and you're not into cutting corners then you test without the new factor and then test with the new factor, whether it's wind, pace, ground loss etc. If win rate improves then you're looking at a net benefit to accuracy which typically you'll want.

I agree 100%.

Knowing CJ for a long time, I'm sure he's testing his ground loss formulas trying to find a way to maximize the results.

I do that myself when I try new things, but ultimately I find that I can't code for everything and wind up being less systematic in practice.

[MJC922]

Originally Posted by classhandicapper

I agree 100%.

Knowing CJ for a long time, I'm sure he's testing his ground loss formulas trying to find a way to maximize the results.

I do that myself when I try new things, but ultimately I find that I can't code for everything and wind up being less systematic in practice.

With folks like you and CJ it's more preaching to the choir than anything else. You guys are obviously willing to jump through more hoops than 99/100. I still need a kick in the pants myself for a reminder. It's more directed at the "I'm going to throw something new together for tomorrows card" crowd. Been there myself many times growing up.

Most of us have all grown up with the same handicapping literature and we all should recognize that because of that alone we're almost conditioned at this point to lean toward the theoretical because almost nobody had sufficient sample sizes back in the day to do anything solid research-wise. I mean take just two examples from perhaps two of best handicapping books ever written, Quirin, look at some of his spot plays at the end of Winning at the Races, what were those a few hundred races and backfit to boot? and that was one of the best handicapping books ever published IMO. Then we have Beyer's Picking Winners, his speed chart based upon proportional time, he will stand in front of his maker to declare the factual distance to distance relationship between what was it 6f and 7f?

I agree proportional time was a great leap ahead, but when you really dig deeper you'll still be left wanting. Leave run-ups out of the discussion for the sake of simplicity. You can have a custom chart for each track and if you're smart you do but certainly for example 6f to 1M out of the chute falls apart by how many fifths if there happens to be a 20mph tailwind that day? Even 10 mph, does anyone not think it's a 1/5th or two? You're running what 45 or more seconds with the wind behind you at the mile and we use the same speed chart, no, that needs adjusting based upon wind conditions. Even wind aside, does anyone think the horse who beats you by 20 lengths at two miles beats you by 40 at four miles. Races are run differently at two miles than 6f. Even on turf he had to tweak the chart. The chart was tweaked for beaten lengths but not final time on turf, correct? so there's a lot there folks do need to rethink and research for themselves. Proportional time works up to a certain point when races are run a certain way but it's not universally applicable.

[classhandicapper]

I agree with everything you are saying.

It's not that I don't strive for greater accuracy. I've just conceded that I'm often working with inaccurate/incomplete figures and/or figures that were impacted by race development in such subtle ways I'm not going to be able to adjust them because I'm not even sure how.

So from a final time perspective, I'm thinking more in terms of "does the horse fit in this field (ballpark)" and then I'm looking at the more subjective areas of the game like the quality of the fields it has been running against, prior trips, projected pace and trip, bias, is the horse going forward or backward, can I find a blatantly wrong figure to exploit etc.. hoping to find a horse that's better/worse than they look on paper so there's a chance I can find some value.

For me, it's getting tougher. There's so much good information in the social media/computer era that the learning curve is very accelerated compared to what it once was. It used to take many years to get pretty good. Now it can happen fast. IMO, there are way more very good players today than years ago and at the same time the weaker players dropped out or switched to lotteries.

[cj]

As good as it sounds in theory, they really need to do better than they did for the Apple Blossom today. Here is the replay, yet the data shows Ce Ce covered just 13 more feet than Ollie's Candy. What is more alarming is it shows she covered 7 feet LESS than Come Dancing, a horse that was inside of her for the entirety of both turns.

YouTube

http://www.equibase.com/static/chart...041820USA.html

[MJC922]

Yea CJ that's crazy. I wonder do they get the transmitters mixed up? Is there a different transmitter for time and another for distance? lol I give you credit for sticking in there with the BS you have to deal with when it comes to the various time-related inputs. For years the typos and mistiming has been going on, when I had to process every track in NA I had no choice but to walk away from it. I had a VBA function that could flag incorrect times. Literally thousands of fractions per year are off. Look at my numbers today, it's all final position based stuff and here I am a pace/trip guy. Very disappointing to still see issues like this in 2020, but I guess they're at least trying some new things, at some point it will be solid hopefully.

[classhandicapper]

Once you find a couple of errors like this that means there are a lot more.

That's kind of why I never accepted Trakus ground loss at face value and continued looking at the replays. Having the information is very helpful, but some of the mistakes are too large to simply accept the data at face value all the time.

[cj]

Originally Posted by MJC922

Yea CJ that's crazy. I wonder do they get the transmitters mixed up? Is there a different transmitter for time and another for distance? lol I give you credit for sticking in there with the BS you have to deal with when it comes to the various time-related inputs. For years the typos and mistiming has been going on, when I had to process every track in NA I had no choice but to walk away from it. I had a VBA function that could flag incorrect times. Literally thousands of fractions per year are off. Look at my numbers today, it's all final position based stuff and here I am a pace/trip guy. Very disappointing to still see issues like this in 2020, but I guess they're at least trying some new things, at some point it will be solid hopefully.

No, there aren't different transmitters. The chip gives the location and then the GPS time is used to determine where the horse was at different points and when. I know there is some experimenting going on to use two chips to improve accuracy.

I do a lot of what you did now with the typos as far as flagging aberrant times, fractions and final It has gotten much, much better this past year. I'd like to think it is at least in part because I was constantly bothering them to fix things.

[cj]

Originally Posted by classhandicapper

Once you find a couple of errors like this that means there are a lot more.

That's kind of why I never accepted Trakus ground loss at face value and continued looking at the replays. Having the information is very helpful, but some of the mistakes are too large to simply accept the data at face value all the time.

Exactly. I don't expect exact precision, but the product should be reliable within reason. It is new and willing to give it some time but I can't trust it right now. The times had a lot of problems in the beginning as well but are much better now so I remain hopeful.

[classhandicapper]

Originally Posted by cj

Exactly. I don't expect exact precision, but the product should be reliable within reason. It is new and willing to give it some time but I can't trust it right now. The times had a lot of problems in the beginning as well but are much better now so I remain hopeful.

Keep us up to date from time to time. I don't have the energy right now to look at it.

[cutchemist42]

Very cool to hear about this.

[Tom]

Discussion of the GPS system on Steve Byk today, second hour.
Interesting stuff planned for workouts down the road.