TPD Feedback and Issues

Integrate live racecourse data into your favourite Betfair trading interface from TPD.
Post Reply
User avatar
jamesedwards
Posts: 1069
Joined: Wed Nov 21, 2018 6:16 pm

TPD-Zone wrote:
Fri Apr 29, 2022 9:44 am
Regarding Ascot 17:10, 27 APR. The relative distances between 1st and 2nd, 2nd and 3rd etc were 3/4L, 3/4L, 1L, nk, 1L, 3/4L, so reasonably tight. All the horses were tracked to 91% or 92% accuracy, as compared to the official course timings, which is not good, we are generally in the high 90%'s.

In our internal systems I can see radio reception and data errors and I suspect the radio reception must have dropped out completely, for a short period of time. In any event the TPD Content Director, who is responsible for the GPS tracking operations is aware.

As an aside, there will be an opportunity to visit the weighing room, at the Southwell Workshop and observe the GPS tracking technology, in action, as it were and chat with the Director mentioned above and his Field Operators.
Looking forward very much to this. No doubt one of the key agenda points from the audience will be to learn more about the process, how accuracy is measured, and plans to improve.
PeterLe
Posts: 3657
Joined: Wed Apr 15, 2009 3:19 pm

I was just looking at the first race at Chester today..
This is a screenshot just as the race started (note the speeds as a verification/indicator) that the horse is moving
The race was 1019.6m
The distance to finish for all horses were never in synch when the horses were in the gates. Absolutely flawless and Glorious Angel appeared to be 30-40 m in front initially just millisecond before the off and as the race is suspended they are all reset to ~1019

I dont think this is how you did it in the past? How long have you been recalibrating the positions at the off?
By the way, I think that this is the best way of doing it, but why would they be off by 30-40 m thats a large margin?
Attachments
Chester1330 4htmay.PNG
SteadySlobbin
Posts: 41
Joined: Wed Mar 09, 2022 8:17 am

PeterLe wrote:
Wed May 04, 2022 1:50 pm
I was just looking at the first race at Chester today..
This is a screenshot just as the race started (note the speeds as a verification/indicator) that the horse is moving
The race was 1019.6m
The distance to finish for all horses were never in synch when the horses were in the gates. Absolutely flawless and Glorious Angel appeared to be 30-40 m in front initially just millisecond before the off and as the race is suspended they are all reset to ~1019

I dont think this is how you did it in the past? How long have you been recalibrating the positions at the off?
By the way, I think that this is the best way of doing it, but why would they be off by 30-40 m thats a large margin?

Is it something to do with the draw positions? AF and GA both drawn widest? Though the graphic shows stall 2 as being closer to stall 1?! Though not sure what I’m looking at to be honest.
User avatar
ShaunWhite
Posts: 7574
Joined: Sat Sep 03, 2016 3:42 am

PeterLe wrote:
Wed May 04, 2022 1:50 pm
I dont think this is how you did it in the past? How long have you been recalibrating the positions at the off?
By the way, I think that this is the best way of doing it, but why would they be off by 30-40 m thats a large margin?
I can't answer that Peter but TPD were making some very encouraging noises today about recent changes and on-going improvements. Two trackers per cloth averaged, additional signal relay antenna at some courses, stalls starts linked to the starters button, centralised monitoring and plans for more operational staff at each meeting, improved gps hardware. I certainly feel much more inclined to invest some time into it that I would have a year or two ago, and it's still work in progress which is enough reason for me to cut them a bit of slack. They know it's 90% which is great because so many firms put out half finished products and call it 100%.
PeterLe
Posts: 3657
Joined: Wed Apr 15, 2009 3:19 pm

Thanks Shaun for the update, hope you guys had a good day.
When it works it is fantastic, when it doesn't, it costs.
I've been recording my screen at 60fps and there are many races where TPD would say that the values are within spec (+/- 1 horse length i think) but the values don't represent the true picture. Often but that figure can be the difference between a 1.01 and a 1000.
You can use TPD as a verification of the odds, but again, there are quirks. I have stuff running that doesn't use TPD and catch some fantastic trades that I wouldn't have caught with TPD
Overall TPD is just additional data, its how you interpret it; model it and apply it that makes the difference.
Regards
Peter
User avatar
jamesedwards
Posts: 1069
Joined: Wed Nov 21, 2018 6:16 pm

What happened in Gulfstream race 8?
  • 10 Dunkle finished 5 lengths back from the winner in 4th yet was first past the post according to TPD.
  • 11 Ken Loves Kitten finished 2nd by a nose yet was around 5 lengths back in 4th according to TPD.

Looking at the data it looks as though the trackers for 10 and 11 might have been transposed throughout the race?!!! :shock:

dunkle2.png
sheffutd1889
Posts: 14
Joined: Mon Jun 05, 2017 3:38 pm

Is there any reason why this is saying unsupported? It's one of the new US tracks but status page doesn't say there's any problems
Attachments
aaa.png
aaa.png (8.16 KiB) Viewed 154 times
deestar
Posts: 123
Joined: Thu Feb 20, 2020 6:17 pm

This might be a silly question but here goes; is it possible from the current data feed to determine when each horse has loaded into the stalls, and therefore the number left to load.

Thanks in advance.
User avatar
TPD-Zone
Posts: 53
Joined: Wed Jan 12, 2022 8:27 am
Location: London, UK
Contact:

Regarding Gulfstream race 8 and the possible transposition of tracking devices - no. 10 Dunkle and no. 11 Ken Loves Kitten.
The TPD UK team are in agreement that the GPS devices we incorrectly allocated and the TPD director responsible for this element of the service is taking the matter up with our US partners.
Apologies for the error.
User avatar
TPD-Zone
Posts: 53
Joined: Wed Jan 12, 2022 8:27 am
Location: London, UK
Contact:

sheffutd1889 wrote:
Sun May 22, 2022 7:34 pm
Is there any reason why this is saying unsupported? It's one of the new US tracks but status page doesn't say there's any problems
The TPD Lead Developer is talking to his Bet Angel contacts and hopefully this will be resolved soonest.
User avatar
TPD-Zone
Posts: 53
Joined: Wed Jan 12, 2022 8:27 am
Location: London, UK
Contact:

deestar wrote:
Mon May 23, 2022 1:18 pm
This might be a silly question but here goes; is it possible from the current data feed to determine when each horse has loaded into the stalls, and therefore the number left to load.

Thanks in advance.
I am afraid that is not possible with the current set-up, but an interesting point.
deestar
Posts: 123
Joined: Thu Feb 20, 2020 6:17 pm

TPD-Zone wrote:
Mon May 23, 2022 2:30 pm
deestar wrote:
Mon May 23, 2022 1:18 pm
This might be a silly question but here goes; is it possible from the current data feed to determine when each horse has loaded into the stalls, and therefore the number left to load.

Thanks in advance.
I am afraid that is not possible with the current set-up, but an interesting point.
OK. Thanks for the response. I would find it useful for pre-off trading close out purposes.
User avatar
Euler
Posts: 22425
Joined: Wed Nov 10, 2010 1:39 pm
Location: Bet Angel HQ

You can derive this information from the speed and stride data. If the horse is still out it will typically be moving, but in the stalls the stride will be close to zero. You get a little jitter on the data, but it's easy to see who is loaded using this data.
User avatar
ShaunWhite
Posts: 7574
Joined: Sat Sep 03, 2016 3:42 am

Euler wrote:
Mon May 23, 2022 3:10 pm
You can derive this information from the speed and stride data. If the horse is still out it will typically be moving, but in the stalls the stride will be close to zero. You get a little jitter on the data, but it's easy to see who is loaded using this data.
Ditto for markets that go in-running late. If >half are moving, it's started. But yes speed and stride are much better than looking for them being in a line, that could be wrong as TPD recalibrate that once they're loaded.
User avatar
TPD-Zone
Posts: 53
Joined: Wed Jan 12, 2022 8:27 am
Location: London, UK
Contact:

Unfortunately it is highly likely that we will not be operating at Rippon today. Our Field Operator, the guy who deals with the on course GPS trackers and associated infrastructure, has come down with Covid. Apologies for the inconvenience.
Post Reply

Return to “Bet Angel - Total Performance Data - FAQ's”