I was reading and testing out the code written by @shapital where the output is speed, position and angle during the race at the given trackus index. I was testing out for Nov 3 2019 Race no. 9 where the program number 5 won by a huge lead but the number 8 and 7 were head-to-head with each other. maybe less than 0.25s between their finish line end yt link to the match. The output of the code was that number 8 came in second position and number 7 3rd but the reality is exact opposite. The overall position and dataset were correct since the code output and actual match data correlated highly, but output during finish line differed which is usually the most imp part to track positions.
\nSo this got me thinking whether the fault is in code or the data, since 0.25s is a high time interval for a match like this where its mostly sprinting and not marathon.
\nHope to get some new information from the dataset as well
I was reading and testing out the code written by @shapital where the output is speed, position and angle during the race at the given trackus index. I was testing out for Nov 3 2019 Race no. 9 where the program number 5 won by a huge lead but the number 8 and 7 were head-to-head with each other. maybe less than 0.25s between their finish line end yt link to the match. The output of the code was that number 8 came in second position and number 7 3rd but the reality is exact opposite. The overall position and dataset were correct since the code output and actual match data correlated highly, but output during finish line differed which is usually the most imp part to track positions.
\nSo this got me thinking whether the fault is in code or the data, since 0.25s is a high time interval for a match like this where its mostly sprinting and not marathon.
\nHope to get some new information from the dataset as well
Hi @kunalmakwana55
\nAccording to the video, horse 8 beat horse 7:
\n![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F6272010%2F5952d3836cc1b6eccd79afcb9e214368%2Frace9.jpg?generation=1662371130110582&alt=media\")
\nAnd according to my code, horse 8 was in second place between trakus indexes 346 and 367, after that horse 7 was in second place.
\n![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F6272010%2F338ac5d161907af886456a071b017faa%2Fpositions.jpg?generation=1662371703787558&alt=media\")
\nI guess that horse 8 crossed the finish line somewhere between the two trakus indexes and then slowed down and allowed horse 7 to surpass him.
\nFor this reason we shouldn't assume the horses position in the last trakus index indicate the true race positions.
First of all Thank you for the notebook really learnt a lot from that and second Excuse for the position written I meant to write the exact opposite according to this
\nAlso looks liks the horses does interchange position in their last two trackus, the first change was captured not on video but the final was. So this raises a question on what percentage of ending trackus positions can we neglect to still be able to make it work.
It is worth noting that the finishing positions are not determined at one specific trakus index.
\nThere are races where the current horses positions and their finishing positions do not match at any trakus index. For example, race number 8 at 2019-01-04 at track AQU - horse number 4 finished first, immediately followed by horse number 3. At the same time horse number 7 was in 3rd position and horse number 6 in 4th position, but right after that horse number 6 surpassed and finished 3rd.
\n![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F6272010%2F4d010e333f3fd935330e890bad84f642%2Fdownload.png?generation=1662966554527246&alt=media\")
If we can somehow calculate the distance to the finish line then we can plot the trakus index in which each horse reached each distance (using interpolation), and at distance 0 we will get the finishing positions (lower trakus index means better finishing position). Obviously, as the title of this discussion suggests, there are cases where the 0.25 seconds resolution is not good enough to precisely determine the finishing positions between two (or more) horses running neck and neck.
", "rawMarkdown": "It is worth noting that the finishing positions are not determined at one specific trakus index.\nThere are races where the current horses positions and their finishing positions do not match at any trakus index. For example, race number 8 at 2019-01-04 at track AQU - horse number 4 finished first, immediately followed by horse number 3. At the same time horse number 7 was in 3rd position and horse number 6 in 4th position, but right after that horse number 6 surpassed and finished 3rd.\n\n\nIf we can somehow calculate the distance to the finish line then we can plot the trakus index in which each horse reached each distance (using interpolation), and at distance 0 we will get the finishing positions (lower trakus index means better finishing position). Obviously, as the title of this discussion suggests, there are cases where the 0.25 seconds resolution is not good enough to precisely determine the finishing positions between two (or more) horses running neck and neck.", "votes": null }, { "id": "1954786", "postDate": "09/25/2022 13:03:02", "content": "Do not forget that each horse travels a different distance according to the starting gate.
\nAlthough they start from the same starting line, they do not start from the same point, therefore the horse that starts from the outermost gate (although it may seem in lat/lon that no advantage has been gained), the horse that runs from the internal is always ahead.
\nThis peculiarity only occurs in long races where the curve is taken, when the race is in a straight line the starting gate does not influence the distances.
", "rawMarkdown": "Do not forget that each horse travels a different distance according to the starting gate.\n\nAlthough they start from the same starting line, they do not start from the same point, therefore the horse that starts from the outermost gate (although it may seem in lat/lon that no advantage has been gained), the horse that runs from the internal is always ahead.\n\nThis peculiarity only occurs in long races where the curve is taken, when the race is in a straight line the starting gate does not influence the distances.", "votes": null } ], "comments": [ { "id": 1926985, "author_name": "shapital", "author_url": "", "post_date": "09/05/2022 09:55:46", "content": "Hi @kunalmakwana55
\nAccording to the video, horse 8 beat horse 7:
\n
\nAnd according to my code, horse 8 was in second place between trakus indexes 346 and 367, after that horse 7 was in second place.
\n
\nI guess that horse 8 crossed the finish line somewhere between the two trakus indexes and then slowed down and allowed horse 7 to surpass him.
\nFor this reason we shouldn't assume the horses position in the last trakus index indicate the true race positions.
First of all Thank you for the notebook really learnt a lot from that and second Excuse for the position written I meant to write the exact opposite according to this
\nAlso looks liks the horses does interchange position in their last two trackus, the first change was captured not on video but the final was. So this raises a question on what percentage of ending trackus positions can we neglect to still be able to make it work.
It is worth noting that the finishing positions are not determined at one specific trakus index.
\nThere are races where the current horses positions and their finishing positions do not match at any trakus index. For example, race number 8 at 2019-01-04 at track AQU - horse number 4 finished first, immediately followed by horse number 3. At the same time horse number 7 was in 3rd position and horse number 6 in 4th position, but right after that horse number 6 surpassed and finished 3rd.
\n
If we can somehow calculate the distance to the finish line then we can plot the trakus index in which each horse reached each distance (using interpolation), and at distance 0 we will get the finishing positions (lower trakus index means better finishing position). Obviously, as the title of this discussion suggests, there are cases where the 0.25 seconds resolution is not good enough to precisely determine the finishing positions between two (or more) horses running neck and neck.
", "votes": null, "replies": [] }, { "id": 1954786, "author_name": "gabrielbeguerie", "author_url": "", "post_date": "09/25/2022 13:03:02", "content": "Do not forget that each horse travels a different distance according to the starting gate.
\nAlthough they start from the same starting line, they do not start from the same point, therefore the horse that starts from the outermost gate (although it may seem in lat/lon that no advantage has been gained), the horse that runs from the internal is always ahead.
\nThis peculiarity only occurs in long races where the curve is taken, when the race is in a straight line the starting gate does not influence the distances.
", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "1926906": "I was reading and testing out the code written by @shapital where the output is speed, position and angle during the race at the given trackus index. I was testing out for Nov 3 2019 Race no. 9 where the program number 5 won by a huge lead but the number 8 and 7 were head-to-head with each other. maybe less than 0.25s between their finish line end [yt link to the match](https://youtu.be/OO1a5MLkZhY). The output of the code was that number 8 came in second position and number 7 3rd but the reality is exact opposite. The overall position and dataset were correct since the code output and actual match data correlated highly, but output during finish line differed which is usually the most imp part to track positions.\nSo this got me thinking whether the fault is in code or the data, since 0.25s is a high time interval for a match like this where its mostly sprinting and not marathon.\nHope to get some new information from the dataset as well", "1926985": "Hi @kunalmakwana55 \nAccording to the video, horse 8 beat horse 7:\n\nAnd according to my code, horse 8 was in second place between trakus indexes 346 and 367, after that horse 7 was in second place.\n\nI guess that horse 8 crossed the finish line somewhere between the two trakus indexes and then slowed down and allowed horse 7 to surpass him.\nFor this reason we shouldn't assume the horses position in the last trakus index indicate the true race positions.", "1927340": "First of all Thank you for the notebook really learnt a lot from that and second Excuse for the position written I meant to write the exact opposite according to [this](https://www.offtrackbetting.com/results/1/aqueduct-20191103.html)\nAlso looks liks the horses does interchange position in their last two trackus, the first change was captured not on video but the final was. So this raises a question on what percentage of ending trackus positions can we neglect to still be able to make it work.", "1935561": "It is worth noting that the finishing positions are not determined at one specific trakus index.\nThere are races where the current horses positions and their finishing positions do not match at any trakus index. For example, race number 8 at 2019-01-04 at track AQU - horse number 4 finished first, immediately followed by horse number 3. At the same time horse number 7 was in 3rd position and horse number 6 in 4th position, but right after that horse number 6 surpassed and finished 3rd.\n\n\nIf we can somehow calculate the distance to the finish line then we can plot the trakus index in which each horse reached each distance (using interpolation), and at distance 0 we will get the finishing positions (lower trakus index means better finishing position). Obviously, as the title of this discussion suggests, there are cases where the 0.25 seconds resolution is not good enough to precisely determine the finishing positions between two (or more) horses running neck and neck.", "1954786": "Do not forget that each horse travels a different distance according to the starting gate.\n\nAlthough they start from the same starting line, they do not start from the same point, therefore the horse that starts from the outermost gate (although it may seem in lat/lon that no advantage has been gained), the horse that runs from the internal is always ahead.\n\nThis peculiarity only occurs in long races where the curve is taken, when the race is in a straight line the starting gate does not influence the distances." }, "source": "meta" }