{ "id": 285670, "title": "There are two types of people: those with LB=0.999 and those who avoid the temptation ;)", "url": "/competitions/g-research-crypto-forecasting/discussion/285670", "author_name": "Tarlan Nazarov", "post_date": "2021-11-05T17:26:18.128000", "votes": 9, "comment_count": 5, "views": 0, "content": "

I guess the percentage of participants with LB=0.999 will be around 33% by the end of this competition. 😁😁 The public LB has already become meaningless… πŸ˜ƒπŸ˜ƒ

", "messages": [ { "id": 1572458, "postDate": "2021-11-05T17:26:18.130Z", "content": "

I guess the percentage of participants with LB=0.999 will be around 33% by the end of this competition. 😁😁 The public LB has already become meaningless… πŸ˜ƒπŸ˜ƒ

", "rawMarkdown": "I guess the percentage of participants with LB=0.999 will be around 33% by the end of this competition. 😁😁 The public LB has already become meaningless... πŸ˜ƒπŸ˜ƒ", "votes": 9 }, { "id": 1572460, "postDate": "2021-11-05T17:28:04.583Z", "content": "

Would be nice to have it cleared (and leak fixed) so we can see how we actually rank against others..

", "rawMarkdown": "Would be nice to have it cleared (and leak fixed) so we can see how we actually rank against others..", "replies": [ { "id": 1572811, "postDate": "2021-11-06T01:34:24.440Z", "content": "

As others have pointed out, the test data is contained within the training data so the leaderboard won't be useful anyways. I think it makes it even more exciting, as we need to trust our own CV :)

", "rawMarkdown": "As others have pointed out, the test data is contained within the training data so the leaderboard won't be useful anyways. I think it makes it even more exciting, as we need to trust our own CV :)", "votes": 1 } ] }, { "id": 1573109, "postDate": "2021-11-06T09:30:59.110Z", "content": "

There's no leak. The organizers have made this clear right from the beginning. I have addressed this thing here as well. Good luck :)
\n\"\"

", "rawMarkdown": "There's no leak. The organizers have made this clear right from the beginning. I have addressed this thing [here](https://www.kaggle.com/c/g-research-crypto-forecasting/discussion/284905) as well. Good luck :)\n![](https://user-images.githubusercontent.com/53314831/140604939-b3a01ee5-24fa-48c5-87ea-28b6f52e09e0.JPG)" }, { "id": 1572802, "postDate": "2021-11-06T01:28:37.030Z", "content": "

Is there actually a data leak?

", "rawMarkdown": "Is there actually a data leak? ", "isDeleted": true, "replies": [ { "id": 1572806, "postDate": "2021-11-06T01:32:18.510Z", "content": "

Well, the past prices of the assets are known so making predictions for them in the past is trivial. This won't be possible on the private data as they use an api which you need to use to make a submission.

", "rawMarkdown": "Well, the past prices of the assets are known so making predictions for them in the past is trivial. This won't be possible on the private data as they use an api which you need to use to make a submission." } ] } ], "comments": [ { "id": 1572460, "author_name": "JM", "author_url": "", "post_date": "2021-11-05T17:28:04.583000", "content": "

Would be nice to have it cleared (and leak fixed) so we can see how we actually rank against others..

", "votes": 0, "replies": [ { "id": 1572811, "author_name": "Fritz Cremer", "author_url": "", "post_date": "2021-11-06T01:34:24.440000", "content": "

As others have pointed out, the test data is contained within the training data so the leaderboard won't be useful anyways. I think it makes it even more exciting, as we need to trust our own CV :)

", "votes": 1, "replies": [] } ] }, { "id": 1573109, "author_name": "Ultron", "author_url": "", "post_date": "2021-11-06T09:30:59.110000", "content": "

There's no leak. The organizers have made this clear right from the beginning. I have addressed this thing here as well. Good luck :)
\n\"\"

", "votes": 0, "replies": [] }, { "id": 1572802, "author_name": "", "author_url": "", "post_date": "2021-11-06T01:28:37.030000", "content": "

Is there actually a data leak?

", "votes": 0, "replies": [ { "id": 1572806, "author_name": "Fritz Cremer", "author_url": "", "post_date": "2021-11-06T01:32:18.510000", "content": "

Well, the past prices of the assets are known so making predictions for them in the past is trivial. This won't be possible on the private data as they use an api which you need to use to make a submission.

", "votes": 0, "replies": [] } ] } ], "raw_markdown_by_id": { "1572458": "I guess the percentage of participants with LB=0.999 will be around 33% by the end of this competition. 😁😁 The public LB has already become meaningless... πŸ˜ƒπŸ˜ƒ", "1572460": "Would be nice to have it cleared (and leak fixed) so we can see how we actually rank against others..", "1573109": "There's no leak. The organizers have made this clear right from the beginning. I have addressed this thing [here](https://www.kaggle.com/c/g-research-crypto-forecasting/discussion/284905) as well. Good luck :)\n![](https://user-images.githubusercontent.com/53314831/140604939-b3a01ee5-24fa-48c5-87ea-28b6f52e09e0.JPG)", "1572802": "Is there actually a data leak? " } }