{
  "id": 540812,
  "title": "Whats with these responders?",
  "url": "/competitions/jane-street-real-time-market-data-forecasting/discussion/540812",
  "author_name": "",
  "post_date": "2024-10-16T05:39:59.289508800Z",
  "votes": null,
  "comment_count": 1,
  "views": 0,
  "content": "<p>I was reviewing that notebook where the host of the competition discussed responders, particularly how we could predict responder_6 based on other responders. I’d like to open up a conversation about these responders to clarify what needs to be predicted and what can be calculated algebraically from the other responders.</p>",
  "messages": [
    {
      "id": "3018893",
      "postDate": "10/16/2024 05:39:59",
      "content": "<p>I was reviewing that notebook where the host of the competition discussed responders, particularly how we could predict responder_6 based on other responders. I’d like to open up a conversation about these responders to clarify what needs to be predicted and what can be calculated algebraically from the other responders.</p>",
      "rawMarkdown": "I was reviewing that notebook where the host of the competition discussed responders, particularly how we could predict responder_6 based on other responders. I’d like to open up a conversation about these responders to clarify what needs to be predicted and what can be calculated algebraically from the other responders.",
      "votes": null
    },
    {
      "id": "3019045",
      "postDate": "10/16/2024 07:59:53",
      "content": "<p>The public kernel claiming that r6 can be linearly calculated using other responders is simply wrong. That notebook mistakenly took meta tags as prediction targets. It's amusing to see many upvotes without checking the fact. I mean, if the target can be so easily decoded, the whole competition would be a joke. </p>\n<p>It would be useful though to check the correlation between two responders. In the last competition, responders (or more precisely, <code>resp_i</code>) are returns of the instrument in different time horizon. It is not clearly explained in this competition what responder actually means, but if we assume it is the same variable, we can expect some kind of correlation between responders. </p>",
      "rawMarkdown": "The public kernel claiming that r6 can be linearly calculated using other responders is simply wrong. That notebook mistakenly took meta tags as prediction targets. It's amusing to see many upvotes without checking the fact. I mean, if the target can be so easily decoded, the whole competition would be a joke. \n\nIt would be useful though to check the correlation between two responders. In the last competition, responders (or more precisely, `resp_i`) are returns of the instrument in different time horizon. It is not clearly explained in this competition what responder actually means, but if we assume it is the same variable, we can expect some kind of correlation between responders.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 3019045,
      "author_name": "shiyili",
      "author_url": "",
      "post_date": "10/16/2024 07:59:53",
      "content": "<p>The public kernel claiming that r6 can be linearly calculated using other responders is simply wrong. That notebook mistakenly took meta tags as prediction targets. It's amusing to see many upvotes without checking the fact. I mean, if the target can be so easily decoded, the whole competition would be a joke. </p>\n<p>It would be useful though to check the correlation between two responders. In the last competition, responders (or more precisely, <code>resp_i</code>) are returns of the instrument in different time horizon. It is not clearly explained in this competition what responder actually means, but if we assume it is the same variable, we can expect some kind of correlation between responders. </p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "3018893": "I was reviewing that notebook where the host of the competition discussed responders, particularly how we could predict responder_6 based on other responders. I’d like to open up a conversation about these responders to clarify what needs to be predicted and what can be calculated algebraically from the other responders.",
    "3019045": "The public kernel claiming that r6 can be linearly calculated using other responders is simply wrong. That notebook mistakenly took meta tags as prediction targets. It's amusing to see many upvotes without checking the fact. I mean, if the target can be so easily decoded, the whole competition would be a joke. \n\nIt would be useful though to check the correlation between two responders. In the last competition, responders (or more precisely, `resp_i`) are returns of the instrument in different time horizon. It is not clearly explained in this competition what responder actually means, but if we assume it is the same variable, we can expect some kind of correlation between responders."
  },
  "source": "meta"
}