{
  "id": 589781,
  "title": "How easy do you think it is to separate the two ?",
  "url": "/competitions/drw-crypto-market-prediction/discussion/589781",
  "author_name": "",
  "post_date": "2025-07-15T11:12:23.931356900Z",
  "votes": null,
  "comment_count": 3,
  "views": 0,
  "content": "<p><img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F11146709%2Fbf9a7dd0a569bc2522a0c04fd80a2015%2Foutput.png?generation=1752577897218149&amp;alt=media\" alt=\"\"></p>",
  "messages": [
    {
      "id": "3248899",
      "postDate": "07/15/2025 11:12:23",
      "content": "<p><img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F11146709%2Fbf9a7dd0a569bc2522a0c04fd80a2015%2Foutput.png?generation=1752577897218149&amp;alt=media\" alt=\"\"></p>",
      "rawMarkdown": "![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F11146709%2Fbf9a7dd0a569bc2522a0c04fd80a2015%2Foutput.png?generation=1752577897218149&alt=media)",
      "votes": null
    },
    {
      "id": "3248928",
      "postDate": "07/15/2025 11:59:14",
      "content": "<p>One should not look only at the scores, since then the best way to cheat might be to aim for a lower score that is still high enough, like 0.2 or something. </p>\n<p>I think a detailed code review is the only way (if it's even possible) to at least try to catch the cheaters. </p>",
      "rawMarkdown": "One should not look only at the scores, since then the best way to cheat might be to aim for a lower score that is still high enough, like 0.2 or something. \n\nI think a detailed code review is the only way (if it's even possible) to at least try to catch the cheaters.",
      "votes": null
    },
    {
      "id": "3249152",
      "postDate": "07/15/2025 20:07:38",
      "content": "<p>The competition description says they will conduct a code review</p>",
      "rawMarkdown": "The competition description says they will conduct a code review",
      "votes": null
    },
    {
      "id": "3249153",
      "postDate": "07/15/2025 20:10:00",
      "content": "<p>This is just a split between time series approaches and non-time series, not a split between cheaters and non-cheaters. Absolutely no way of knowing whether a 0.15 used the ordered test offline to choose features or not.</p>",
      "rawMarkdown": "This is just a split between time series approaches and non-time series, not a split between cheaters and non-cheaters. Absolutely no way of knowing whether a 0.15 used the ordered test offline to choose features or not.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 3248928,
      "author_name": "ern711",
      "author_url": "",
      "post_date": "07/15/2025 11:59:14",
      "content": "<p>One should not look only at the scores, since then the best way to cheat might be to aim for a lower score that is still high enough, like 0.2 or something. </p>\n<p>I think a detailed code review is the only way (if it's even possible) to at least try to catch the cheaters. </p>",
      "votes": null,
      "replies": [
        {
          "id": 3249152,
          "author_name": "paperxd",
          "author_url": "",
          "post_date": "07/15/2025 20:07:38",
          "content": "<p>The competition description says they will conduct a code review</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 3249153,
      "author_name": "greatestcutie",
      "author_url": "",
      "post_date": "07/15/2025 20:10:00",
      "content": "<p>This is just a split between time series approaches and non-time series, not a split between cheaters and non-cheaters. Absolutely no way of knowing whether a 0.15 used the ordered test offline to choose features or not.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "3248899": "![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F11146709%2Fbf9a7dd0a569bc2522a0c04fd80a2015%2Foutput.png?generation=1752577897218149&alt=media)",
    "3248928": "One should not look only at the scores, since then the best way to cheat might be to aim for a lower score that is still high enough, like 0.2 or something. \n\nI think a detailed code review is the only way (if it's even possible) to at least try to catch the cheaters.",
    "3249152": "The competition description says they will conduct a code review",
    "3249153": "This is just a split between time series approaches and non-time series, not a split between cheaters and non-cheaters. Absolutely no way of knowing whether a 0.15 used the ordered test offline to choose features or not."
  },
  "source": "meta"
}