{
  "id": 583349,
  "title": "What is current the best LB score without using external data? ",
  "url": "/competitions/drw-crypto-market-prediction/discussion/583349",
  "author_name": "Rafał Pawłowski",
  "post_date": "2025-06-06T07:20:41.622000",
  "votes": 3,
  "comment_count": 11,
  "views": 0,
  "content": "<p>It seems to be about 0.12 now. </p>",
  "messages": [
    {
      "id": 3218621,
      "postDate": "2025-06-06T12:58:45.070Z",
      "content": "<p>My guess is around 0.1x. Anything above is using external data. </p>",
      "rawMarkdown": "My guess is around 0.1x. Anything above is using external data. ",
      "votes": 3
    },
    {
      "id": 3218625,
      "postDate": "2025-06-06T13:06:34.150Z",
      "content": "<p>Well, actually I think it's quite boring if it's allowed to use external newer data. It definitely alters what to optimize for, making it less interesting IMO.  </p>",
      "rawMarkdown": "Well, actually I think it's quite boring if it's allowed to use external newer data. It definitely alters what to optimize for, making it less interesting IMO.  ",
      "votes": 1,
      "replies": [
        {
          "id": 3218632,
          "postDate": "2025-06-06T13:17:04.470Z",
          "content": "<p>Also, I think it was less interesting that they masked time…</p>",
          "rawMarkdown": "Also, I think it was less interesting that they masked time...",
          "votes": 3,
          "replies": [
            {
              "id": 3218638,
              "postDate": "2025-06-06T13:28:30.013Z",
              "content": "<p>Yes, it's also really vague what's allowed or not.</p>\n<p>They state:</p>\n<ul>\n<li><p>You are NOT allowed to use the test dataset to aid the modeling process, as all test data—except for the specific query data point—contains future information that would not be available at prediction time in a real-world setting.</p></li>\n<li><p>Similar to the note about using future data, any external dataset containing future information will be considered a rule violation. So make sure everything you use would have been available at the time of test data prediction!</p></li>\n</ul>\n<p>Does this mean I'm not allowed / allowed to use adversarial validation? The features for the test data points are clearly available at prediction time (otherwise we couldn't use them to predict) meaning I could perhaps use them to improve my model? </p>\n<p>What do they mean by \"the specific query data point\" etc? </p>",
              "rawMarkdown": "Yes, it's also really vague what's allowed or not.\n\nThey state:\n\n* You are NOT allowed to use the test dataset to aid the modeling process, as all test data—except for the specific query data point—contains future information that would not be available at prediction time in a real-world setting.\n\n* Similar to the note about using future data, any external dataset containing future information will be considered a rule violation. So make sure everything you use would have been available at the time of test data prediction!\n\nDoes this mean I'm not allowed / allowed to use adversarial validation? The features for the test data points are clearly available at prediction time (otherwise we couldn't use them to predict) meaning I could perhaps use them to improve my model? \n\n What do they mean by \"the specific query data point\" etc? \n\n\n",
              "votes": 3
            },
            {
              "id": 3218649,
              "postDate": "2025-06-06T13:46:03.060Z",
              "content": "<p>I'm just using train.parquet to predict. Yes, it is really vague. </p>",
              "rawMarkdown": "I'm just using train.parquet to predict. Yes, it is really vague. ",
              "votes": 3
            }
          ]
        }
      ]
    },
    {
      "id": 3218430,
      "postDate": "2025-06-06T07:20:41.623Z",
      "content": "<p>It seems to be about 0.12 now. </p>",
      "rawMarkdown": "It seems to be about 0.12 now. ",
      "votes": 2
    },
    {
      "id": 3218676,
      "postDate": "2025-06-06T14:53:55.380Z",
      "content": "<p>how do you know that top players use external data?</p>",
      "rawMarkdown": "how do you know that top players use external data?\n",
      "replies": [
        {
          "id": 3221225,
          "postDate": "2025-06-10T15:33:48.547Z",
          "rawMarkdown": "",
          "isDeleted": true
        }
      ]
    },
    {
      "id": 3218463,
      "postDate": "2025-06-06T07:58:41.843Z",
      "content": "<p>Has it been confirmed that the current top scores are using external data? </p>",
      "rawMarkdown": "Has it been confirmed that the current top scores are using external data? ",
      "replies": [
        {
          "id": 3218479,
          "postDate": "2025-06-06T08:22:24.637Z",
          "content": "<p>No. Maybe top teams will reveal this, but  I think they use extra data.  </p>",
          "rawMarkdown": "No. Maybe top teams will reveal this, but  I think they use extra data.  ",
          "replies": [
            {
              "id": 3218487,
              "postDate": "2025-06-06T08:38:40.727Z",
              "content": "<p>Without knowing, I agree that 0.27 sounds higher than I would expect, given that the test period is future data from a different date. But who knows.</p>",
              "rawMarkdown": "Without knowing, I agree that 0.27 sounds higher than I would expect, given that the test period is future data from a different date. But who knows."
            }
          ]
        }
      ]
    },
    {
      "id": 3221516,
      "postDate": "2025-06-11T05:42:17.373Z",
      "rawMarkdown": "",
      "votes": 2,
      "isDeleted": true
    }
  ],
  "comments": [
    {
      "id": 3218621,
      "author_name": "byunjins",
      "author_url": "",
      "post_date": "2025-06-06T12:58:45.070000",
      "content": "<p>My guess is around 0.1x. Anything above is using external data. </p>",
      "votes": 3,
      "replies": []
    },
    {
      "id": 3218625,
      "author_name": "Ern711",
      "author_url": "",
      "post_date": "2025-06-06T13:06:34.150000",
      "content": "<p>Well, actually I think it's quite boring if it's allowed to use external newer data. It definitely alters what to optimize for, making it less interesting IMO.  </p>",
      "votes": 1,
      "replies": [
        {
          "id": 3218632,
          "author_name": "byunjins",
          "author_url": "",
          "post_date": "2025-06-06T13:17:04.470000",
          "content": "<p>Also, I think it was less interesting that they masked time…</p>",
          "votes": 3,
          "replies": [
            {
              "id": 3218638,
              "author_name": "Ern711",
              "author_url": "",
              "post_date": "2025-06-06T13:28:30.013000",
              "content": "<p>Yes, it's also really vague what's allowed or not.</p>\n<p>They state:</p>\n<ul>\n<li><p>You are NOT allowed to use the test dataset to aid the modeling process, as all test data—except for the specific query data point—contains future information that would not be available at prediction time in a real-world setting.</p></li>\n<li><p>Similar to the note about using future data, any external dataset containing future information will be considered a rule violation. So make sure everything you use would have been available at the time of test data prediction!</p></li>\n</ul>\n<p>Does this mean I'm not allowed / allowed to use adversarial validation? The features for the test data points are clearly available at prediction time (otherwise we couldn't use them to predict) meaning I could perhaps use them to improve my model? </p>\n<p>What do they mean by \"the specific query data point\" etc? </p>",
              "votes": 3,
              "replies": []
            },
            {
              "id": 3218649,
              "author_name": "byunjins",
              "author_url": "",
              "post_date": "2025-06-06T13:46:03.060000",
              "content": "<p>I'm just using train.parquet to predict. Yes, it is really vague. </p>",
              "votes": 3,
              "replies": []
            }
          ]
        }
      ]
    },
    {
      "id": 3218676,
      "author_name": "Yourui Wang",
      "author_url": "",
      "post_date": "2025-06-06T14:53:55.380000",
      "content": "<p>how do you know that top players use external data?</p>",
      "votes": 0,
      "replies": [
        {
          "id": 3221225,
          "author_name": "",
          "author_url": "",
          "post_date": "2025-06-10T15:33:48.547000",
          "content": "",
          "votes": 0,
          "replies": []
        }
      ]
    },
    {
      "id": 3218463,
      "author_name": "Ern711",
      "author_url": "",
      "post_date": "2025-06-06T07:58:41.843000",
      "content": "<p>Has it been confirmed that the current top scores are using external data? </p>",
      "votes": 0,
      "replies": [
        {
          "id": 3218479,
          "author_name": "Rafał Pawłowski",
          "author_url": "",
          "post_date": "2025-06-06T08:22:24.637000",
          "content": "<p>No. Maybe top teams will reveal this, but  I think they use extra data.  </p>",
          "votes": 0,
          "replies": [
            {
              "id": 3218487,
              "author_name": "Ern711",
              "author_url": "",
              "post_date": "2025-06-06T08:38:40.727000",
              "content": "<p>Without knowing, I agree that 0.27 sounds higher than I would expect, given that the test period is future data from a different date. But who knows.</p>",
              "votes": 0,
              "replies": []
            }
          ]
        }
      ]
    },
    {
      "id": 3221516,
      "author_name": "",
      "author_url": "",
      "post_date": "2025-06-11T05:42:17.373000",
      "content": "",
      "votes": 2,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "3218621": "My guess is around 0.1x. Anything above is using external data. ",
    "3218625": "Well, actually I think it's quite boring if it's allowed to use external newer data. It definitely alters what to optimize for, making it less interesting IMO.  ",
    "3218430": "It seems to be about 0.12 now. ",
    "3218676": "how do you know that top players use external data?\n",
    "3218463": "Has it been confirmed that the current top scores are using external data? ",
    "3221516": ""
  }
}