{
  "id": 553421,
  "title": "Any insights on using symbol_id and time_id?",
  "url": "/competitions/jane-street-real-time-market-data-forecasting/discussion/553421",
  "author_name": "",
  "post_date": "2024-12-26T00:29:38.410657Z",
  "votes": 3,
  "comment_count": 2,
  "views": 0,
  "content": "<p>Does anyone have any insights on using time_id and/or symbol_id as additional features (on top of the 79 raw features that we're already given)? Or are these things somehow already embedded into the existing feature set? Are there any posts with analysis on this?</p>",
  "messages": [
    {
      "id": "3080924",
      "postDate": "12/26/2024 00:29:38",
      "content": "<p>Does anyone have any insights on using time_id and/or symbol_id as additional features (on top of the 79 raw features that we're already given)? Or are these things somehow already embedded into the existing feature set? Are there any posts with analysis on this?</p>",
      "rawMarkdown": "Does anyone have any insights on using time_id and/or symbol_id as additional features (on top of the 79 raw features that we're already given)? Or are these things somehow already embedded into the existing feature set? Are there any posts with analysis on this?",
      "votes": null
    },
    {
      "id": "3081050",
      "postDate": "12/26/2024 06:26:55",
      "content": "<p>Those are the oof scores per <code>time_id</code> for me. I think using <code>time_id</code> helps a bit since models can distinguish regular trading hours, pre-market and after hours.<br>\n<img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F2706866%2Ff33b6519207b66811c35136d2de9255f%2Foof_time_id_scores.png?generation=1735194303166519&amp;alt=media\" alt=\"time_id scores\"></p>",
      "rawMarkdown": "Those are the oof scores per `time_id` for me. I think using `time_id` helps a bit since models can distinguish regular trading hours, pre-market and after hours.\n![time_id scores](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F2706866%2Ff33b6519207b66811c35136d2de9255f%2Foof_time_id_scores.png?generation=1735194303166519&alt=media)",
      "votes": null
    },
    {
      "id": "3081484",
      "postDate": "12/26/2024 19:33:56",
      "content": "<p>dropping time id did drop my xgboost model by quite a bit on the leaderboard score, so i figure it's probably important.</p>\n<p>any insights on keeping symbol id as a feature, or is that somehow already embedded into the features?</p>",
      "rawMarkdown": "dropping time id did drop my xgboost model by quite a bit on the leaderboard score, so i figure it's probably important.\n\nany insights on keeping symbol id as a feature, or is that somehow already embedded into the features?",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 3081050,
      "author_name": "gunesevitan",
      "author_url": "",
      "post_date": "12/26/2024 06:26:55",
      "content": "<p>Those are the oof scores per <code>time_id</code> for me. I think using <code>time_id</code> helps a bit since models can distinguish regular trading hours, pre-market and after hours.<br>\n<img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F2706866%2Ff33b6519207b66811c35136d2de9255f%2Foof_time_id_scores.png?generation=1735194303166519&amp;alt=media\" alt=\"time_id scores\"></p>",
      "votes": null,
      "replies": [
        {
          "id": 3081484,
          "author_name": "rulangu",
          "author_url": "",
          "post_date": "12/26/2024 19:33:56",
          "content": "<p>dropping time id did drop my xgboost model by quite a bit on the leaderboard score, so i figure it's probably important.</p>\n<p>any insights on keeping symbol id as a feature, or is that somehow already embedded into the features?</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "3080924": "Does anyone have any insights on using time_id and/or symbol_id as additional features (on top of the 79 raw features that we're already given)? Or are these things somehow already embedded into the existing feature set? Are there any posts with analysis on this?",
    "3081050": "Those are the oof scores per `time_id` for me. I think using `time_id` helps a bit since models can distinguish regular trading hours, pre-market and after hours.\n![time_id scores](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F2706866%2Ff33b6519207b66811c35136d2de9255f%2Foof_time_id_scores.png?generation=1735194303166519&alt=media)",
    "3081484": "dropping time id did drop my xgboost model by quite a bit on the leaderboard score, so i figure it's probably important.\n\nany insights on keeping symbol id as a feature, or is that somehow already embedded into the features?"
  },
  "source": "meta"
}