{
  "id": 541671,
  "title": "Need some clarifications on the lag data!",
  "url": "/competitions/jane-street-real-time-market-data-forecasting/discussion/541671",
  "author_name": "",
  "post_date": "2024-10-20T19:29:50.495492300Z",
  "votes": 2,
  "comment_count": 1,
  "views": 0,
  "content": "<p>The description says the lag data is lagged by 1 date_id, and will be served on time step 0. But each day for a single financial instrument, there will be around 900 time steps and that many set of 9 responders for a single symbol_id. Which one is served? </p>\n<p>Or you mean the lag data is lagged by 1 time step, and even within a single day, if we want to predict at time_id 100, the lagged responders of time_99 will be served already?</p>",
  "messages": [
    {
      "id": "3023637",
      "postDate": "10/20/2024 19:29:50",
      "content": "<p>The description says the lag data is lagged by 1 date_id, and will be served on time step 0. But each day for a single financial instrument, there will be around 900 time steps and that many set of 9 responders for a single symbol_id. Which one is served? </p>\n<p>Or you mean the lag data is lagged by 1 time step, and even within a single day, if we want to predict at time_id 100, the lagged responders of time_99 will be served already?</p>",
      "rawMarkdown": "The description says the lag data is lagged by 1 date_id, and will be served on time step 0. But each day for a single financial instrument, there will be around 900 time steps and that many set of 9 responders for a single symbol_id. Which one is served? \n\nOr you mean the lag data is lagged by 1 time step, and even within a single day, if we want to predict at time_id 100, the lagged responders of time_99 will be served already?",
      "votes": null
    },
    {
      "id": "3023717",
      "postDate": "10/20/2024 22:18:05",
      "content": "<p>at time=0 of date n+1, responders of all time steps &amp; all symbols in date n will be served. </p>\n<p>it is not lagged by one time step. It is lagged by one-day.</p>\n<p>you can check my synthetic test data kernel to understand how lags are served.</p>",
      "rawMarkdown": "at time=0 of date n+1, responders of all time steps & all symbols in date n will be served. \n\nit is not lagged by one time step. It is lagged by one-day.\n\nyou can check my synthetic test data kernel to understand how lags are served.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 3023717,
      "author_name": "shiyili",
      "author_url": "",
      "post_date": "10/20/2024 22:18:05",
      "content": "<p>at time=0 of date n+1, responders of all time steps &amp; all symbols in date n will be served. </p>\n<p>it is not lagged by one time step. It is lagged by one-day.</p>\n<p>you can check my synthetic test data kernel to understand how lags are served.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "3023637": "The description says the lag data is lagged by 1 date_id, and will be served on time step 0. But each day for a single financial instrument, there will be around 900 time steps and that many set of 9 responders for a single symbol_id. Which one is served? \n\nOr you mean the lag data is lagged by 1 time step, and even within a single day, if we want to predict at time_id 100, the lagged responders of time_99 will be served already?",
    "3023717": "at time=0 of date n+1, responders of all time steps & all symbols in date n will be served. \n\nit is not lagged by one time step. It is lagged by one-day.\n\nyou can check my synthetic test data kernel to understand how lags are served."
  },
  "source": "meta"
}