{
  "id": 130845,
  "title": "How are predictions for non-existant games evaluated?",
  "url": "/competitions/google-cloud-ncaa-march-madness-2020-division-1-mens-tournament/discussion/130845",
  "author_name": "",
  "post_date": "2020-02-16T18:33:05.500028400Z",
  "votes": 2,
  "comment_count": 2,
  "views": 0,
  "content": "<p>I might be missing something, but if we need to predict the win probabilities for all possible matchups does this not mean that some of the matchups will never take place and not all of the predictions will be possible to evaluate?</p>\n\n<p>Will the predictions for the games that do not happen be simply disregarded?</p>",
  "messages": [
    {
      "id": "747705",
      "postDate": "02/16/2020 18:33:05",
      "content": "<p>I might be missing something, but if we need to predict the win probabilities for all possible matchups does this not mean that some of the matchups will never take place and not all of the predictions will be possible to evaluate?</p>\n\n<p>Will the predictions for the games that do not happen be simply disregarded?</p>",
      "rawMarkdown": "I might be missing something, but if we need to predict the win probabilities for all possible matchups does this not mean that some of the matchups will never take place and not all of the predictions will be possible to evaluate?\n\nWill the predictions for the games that do not happen be simply disregarded?",
      "votes": null
    },
    {
      "id": "747710",
      "postDate": "02/16/2020 18:41:12",
      "content": "<p>That's exactly right. You will only be scored on games that happen. </p>",
      "rawMarkdown": "That's exactly right. You will only be scored on games that happen.",
      "votes": null
    },
    {
      "id": "747753",
      "postDate": "02/16/2020 19:51:07",
      "content": "<p>The overall idea of the contest is to have people predict games before they are played, and then to evaluate those predictions after the outcomes are known.  There would really be only two good ways to do this - have people submit their predictions for the first 32 games, then once we know who advanced, submit their predictions for the next 16 games, and so on.  This is a logistical nightmare given the quick turnaround time required, and has the possibility to doom the entire contest if something goes wrong technically mid-tournament.  Some people would probably fail to get all their predictions submitted each time.  So years ago, we decided the best practical approach would be to have you make predictions only once, but for all possible matchups, since we won't know which matchups really happened until the games are played.  So there is a submission window spanning several days, after Selection Sunday brackets are announced and Stage 2 data is made available.  This window overlaps the \"play-in\" games where the men's field goes from 68 teams to 64 teams.  So in both the men's and women's contests, you are only scored on the 63 games that were actually played among those 64 teams, and all other predictions are ignored by the scoring metric.  All of those 63 games are played after the submission window has closed.</p>",
      "rawMarkdown": "The overall idea of the contest is to have people predict games before they are played, and then to evaluate those predictions after the outcomes are known.  There would really be only two good ways to do this - have people submit their predictions for the first 32 games, then once we know who advanced, submit their predictions for the next 16 games, and so on.  This is a logistical nightmare given the quick turnaround time required, and has the possibility to doom the entire contest if something goes wrong technically mid-tournament.  Some people would probably fail to get all their predictions submitted each time.  So years ago, we decided the best practical approach would be to have you make predictions only once, but for all possible matchups, since we won't know which matchups really happened until the games are played.  So there is a submission window spanning several days, after Selection Sunday brackets are announced and Stage 2 data is made available.  This window overlaps the \"play-in\" games where the men's field goes from 68 teams to 64 teams.  So in both the men's and women's contests, you are only scored on the 63 games that were actually played among those 64 teams, and all other predictions are ignored by the scoring metric.  All of those 63 games are played after the submission window has closed.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 747710,
      "author_name": "btesch",
      "author_url": "",
      "post_date": "02/16/2020 18:41:12",
      "content": "<p>That's exactly right. You will only be scored on games that happen. </p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 747753,
      "author_name": "jeffsonas",
      "author_url": "",
      "post_date": "02/16/2020 19:51:07",
      "content": "<p>The overall idea of the contest is to have people predict games before they are played, and then to evaluate those predictions after the outcomes are known.  There would really be only two good ways to do this - have people submit their predictions for the first 32 games, then once we know who advanced, submit their predictions for the next 16 games, and so on.  This is a logistical nightmare given the quick turnaround time required, and has the possibility to doom the entire contest if something goes wrong technically mid-tournament.  Some people would probably fail to get all their predictions submitted each time.  So years ago, we decided the best practical approach would be to have you make predictions only once, but for all possible matchups, since we won't know which matchups really happened until the games are played.  So there is a submission window spanning several days, after Selection Sunday brackets are announced and Stage 2 data is made available.  This window overlaps the \"play-in\" games where the men's field goes from 68 teams to 64 teams.  So in both the men's and women's contests, you are only scored on the 63 games that were actually played among those 64 teams, and all other predictions are ignored by the scoring metric.  All of those 63 games are played after the submission window has closed.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "747705": "I might be missing something, but if we need to predict the win probabilities for all possible matchups does this not mean that some of the matchups will never take place and not all of the predictions will be possible to evaluate?\n\nWill the predictions for the games that do not happen be simply disregarded?",
    "747710": "That's exactly right. You will only be scored on games that happen.",
    "747753": "The overall idea of the contest is to have people predict games before they are played, and then to evaluate those predictions after the outcomes are known.  There would really be only two good ways to do this - have people submit their predictions for the first 32 games, then once we know who advanced, submit their predictions for the next 16 games, and so on.  This is a logistical nightmare given the quick turnaround time required, and has the possibility to doom the entire contest if something goes wrong technically mid-tournament.  Some people would probably fail to get all their predictions submitted each time.  So years ago, we decided the best practical approach would be to have you make predictions only once, but for all possible matchups, since we won't know which matchups really happened until the games are played.  So there is a submission window spanning several days, after Selection Sunday brackets are announced and Stage 2 data is made available.  This window overlaps the \"play-in\" games where the men's field goes from 68 teams to 64 teams.  So in both the men's and women's contests, you are only scored on the 63 games that were actually played among those 64 teams, and all other predictions are ignored by the scoring metric.  All of those 63 games are played after the submission window has closed."
  },
  "source": "meta"
}