![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F175917%2F6bdebcf61e80f8020e474cad233449ca%2Fdownload%20(2).png?generation=1729845529495512&alt=media\")
I'm not very familiar with quantitative trading, so I'm interested in understanding what 'weight' and 'responder' might represent in real-world applications.
\nFor 'responder,' I assume it could relate to price change, but I’m curious about other indicators traders might focus on in practice, especially given that there are nine responder metrics in this competition.
\nAs for 'weight,' could it indicate trading volume, since higher volume could suggest easier transaction execution? Are there other indicators that might highlight an instrument's importance?
\nI'd really appreciate any insights!
", "messages": [ { "id": "3027712", "postDate": "10/25/2024 07:21:15", "content": "I'm not very familiar with quantitative trading, so I'm interested in understanding what 'weight' and 'responder' might represent in real-world applications.
\nFor 'responder,' I assume it could relate to price change, but I’m curious about other indicators traders might focus on in practice, especially given that there are nine responder metrics in this competition.
\nAs for 'weight,' could it indicate trading volume, since higher volume could suggest easier transaction execution? Are there other indicators that might highlight an instrument's importance?
\nI'd really appreciate any insights!
", "rawMarkdown": "I'm not very familiar with quantitative trading, so I'm interested in understanding what 'weight' and 'responder' might represent in real-world applications.\n\nFor 'responder,' I assume it could relate to price change, but I’m curious about other indicators traders might focus on in practice, especially given that there are nine responder metrics in this competition.\n\nAs for 'weight,' could it indicate trading volume, since higher volume could suggest easier transaction execution? Are there other indicators that might highlight an instrument's importance?\n\nI'd really appreciate any insights!", "votes": null }, { "id": "3027757", "postDate": "10/25/2024 08:43:41", "content": "My understanding is that a responder represents a metric that reflects a behaviour of a financial instrument.
\nWeight represents the importance of a prediction to the competition metric. Some dates and instruments are more important than others in determining our overall score.
\nduckdb.query(f'''\n SELECT \n -- rows with any variation of weight within a date?\n date_id, symbol_id\n FROM read_parquet('{train_path}', union_by_name=true, filename=true)\n GROUP BY date_id, symbol_id\n HAVING stddev(weight) > 0\n ''')\n\n> 0 rows\n
Thank you, that's quite helpful… So if I understand correctly responders are probably some forward returns…. right? I guess the responder 6 we need to predict should be the future return of each instrument?
", "rawMarkdown": "Thank you, that's quite helpful... So if I understand correctly responders are probably some forward returns.... right? I guess the responder 6 we need to predict should be the future return of each instrument?", "votes": null } ], "comments": [ { "id": 3027757, "author_name": "paddykb", "author_url": "", "post_date": "10/25/2024 08:43:41", "content": "My understanding is that a responder represents a metric that reflects a behaviour of a financial instrument.
\nWeight represents the importance of a prediction to the competition metric. Some dates and instruments are more important than others in determining our overall score.
\nduckdb.query(f'''\n SELECT \n -- rows with any variation of weight within a date?\n date_id, symbol_id\n FROM read_parquet('{train_path}', union_by_name=true, filename=true)\n GROUP BY date_id, symbol_id\n HAVING stddev(weight) > 0\n ''')\n\n> 0 rows\n
Thank you, that's quite helpful… So if I understand correctly responders are probably some forward returns…. right? I guess the responder 6 we need to predict should be the future return of each instrument?
", "votes": null, "replies": [] } ] } ], "raw_markdown_by_id": { "3027712": "I'm not very familiar with quantitative trading, so I'm interested in understanding what 'weight' and 'responder' might represent in real-world applications.\n\nFor 'responder,' I assume it could relate to price change, but I’m curious about other indicators traders might focus on in practice, especially given that there are nine responder metrics in this competition.\n\nAs for 'weight,' could it indicate trading volume, since higher volume could suggest easier transaction execution? Are there other indicators that might highlight an instrument's importance?\n\nI'd really appreciate any insights!", "3027757": "My understanding is that a responder represents a metric that reflects a behaviour of a financial instrument. \n\nWeight represents the importance of a prediction to the competition metric. Some dates and instruments are more important than others in determining our overall score.\n\n- weight is the same for all time_ids within a date_id\n```text\nduckdb.query(f'''\n SELECT \n -- rows with any variation of weight within a date?\n date_id, symbol_id\n FROM read_parquet('{train_path}', union_by_name=true, filename=true)\n GROUP BY date_id, symbol_id\n HAVING stddev(weight) > 0\n ''')\n\n> 0 rows\n```\n- Weight has a lot of variation day by day. Historically it has been more important to get symbol 19 right than symbol 6 :) You can also see here that new instruments (symbols) are added over time.\n - Symbols 4, 24 & 31 are first monitored at date_id = 952\n - Symbols 6, 18 & 32 are first monitored at date_id = 1063\n\n.png?generation=1729845529495512&alt=media)", "3028799": "Thank you, that's quite helpful... So if I understand correctly responders are probably some forward returns.... right? I guess the responder 6 we need to predict should be the future return of each instrument?" }, "source": "meta" }