Hello all,
\nWishing you the best for the competition! This is my favourite data science topic, as a Finance post-grad and I am excited to witness such a competition after a long wait! Hope the below materials help you onboard well and efficiently-
\nWishing you the best for the assignment and happy learning!
", "messages": [ { "id": "3207544", "postDate": "05/22/2025 23:29:45", "content": "Hello all,
\nWishing you the best for the competition! This is my favourite data science topic, as a Finance post-grad and I am excited to witness such a competition after a long wait! Hope the below materials help you onboard well and efficiently-
\nWishing you the best for the assignment and happy learning!
", "rawMarkdown": "Hello all,\n\nWishing you the best for the competition! This is my favourite data science topic, as a Finance post-grad and I am excited to witness such a competition after a long wait! Hope the below materials help you onboard well and efficiently-\n\n# [Jane Street Market Prediction](https://www.kaggle.com/competitions/jane-street-market-prediction/overview)\n## Most voted kernels\n- https://www.kaggle.com/code/carlmcbrideellis/jane-street-eda-of-day-0-and-feature-importance\n- https://www.kaggle.com/code/hamditarek/market-prediction-xgboost-with-gpu-fit-in-1min\n- https://www.kaggle.com/code/aimind/bottleneck-encoder-mlp-keras-tuner-8601c5\n- https://www.kaggle.com/code/gogo827jz/jane-street-neural-network-starter\n- https://www.kaggle.com/code/muhammadmelsherbini/jane-street-extensive-eda-pca-starter\n- https://www.kaggle.com/code/tarlannazarov/own-jane-street-with-keras-nn\n- https://www.kaggle.com/code/odins0n/exploring-time-series-plots-beginners-guide\n- https://www.kaggle.com/code/jorijnsmit/found-the-holy-grail-grouptimeseriessplit\n- https://www.kaggle.com/code/snippsy/bottleneck-encoder-mlp-keras-tuner\n- https://www.kaggle.com/code/a763337092/blending-tensorflow-and-pytorch\n\n## Top ranked solutions\n- https://www.kaggle.com/competitions/jane-street-market-prediction/discussion/224348 -- rank 1\n- https://www.kaggle.com/competitions/jane-street-market-prediction/discussion/224713 -- rank 3\n- https://www.kaggle.com/competitions/jane-street-market-prediction/discussion/226837 -- rank 10\n- https://www.kaggle.com/competitions/jane-street-market-prediction/discussion/269181 -- rank 15\n- https://www.kaggle.com/competitions/jane-street-market-prediction/discussion/224079 -- rank 23\n\n# [Jane Street Real-Time Market Data Forecasting](https://www.kaggle.com/competitions/jane-street-real-time-market-data-forecasting)\n## Most voted kernels\n- https://www.kaggle.com/code/yuanzhezhou/jane-street-baseline-lgb-xgb-and-catboost\n- https://www.kaggle.com/code/allegich/jane-street-time-series-analysis-eda-ensemble\n- https://www.kaggle.com/code/voix97/jane-street-rmf-inference-nn-xgb\n- https://www.kaggle.com/code/motono0223/eda-jane-street-real-time-market-data-forecasting\n- https://www.kaggle.com/code/yongsukprasertsuk/0-008-js-rmf-ensemble-xgb-nn-tabm-ridge\n- https://www.kaggle.com/code/eivolkova/public-lb-6th\n- https://www.kaggle.com/code/simonedegasperis/online-retrain-poc\n- https://www.kaggle.com/code/motono0223/js24-inference-gbdt-with-lags-singlemodel\n\n# [Optiver- Trading at the Close](https://www.kaggle.com/competitions/optiver-trading-at-the-close)\n## Kernels- most voted \n1. https://www.kaggle.com/code/ravi20076/optiver-baseline-models\n2. https://www.kaggle.com/code/yuanzhezhou/baseline-lgb-xgb-and-catboost\n3. https://www.kaggle.com/code/a27182818/explain-the-data-lightgbm-baseline\n4. https://www.kaggle.com/code/lblhandsome/optiver-robust-best-single-model\n5. https://www.kaggle.com/code/verracodeguacas/fold-cv\n6. https://www.kaggle.com/code/peizhengwang/best-public-score\n7. https://www.kaggle.com/code/siddhvr/optiver-trading-at-the-close-sub\n\n## High scoring approaches and discussions-\n1. https://www.kaggle.com/competitions/optiver-trading-at-the-close/discussion/487446\n2. https://www.kaggle.com/competitions/optiver-trading-at-the-close/discussion/486868\n3. https://www.kaggle.com/competitions/optiver-trading-at-the-close/discussion/462653\n4. https://www.kaggle.com/competitions/optiver-trading-at-the-close/discussion/486086\n\n# [Enefit - Predict Energy Behavior of Prosumers](https://www.kaggle.com/competitions/predict-energy-behavior-of-prosumers)\n## Kernels- most voted \n1. https://www.kaggle.com/code/rafiko1/enefit-xgboost-starter\n2. https://www.kaggle.com/code/vitalykudelya/explain-dataset-and-baseline\n3. https://www.kaggle.com/code/vincentschuler/enefit-baseline-cross-validation\n4. https://www.kaggle.com/code/greysky/enefit-generic-notebook\n5. https://www.kaggle.com/code/vitalykudelya/enefit-target-diff\n6. https://www.kaggle.com/code/ahmedabdulwahab/pandas-data-description-and-starters-guide\n\n## High scoring approaches and discussions-\n1. https://www.kaggle.com/competitions/predict-energy-behavior-of-prosumers/discussion/472793\n2. https://www.kaggle.com/competitions/predict-energy-behavior-of-prosumers/discussion/499938\n3. https://www.kaggle.com/competitions/predict-energy-behavior-of-prosumers/discussion/499397\n4. https://www.kaggle.com/competitions/predict-energy-behavior-of-prosumers/discussion/499649\n5. https://www.kaggle.com/competitions/predict-energy-behavior-of-prosumers/discussion/472537\n6. https://www.kaggle.com/competitions/predict-energy-behavior-of-prosumers/discussion/472537\n\n# Playground Time Series Forecasting challenges\n## [Season3-Episode 20](https://www.kaggle.com/competitions/playground-series-s3e20)\n### Kernels- most voted \n1. https://www.kaggle.com/code/ambrosm/pss3e20-eda-which-makes-sense\n2. https://www.kaggle.com/code/kacperrabczewski/rwanda-co2-step-by-step-guide\n3. https://www.kaggle.com/code/yaaangzhou/pg-s3-e20-eda-modeling\n4. https://www.kaggle.com/code/dmitryuarov/ps3e20-rwanda-emission-advanced-fe-20-88\n5. https://www.kaggle.com/code/iqbalsyahakbar/ps3e20-time-series-for-beginners\n\n## [Season3-Episode 19](https://www.kaggle.com/competitions/playground-series-s3e19)\n### Kernels- most voted \n1. https://www.kaggle.com/code/tumpanjawat/s3e19-course-eda-fe-lightgbm\n2. https://www.kaggle.com/code/iqbalsyahakbar/ps3e19-time-series-for-beginners\n3. https://www.kaggle.com/code/ivyzang/1st-place-solution-less-is-more\n4. https://www.kaggle.com/code/tetsutani/ps3e19-eda-ensemble-ml-pipeline-rnn-by-skorch\n5. https://www.kaggle.com/code/kacperrabczewski/last-minute-forecasting\n\n## [TPS- September2022](https://www.kaggle.com/competitions/tabular-playground-series-sep-2022)\n### Kernels- most voted \n1. https://www.kaggle.com/code/azminetoushikwasi/time-series-analysis-forecasting\n2. https://www.kaggle.com/code/cabaxiom/tps-sep-22-eda-and-linear-regression-baseline\n3. https://www.kaggle.com/code/khashayarrahimi94/why-you-should-not-use-correlation\n4. https://www.kaggle.com/code/vencerlanz09/tps-eda-9-models-explanation\n5. https://www.kaggle.com/code/samuelcortinhas/tps-sept-22-timeseries-analysis\n\n## [TPS- January2022](https://www.kaggle.com/competitions/tabular-playground-series-jan-2022)\n### Kernels- most voted \n1. https://www.kaggle.com/code/ambrosm/tpsjan22-03-linear-model\n2. https://www.kaggle.com/code/ambrosm/tpsjan22-01-eda-which-makes-sense\n3. https://www.kaggle.com/code/usharengaraju/tensorflow-tf-data-keraspreprocessinglayers-w-b\n4. https://www.kaggle.com/code/teckmengwong/tps2201-hybrid-time-series\n5. https://www.kaggle.com/code/lordozvlad/tps-jan-fast-pycaret-with-scikit-learn-intelex\n\n# [GoDaddy - Microbusiness Density Forecasting](https://www.kaggle.com/competitions/godaddy-microbusiness-density-forecasting/overview)\n## Kernels- most voted \n1. https://www.kaggle.com/code/titericz/better-xgb-baseline\n2. https://www.kaggle.com/code/cdeotte/linear-regression-baseline-lb-1-092\n3. https://www.kaggle.com/code/tanmay111999/gdmbf-ar-ma-arma-arima-sarima-auto-arima\n4. https://www.kaggle.com/code/cdeotte/seasonal-model-with-validation-lb-1-091\n5. https://www.kaggle.com/code/kimtaehun/complete-baseline-code-with-various-ml-model\n\n## High scoring approaches and discussions-\n1. https://www.kaggle.com/competitions/godaddy-microbusiness-density-forecasting/discussion/395131 -- rank1\n2. https://www.kaggle.com/competitions/godaddy-microbusiness-density-forecasting/discussion/395264 -- rank2\n3. https://www.kaggle.com/competitions/godaddy-microbusiness-density-forecasting/discussion/418287 -- rank3\n4. https://www.kaggle.com/competitions/godaddy-microbusiness-density-forecasting/discussion/394821 -- rank4\n\n# Miscellaneous kernel references on time series- \n1. https://www.kaggle.com/code/ravi20076/optiver-baseline-models\n2. https://www.kaggle.com/code/kanncaa1/time-series-prediction-tutorial-with-eda\n3. https://www.kaggle.com/code/freespirit08/time-series-for-beginners-with-arima\n4. https://www.kaggle.com/code/robikscube/time-series-forecasting-with-prophet\n5. https://www.kaggle.com/code/iamleonie/intro-to-time-series-forecasting\n6. https://www.kaggle.com/code/robikscube/time-series-forecasting-with-machine-learning-yt\n7. https://www.kaggle.com/code/rohanrao/a-modern-time-series-tutorial\n8. https://www.kaggle.com/code/cdeotte/time-series-eda\n9. https://www.kaggle.com/code/janiobachmann/s-p-500-time-series-forecasting-with-prophet\n10. https://www.kaggle.com/code/robikscube/pt2-time-series-forecasting-with-xgboost\n\nWishing you the best for the assignment and happy learning!", "votes": null }, { "id": "3209385", "postDate": "05/25/2025 17:37:42", "content": "Thanks for sharing
", "rawMarkdown": "Thanks for sharing", "votes": null }, { "id": "3209407", "postDate": "05/25/2025 18:19:20", "content": "Excellent compilation @ravi20076! This is exactly what the community needs. Building on your fantastic resource list, I'd like to add some crypto-specific considerations and advanced techniques that could be particularly valuable for this competition:
\n1. Microstructure Features for Crypto Markets:
\n2. Volatility Regime Detection:
\n\n sklearn.mixture GaussianMixture\n\nvolatility_regimes = GaussianMixture(n_components=).fit(rolling_volatility)\n3. Multi-timeframe Feature Engineering:
\n1. Transformer-based Approaches:
\n2. Graph Neural Networks:
\n3. Ensemble Strategies:
\n1. Time-Aware Cross-Validation:
\n\n sklearn.model_selection TimeSeriesSplit\n\n2. Walk-Forward Analysis:
\n1. Feature Selection for High-Frequency Data:
\n2. Risk Management Integration:
\n3. Computational Efficiency:
\nCrypto-Specific Papers:
\nAdvanced Time Series Libraries:
\ndarts - Modern forecasting librarysktime - Unified time series MLtslearn - Time series clustering/classificationThis competition is a perfect opportunity to combine traditional quantitative finance techniques with modern ML approaches. The key is balancing model complexity with interpretability, especially given the unique characteristics of crypto markets.
\nHappy modeling, and may the best algorithm win! 🏆
", "rawMarkdown": "Excellent compilation @ravi20076! This is exactly what the community needs. Building on your fantastic resource list, I'd like to add some **crypto-specific considerations** and **advanced techniques** that could be particularly valuable for this competition:\n\n## 🚀 **Crypto-Specific Enhancements**\n\n**1. Microstructure Features for Crypto Markets:**\n- Order book imbalance ratios\n- Bid-ask spread dynamics\n- Volume-weighted average price (VWAP) deviations\n- Tick-by-tick momentum indicators\n- Cross-exchange arbitrage signals\n\n**2. Volatility Regime Detection:**\n```python\n# Regime-switching models for crypto volatility\nfrom sklearn.mixture import GaussianMixture\n# Detect high/low volatility regimes\nvolatility_regimes = GaussianMixture(n_components=3).fit(rolling_volatility)\n```\n\n**3. Multi-timeframe Feature Engineering:**\n- 1-minute, 5-minute, 15-minute, 1-hour aggregations\n- Fractal dimension analysis across timeframes\n- Wavelet decomposition for multi-resolution analysis\n\n## 🧠 **Advanced Model Architectures**\n\n**1. Transformer-based Approaches:**\n- Temporal Fusion Transformers (TFT) for multi-horizon forecasting\n- Attention mechanisms for feature importance\n- Positional encoding for time-aware learning\n\n**2. Graph Neural Networks:**\n- Model cross-asset correlations as graph structures\n- Capture market contagion effects\n- Dynamic graph learning for evolving relationships\n\n**3. Ensemble Strategies:**\n- Bayesian Model Averaging with uncertainty quantification\n- Dynamic ensemble weights based on market regime\n- Multi-objective optimization (return vs. risk)\n\n## 📊 **Validation Strategies for Crypto**\n\n**1. Time-Aware Cross-Validation:**\n```python\n# Purged Group Time Series Split\nfrom sklearn.model_selection import TimeSeriesSplit\n# Account for overlapping predictions and market microstructure\n```\n\n**2. Walk-Forward Analysis:**\n- Rolling window retraining\n- Adaptive lookback periods\n- Online learning with concept drift detection\n\n## 🔧 **Implementation Tips**\n\n**1. Feature Selection for High-Frequency Data:**\n- Mutual Information with time lags\n- Recursive Feature Elimination with cross-validation\n- SHAP values for interpretability\n\n**2. Risk Management Integration:**\n- Kelly Criterion for position sizing\n- Value-at-Risk (VaR) constraints\n- Maximum Drawdown controls\n\n**3. Computational Efficiency:**\n- Polars for fast data processing\n- Numba JIT compilation for custom indicators\n- GPU acceleration with CuPy/RAPIDS\n\n## 📚 **Additional Resources**\n\n**Crypto-Specific Papers:**\n- \"Deep Learning for Cryptocurrency Forecasting\" (2021)\n- \"High-Frequency Trading in Cryptocurrency Markets\" (2022)\n- \"Market Microstructure in Digital Asset Markets\" (2023)\n\n**Advanced Time Series Libraries:**\n- `darts` - Modern forecasting library\n- `sktime` - Unified time series ML\n- `tslearn` - Time series clustering/classification\n\nThis competition is a perfect opportunity to combine traditional quantitative finance techniques with modern ML approaches. The key is balancing model complexity with interpretability, especially given the unique characteristics of crypto markets.\n\nHappy modeling, and may the best algorithm win! 🏆", "votes": null }, { "id": "3209422", "postDate": "05/25/2025 19:04:01", "content": "@harshithvaddiparthy thanks for the added resources.
\nConsidering that the test data is not in temporal order, do you think these models will work? Perhaps they will work well for a good offline CV scheme though.
This is a great compilation! Thank you very much for sharing!
", "rawMarkdown": "This is a great compilation! Thank you very much for sharing!", "votes": null }, { "id": "3209766", "postDate": "05/26/2025 09:32:25", "content": "Thanks for sharing
", "rawMarkdown": "Thanks for sharing", "votes": null }, { "id": "3210113", "postDate": "05/26/2025 18:30:39", "content": "Amazing work bro, thanks for your information. Do you have winner-position in leaderboard for Playground Time Series Forecasting challenges?
", "rawMarkdown": "Amazing work bro, thanks for your information. Do you have winner-position in leaderboard for Playground Time Series Forecasting challenges?", "votes": null }, { "id": "3210906", "postDate": "05/27/2025 20:46:07", "content": "This is super helpful, thank you!
", "rawMarkdown": "This is super helpful, thank you!", "votes": null }, { "id": "3211361", "postDate": "05/28/2025 10:45:43", "content": "Thanks for the educational resources!
", "rawMarkdown": "Thanks for the educational resources!", "votes": null }, { "id": "3212061", "postDate": "05/29/2025 08:54:55", "content": "kind of fantastic
", "rawMarkdown": "kind of fantastic", "votes": null }, { "id": "3213509", "postDate": "05/30/2025 05:32:06", "content": "This is something that I really needed 🙌. Thank you sir..!
", "rawMarkdown": "This is something that I really needed 🙌. Thank you sir..!", "votes": null }, { "id": "3218368", "postDate": "06/06/2025 05:45:28", "content": "Thanks for sharing!!
", "rawMarkdown": "Thanks for sharing!!", "votes": null }, { "id": "3221918", "postDate": "06/11/2025 15:40:42", "content": "Thanks so much for the sharing, bro! Your compilation really helped me a lot!
", "rawMarkdown": "Thanks so much for the sharing, bro! Your compilation really helped me a lot!", "votes": null }, { "id": "3239660", "postDate": "07/03/2025 03:07:19", "content": "Thanks for sharing
", "rawMarkdown": "Thanks for sharing", "votes": null }, { "id": "3247065", "postDate": "07/12/2025 03:22:50", "content": "The Jane street code is fascinating to review!
", "rawMarkdown": "The Jane street code is fascinating to review!", "votes": null } ], "comments": [ { "id": 3209385, "author_name": "sharmajicoder", "author_url": "", "post_date": "05/25/2025 17:37:42", "content": "Thanks for sharing
", "votes": null, "replies": [] }, { "id": 3209407, "author_name": "harshithvaddiparthy", "author_url": "", "post_date": "05/25/2025 18:19:20", "content": "Excellent compilation @ravi20076! This is exactly what the community needs. Building on your fantastic resource list, I'd like to add some crypto-specific considerations and advanced techniques that could be particularly valuable for this competition:
\n1. Microstructure Features for Crypto Markets:
\n2. Volatility Regime Detection:
\n\n sklearn.mixture GaussianMixture\n\nvolatility_regimes = GaussianMixture(n_components=).fit(rolling_volatility)\n3. Multi-timeframe Feature Engineering:
\n1. Transformer-based Approaches:
\n2. Graph Neural Networks:
\n3. Ensemble Strategies:
\n1. Time-Aware Cross-Validation:
\n\n sklearn.model_selection TimeSeriesSplit\n\n2. Walk-Forward Analysis:
\n1. Feature Selection for High-Frequency Data:
\n2. Risk Management Integration:
\n3. Computational Efficiency:
\nCrypto-Specific Papers:
\nAdvanced Time Series Libraries:
\ndarts - Modern forecasting librarysktime - Unified time series MLtslearn - Time series clustering/classificationThis competition is a perfect opportunity to combine traditional quantitative finance techniques with modern ML approaches. The key is balancing model complexity with interpretability, especially given the unique characteristics of crypto markets.
\nHappy modeling, and may the best algorithm win! 🏆
", "votes": null, "replies": [ { "id": 3209422, "author_name": "ravi20076", "author_url": "", "post_date": "05/25/2025 19:04:01", "content": "@harshithvaddiparthy thanks for the added resources.
\nConsidering that the test data is not in temporal order, do you think these models will work? Perhaps they will work well for a good offline CV scheme though.
This is a great compilation! Thank you very much for sharing!
", "votes": null, "replies": [] }, { "id": 3209766, "author_name": "nancyalaswad90", "author_url": "", "post_date": "05/26/2025 09:32:25", "content": "Thanks for sharing
", "votes": null, "replies": [] }, { "id": 3210113, "author_name": "poitrew", "author_url": "", "post_date": "05/26/2025 18:30:39", "content": "Amazing work bro, thanks for your information. Do you have winner-position in leaderboard for Playground Time Series Forecasting challenges?
", "votes": null, "replies": [] }, { "id": 3210906, "author_name": "zakrashad", "author_url": "", "post_date": "05/27/2025 20:46:07", "content": "This is super helpful, thank you!
", "votes": null, "replies": [] }, { "id": 3211361, "author_name": "olowedaniel", "author_url": "", "post_date": "05/28/2025 10:45:43", "content": "Thanks for the educational resources!
", "votes": null, "replies": [] }, { "id": 3212061, "author_name": "chatgptforus", "author_url": "", "post_date": "05/29/2025 08:54:55", "content": "kind of fantastic
", "votes": null, "replies": [] }, { "id": 3213509, "author_name": "vishalpainjane", "author_url": "", "post_date": "05/30/2025 05:32:06", "content": "This is something that I really needed 🙌. Thank you sir..!
", "votes": null, "replies": [] }, { "id": 3218368, "author_name": "xiaoleilian", "author_url": "", "post_date": "06/06/2025 05:45:28", "content": "Thanks for sharing!!
", "votes": null, "replies": [] }, { "id": 3221918, "author_name": "foursevendong", "author_url": "", "post_date": "06/11/2025 15:40:42", "content": "Thanks so much for the sharing, bro! Your compilation really helped me a lot!
", "votes": null, "replies": [] }, { "id": 3239660, "author_name": "yingjunmao", "author_url": "", "post_date": "07/03/2025 03:07:19", "content": "Thanks for sharing
", "votes": null, "replies": [] }, { "id": 3247065, "author_name": "taylorsamarel", "author_url": "", "post_date": "07/12/2025 03:22:50", "content": "The Jane street code is fascinating to review!
", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "3207544": "Hello all,\n\nWishing you the best for the competition! This is my favourite data science topic, as a Finance post-grad and I am excited to witness such a competition after a long wait! Hope the below materials help you onboard well and efficiently-\n\n# [Jane Street Market Prediction](https://www.kaggle.com/competitions/jane-street-market-prediction/overview)\n## Most voted kernels\n- https://www.kaggle.com/code/carlmcbrideellis/jane-street-eda-of-day-0-and-feature-importance\n- https://www.kaggle.com/code/hamditarek/market-prediction-xgboost-with-gpu-fit-in-1min\n- https://www.kaggle.com/code/aimind/bottleneck-encoder-mlp-keras-tuner-8601c5\n- https://www.kaggle.com/code/gogo827jz/jane-street-neural-network-starter\n- https://www.kaggle.com/code/muhammadmelsherbini/jane-street-extensive-eda-pca-starter\n- https://www.kaggle.com/code/tarlannazarov/own-jane-street-with-keras-nn\n- https://www.kaggle.com/code/odins0n/exploring-time-series-plots-beginners-guide\n- https://www.kaggle.com/code/jorijnsmit/found-the-holy-grail-grouptimeseriessplit\n- https://www.kaggle.com/code/snippsy/bottleneck-encoder-mlp-keras-tuner\n- https://www.kaggle.com/code/a763337092/blending-tensorflow-and-pytorch\n\n## Top ranked solutions\n- https://www.kaggle.com/competitions/jane-street-market-prediction/discussion/224348 -- rank 1\n- https://www.kaggle.com/competitions/jane-street-market-prediction/discussion/224713 -- rank 3\n- https://www.kaggle.com/competitions/jane-street-market-prediction/discussion/226837 -- rank 10\n- https://www.kaggle.com/competitions/jane-street-market-prediction/discussion/269181 -- rank 15\n- https://www.kaggle.com/competitions/jane-street-market-prediction/discussion/224079 -- rank 23\n\n# [Jane Street Real-Time Market Data Forecasting](https://www.kaggle.com/competitions/jane-street-real-time-market-data-forecasting)\n## Most voted kernels\n- https://www.kaggle.com/code/yuanzhezhou/jane-street-baseline-lgb-xgb-and-catboost\n- https://www.kaggle.com/code/allegich/jane-street-time-series-analysis-eda-ensemble\n- https://www.kaggle.com/code/voix97/jane-street-rmf-inference-nn-xgb\n- https://www.kaggle.com/code/motono0223/eda-jane-street-real-time-market-data-forecasting\n- https://www.kaggle.com/code/yongsukprasertsuk/0-008-js-rmf-ensemble-xgb-nn-tabm-ridge\n- https://www.kaggle.com/code/eivolkova/public-lb-6th\n- https://www.kaggle.com/code/simonedegasperis/online-retrain-poc\n- https://www.kaggle.com/code/motono0223/js24-inference-gbdt-with-lags-singlemodel\n\n# [Optiver- Trading at the Close](https://www.kaggle.com/competitions/optiver-trading-at-the-close)\n## Kernels- most voted \n1. https://www.kaggle.com/code/ravi20076/optiver-baseline-models\n2. https://www.kaggle.com/code/yuanzhezhou/baseline-lgb-xgb-and-catboost\n3. https://www.kaggle.com/code/a27182818/explain-the-data-lightgbm-baseline\n4. https://www.kaggle.com/code/lblhandsome/optiver-robust-best-single-model\n5. https://www.kaggle.com/code/verracodeguacas/fold-cv\n6. https://www.kaggle.com/code/peizhengwang/best-public-score\n7. https://www.kaggle.com/code/siddhvr/optiver-trading-at-the-close-sub\n\n## High scoring approaches and discussions-\n1. https://www.kaggle.com/competitions/optiver-trading-at-the-close/discussion/487446\n2. https://www.kaggle.com/competitions/optiver-trading-at-the-close/discussion/486868\n3. https://www.kaggle.com/competitions/optiver-trading-at-the-close/discussion/462653\n4. https://www.kaggle.com/competitions/optiver-trading-at-the-close/discussion/486086\n\n# [Enefit - Predict Energy Behavior of Prosumers](https://www.kaggle.com/competitions/predict-energy-behavior-of-prosumers)\n## Kernels- most voted \n1. https://www.kaggle.com/code/rafiko1/enefit-xgboost-starter\n2. https://www.kaggle.com/code/vitalykudelya/explain-dataset-and-baseline\n3. https://www.kaggle.com/code/vincentschuler/enefit-baseline-cross-validation\n4. https://www.kaggle.com/code/greysky/enefit-generic-notebook\n5. https://www.kaggle.com/code/vitalykudelya/enefit-target-diff\n6. https://www.kaggle.com/code/ahmedabdulwahab/pandas-data-description-and-starters-guide\n\n## High scoring approaches and discussions-\n1. https://www.kaggle.com/competitions/predict-energy-behavior-of-prosumers/discussion/472793\n2. https://www.kaggle.com/competitions/predict-energy-behavior-of-prosumers/discussion/499938\n3. https://www.kaggle.com/competitions/predict-energy-behavior-of-prosumers/discussion/499397\n4. https://www.kaggle.com/competitions/predict-energy-behavior-of-prosumers/discussion/499649\n5. https://www.kaggle.com/competitions/predict-energy-behavior-of-prosumers/discussion/472537\n6. https://www.kaggle.com/competitions/predict-energy-behavior-of-prosumers/discussion/472537\n\n# Playground Time Series Forecasting challenges\n## [Season3-Episode 20](https://www.kaggle.com/competitions/playground-series-s3e20)\n### Kernels- most voted \n1. https://www.kaggle.com/code/ambrosm/pss3e20-eda-which-makes-sense\n2. https://www.kaggle.com/code/kacperrabczewski/rwanda-co2-step-by-step-guide\n3. https://www.kaggle.com/code/yaaangzhou/pg-s3-e20-eda-modeling\n4. https://www.kaggle.com/code/dmitryuarov/ps3e20-rwanda-emission-advanced-fe-20-88\n5. https://www.kaggle.com/code/iqbalsyahakbar/ps3e20-time-series-for-beginners\n\n## [Season3-Episode 19](https://www.kaggle.com/competitions/playground-series-s3e19)\n### Kernels- most voted \n1. https://www.kaggle.com/code/tumpanjawat/s3e19-course-eda-fe-lightgbm\n2. https://www.kaggle.com/code/iqbalsyahakbar/ps3e19-time-series-for-beginners\n3. https://www.kaggle.com/code/ivyzang/1st-place-solution-less-is-more\n4. https://www.kaggle.com/code/tetsutani/ps3e19-eda-ensemble-ml-pipeline-rnn-by-skorch\n5. https://www.kaggle.com/code/kacperrabczewski/last-minute-forecasting\n\n## [TPS- September2022](https://www.kaggle.com/competitions/tabular-playground-series-sep-2022)\n### Kernels- most voted \n1. https://www.kaggle.com/code/azminetoushikwasi/time-series-analysis-forecasting\n2. https://www.kaggle.com/code/cabaxiom/tps-sep-22-eda-and-linear-regression-baseline\n3. https://www.kaggle.com/code/khashayarrahimi94/why-you-should-not-use-correlation\n4. https://www.kaggle.com/code/vencerlanz09/tps-eda-9-models-explanation\n5. https://www.kaggle.com/code/samuelcortinhas/tps-sept-22-timeseries-analysis\n\n## [TPS- January2022](https://www.kaggle.com/competitions/tabular-playground-series-jan-2022)\n### Kernels- most voted \n1. https://www.kaggle.com/code/ambrosm/tpsjan22-03-linear-model\n2. https://www.kaggle.com/code/ambrosm/tpsjan22-01-eda-which-makes-sense\n3. https://www.kaggle.com/code/usharengaraju/tensorflow-tf-data-keraspreprocessinglayers-w-b\n4. https://www.kaggle.com/code/teckmengwong/tps2201-hybrid-time-series\n5. https://www.kaggle.com/code/lordozvlad/tps-jan-fast-pycaret-with-scikit-learn-intelex\n\n# [GoDaddy - Microbusiness Density Forecasting](https://www.kaggle.com/competitions/godaddy-microbusiness-density-forecasting/overview)\n## Kernels- most voted \n1. https://www.kaggle.com/code/titericz/better-xgb-baseline\n2. https://www.kaggle.com/code/cdeotte/linear-regression-baseline-lb-1-092\n3. https://www.kaggle.com/code/tanmay111999/gdmbf-ar-ma-arma-arima-sarima-auto-arima\n4. https://www.kaggle.com/code/cdeotte/seasonal-model-with-validation-lb-1-091\n5. https://www.kaggle.com/code/kimtaehun/complete-baseline-code-with-various-ml-model\n\n## High scoring approaches and discussions-\n1. https://www.kaggle.com/competitions/godaddy-microbusiness-density-forecasting/discussion/395131 -- rank1\n2. https://www.kaggle.com/competitions/godaddy-microbusiness-density-forecasting/discussion/395264 -- rank2\n3. https://www.kaggle.com/competitions/godaddy-microbusiness-density-forecasting/discussion/418287 -- rank3\n4. https://www.kaggle.com/competitions/godaddy-microbusiness-density-forecasting/discussion/394821 -- rank4\n\n# Miscellaneous kernel references on time series- \n1. https://www.kaggle.com/code/ravi20076/optiver-baseline-models\n2. https://www.kaggle.com/code/kanncaa1/time-series-prediction-tutorial-with-eda\n3. https://www.kaggle.com/code/freespirit08/time-series-for-beginners-with-arima\n4. https://www.kaggle.com/code/robikscube/time-series-forecasting-with-prophet\n5. https://www.kaggle.com/code/iamleonie/intro-to-time-series-forecasting\n6. https://www.kaggle.com/code/robikscube/time-series-forecasting-with-machine-learning-yt\n7. https://www.kaggle.com/code/rohanrao/a-modern-time-series-tutorial\n8. https://www.kaggle.com/code/cdeotte/time-series-eda\n9. https://www.kaggle.com/code/janiobachmann/s-p-500-time-series-forecasting-with-prophet\n10. https://www.kaggle.com/code/robikscube/pt2-time-series-forecasting-with-xgboost\n\nWishing you the best for the assignment and happy learning!", "3209385": "Thanks for sharing", "3209407": "Excellent compilation @ravi20076! This is exactly what the community needs. Building on your fantastic resource list, I'd like to add some **crypto-specific considerations** and **advanced techniques** that could be particularly valuable for this competition:\n\n## 🚀 **Crypto-Specific Enhancements**\n\n**1. Microstructure Features for Crypto Markets:**\n- Order book imbalance ratios\n- Bid-ask spread dynamics\n- Volume-weighted average price (VWAP) deviations\n- Tick-by-tick momentum indicators\n- Cross-exchange arbitrage signals\n\n**2. Volatility Regime Detection:**\n```python\n# Regime-switching models for crypto volatility\nfrom sklearn.mixture import GaussianMixture\n# Detect high/low volatility regimes\nvolatility_regimes = GaussianMixture(n_components=3).fit(rolling_volatility)\n```\n\n**3. Multi-timeframe Feature Engineering:**\n- 1-minute, 5-minute, 15-minute, 1-hour aggregations\n- Fractal dimension analysis across timeframes\n- Wavelet decomposition for multi-resolution analysis\n\n## 🧠 **Advanced Model Architectures**\n\n**1. Transformer-based Approaches:**\n- Temporal Fusion Transformers (TFT) for multi-horizon forecasting\n- Attention mechanisms for feature importance\n- Positional encoding for time-aware learning\n\n**2. Graph Neural Networks:**\n- Model cross-asset correlations as graph structures\n- Capture market contagion effects\n- Dynamic graph learning for evolving relationships\n\n**3. Ensemble Strategies:**\n- Bayesian Model Averaging with uncertainty quantification\n- Dynamic ensemble weights based on market regime\n- Multi-objective optimization (return vs. risk)\n\n## 📊 **Validation Strategies for Crypto**\n\n**1. Time-Aware Cross-Validation:**\n```python\n# Purged Group Time Series Split\nfrom sklearn.model_selection import TimeSeriesSplit\n# Account for overlapping predictions and market microstructure\n```\n\n**2. Walk-Forward Analysis:**\n- Rolling window retraining\n- Adaptive lookback periods\n- Online learning with concept drift detection\n\n## 🔧 **Implementation Tips**\n\n**1. Feature Selection for High-Frequency Data:**\n- Mutual Information with time lags\n- Recursive Feature Elimination with cross-validation\n- SHAP values for interpretability\n\n**2. Risk Management Integration:**\n- Kelly Criterion for position sizing\n- Value-at-Risk (VaR) constraints\n- Maximum Drawdown controls\n\n**3. Computational Efficiency:**\n- Polars for fast data processing\n- Numba JIT compilation for custom indicators\n- GPU acceleration with CuPy/RAPIDS\n\n## 📚 **Additional Resources**\n\n**Crypto-Specific Papers:**\n- \"Deep Learning for Cryptocurrency Forecasting\" (2021)\n- \"High-Frequency Trading in Cryptocurrency Markets\" (2022)\n- \"Market Microstructure in Digital Asset Markets\" (2023)\n\n**Advanced Time Series Libraries:**\n- `darts` - Modern forecasting library\n- `sktime` - Unified time series ML\n- `tslearn` - Time series clustering/classification\n\nThis competition is a perfect opportunity to combine traditional quantitative finance techniques with modern ML approaches. The key is balancing model complexity with interpretability, especially given the unique characteristics of crypto markets.\n\nHappy modeling, and may the best algorithm win! 🏆", "3209422": "harshithvaddiparthy thanks for the added resources. \nConsidering that the test data is not in temporal order, do you think these models will work? Perhaps they will work well for a good offline CV scheme though.", "3209741": "This is a great compilation! Thank you very much for sharing!", "3209766": "Thanks for sharing", "3210113": "Amazing work bro, thanks for your information. Do you have winner-position in leaderboard for Playground Time Series Forecasting challenges?", "3210906": "This is super helpful, thank you!", "3211361": "Thanks for the educational resources!", "3212061": "kind of fantastic", "3213509": "This is something that I really needed 🙌. Thank you sir..!", "3218368": "Thanks for sharing!!", "3221918": "Thanks so much for the sharing, bro! Your compilation really helped me a lot!", "3239660": "Thanks for sharing", "3247065": "The Jane street code is fascinating to review!" }, "source": "meta" }