The market is projected to reach USD 1,074 Million in 2025 and is expected to grow to USD 7,975.7 Million by 2035, registering a CAGR of 22.2% over the forecast period. The expansion of Web3 infrastructure, advancements in multi-chain API solutions, and increasing demand for secure and scalable blockchain integrations are fueling market expansion. Additionally, rising adoption of tokenization, cross-chain interoperability, and API-driven NFT marketplaces is shaping the industry's future.

Country-wise Insights

Competitive Outlook

Cryptocurrency historical datasets from January 2012 (if available) to October 2021 were obtained and integrated from various sources and Application Programming Interfaces (APIs) including Yahoo Finance, Cryptodownload, CoinMarketCap, various Kaggle datasets, and multiple APIs. While these datasets used various formats of time (e.g., minutes, hours, days), in order to integrate the datasets days format was used for in this research study. The integrated cryptocurrency historical datasets for 80 cryptocurrencies including but not limited to Bitcoin (BTC), Ethereum (ETH), Binance Coin (BNB), Cardano (ADA), Tether (USDT), Ripple (XRP), Solana (SOL), Polkadot (DOT), USD Coin (USDC), Dogecoin (DOGE), Tron (TRX), Bitcoin Cash (BCH), Litecoin (LTC), EOS (EOS), Cosmos (ATOM), Stellar (XLM), Wrapped Bitcoin (WBTC), Uniswap (UNI), Terra (LUNA), SHIBA INU (SHIB), and 60 more cryptocurrencies were uploaded in this online Mendeley data repository. Although the primary attribute of including the mentioned cryptocurrencies was the Market Capitalization, a subject matter expert i.e., a professional trader has also guided the initial selection of the cryptocurrencies by analyzing various indicators such as Relative Strength Index (RSI), Moving Average Convergence/Divergence (MACD), MYC Signals, Bollinger Bands, Fibonacci Retracement, Stochastic Oscillator and Ichimoku Cloud. The primary features of this dataset that were used as the decision-making criteria of the CLUS-MCDA II approach are Timestamps, Open, High, Low, Closed, Volume (Currency), % Change (7 days and 24 hours), Market Cap and Weighted Price values. The available excel and CSV files in this data set are just part of the integrated data and other databases, datasets and API References that was used in this study are as follows: [1] https://finance.yahoo.com/ [2] https://coinmarketcap.com/historical/ [3] https://cryptodatadownload.com/ [4] https://kaggle.com/philmohun/cryptocurrency-financial-data [5] https://kaggle.com/deepshah16/meme-cryptocurrency-historical-data [6] https://kaggle.com/sudalairajkumar/cryptocurrencypricehistory [7] https://min-api.cryptocompare.com/data/price?fsym=BTC&tsyms=USD [8] https://min-api.cryptocompare.com/ [9] https://p.nomics.com/cryptocurrency-bitcoin-api [10] https://www.coinapi.io/ [11] https://www.coingecko.com/en/api [12] https://cryptowat.ch/ [13] https://www.alphavantage.co/ This dataset is part of the CLUS-MCDA (Cluster analysis for improving Multiple Criteria Decision Analysis) and CLUS-MCDAII Project: https://aimaghsoodi.github.io/CLUSMCDA-R-Package/ https://github.com/Aimaghsoodi/CLUS-MCDA-II https://github.com/azadkavian/CLUS-MCDA

CoinAPI's Level 1 Crypto Quote Data delivers essential digital asset market intelligence, capturing real-time bid/ask prices and volumes across 350+ exchanges including both CEX and DEX platforms.

This comprehensive data stream provides precise market snapshots with microsecond-accurate timestamps, perfect for applications demanding rapid price discovery and effective market monitoring.

Designed for minimal latency and maximum update frequency, our feed powers everything from sophisticated trading algorithms and responsive price widgets to in-depth market analysis tools.

You can access data through FIX or WebSocket for instant streaming or REST API for historical analysis and backtesting.

Why work with us?

Market Coverage & Data Types: - Real-time and historical data since 2010 (for chosen assets) - Full order book depth (L2/L3) - Tick-by-tick data - OHLCV across multiple timeframes - Market indexes (VWAP, PRIMKT) - Exchange rates with fiat pairs - Spot, futures, options, and perpetual contracts - Coverage of 90%+ global trading volume - Full Cryptocurrency Investor Data

Technical Excellence: - 99,9% uptime guarantee - Multiple delivery methods: REST, WebSocket, FIX, S3 - Standardized data format across exchanges - Ultra-low latency data streaming - Detailed documentation - Custom integration assistance

CoinAPI is trusted by financial institutions, trading firms, hedge funds, researchers, and technology developers worldwide. We provide reliable cryptocurrency market data through our commitment to quality and technical performance.

When you need to analyze crypto market history, batch processing often beats streaming APIs. That's why we built the Flat Files S3 API - giving analysts and researchers direct access to structured historical cryptocurrency data without the integration complexity of traditional APIs.

Pull comprehensive historical data across 800+ cryptocurrencies and their trading pairs, delivered in clean, ready-to-use CSV formats that drop straight into your analysis tools. Whether you're building backtest environments, training machine learning models, or running complex market studies, our flat file approach gives you the flexibility to work with massive datasets efficiently.

Why work with us?

Market Coverage & Data Types: - Comprehensive historical data since 2010 (for chosen assets) - Comprehensive order book snapshots and updates - Trade-by-trade data

Technical Excellence: - 99,9% uptime guarantee - Standardized data format across exchanges - Flexible Integration - Detailed documentation - Scalable Architecture

CoinAPI serves hundreds of institutions worldwide, from trading firms and hedge funds to research organizations and technology providers. Our S3 delivery method easily integrates with your existing workflows, offering familiar access patterns, reliable downloads, and straightforward automation for your data team. Our commitment to data quality and technical excellence, combined with accessible delivery options, makes us the trusted choice for institutions that demand both comprehensive historical data and real-time market intelligence

Context

Since the launch of Bitcoin in 2008, hundreds of similar projects based on the blockchain technology have emerged. We call these cryptocurrencies (also coins or cryptos in the Internet slang). Some are extremely valuable nowadays, and others may have the potential to become extremely valuable in the future1. In fact, on the 6th of December of 2017, Bitcoin has a market capitalization above $200 billion.

Content

The cryptocurrency market is exceptionally volatile2 and any money you put in might disappear into thin air. Cryptocurrencies mentioned here might be scams similar to Ponzi Schemes or have many other issues (overvaluation, technical, etc.). Please do not mistake this for investment advice. *

2 Update on March 2020: Well, it turned out to be volatile indeed :D

That said, let's get to business. We will start with a CSV we conveniently downloaded on the 6th of December of 2017 using the coinmarketcap API (NOTE: The public API went private in 2020 and is no longer available) named datasets/coinmarketcap_06122017.csv.

Acknowledgements

Data set is from DataCamp

Cryptocurrencies

Finage offers you more than 1700+ cryptocurrency data in real time.

With Finage, you can react to the cryptocurrency data in Real-Time via WebSocket or unlimited API calls. Also, we offer you a 7-year historical data API.

You can view the full Cryptocurrency market coverage with the link given below. https://finage.s3.eu-west-2.amazonaws.com/Finage_Crypto_Coverage.pdf

Context

The past two months were crazy in the crypto market. The goal is to allow analyze correlations between Bitcoin and other Crypto Currencies in order to do smarter day-trading.

Content

This data set was updated every 15 min using Coin Market Cap API and includes the top 100 coins market cap, price in USD and price in BTC. Every row has its update time in EST Time zone

Acknowledgements

Coin Market Cap API

Inspiration

Who are the followers and leaders in the crypto market? When BTC goes down - what coins should be bought and when? When it goes up - which coins start to rise following it but still giving us enough time to buy them?

This dataset provides comprehensive access to financial market data from Google Finance in real-time. Get detailed information on stocks, market quotes, trends, ETFs, international exchanges, forex, crypto, and related news. Perfect for financial applications, trading platforms, and market analysis tools. The dataset is delivered in a JSON format via REST API.

Context:

This dataset is an extra updating dataset for the G-Research Crypto Forecasting competition.

Introduction

This is a daily updated dataset, automaticlly collecting market data for G-Research crypto forecasting competition. The data is of the 1-minute resolution, collected for all competition assets and both retrieval and uploading are fully automated. see discussion topic.

The Data

For every asset in the competition, the following fields from Binance's official API endpoint for historical candlestick data are collected, saved, and processed.

1. **timestamp** - A timestamp for the minute covered by the row.
2. **Asset_ID** - An ID code for the cryptoasset.
3. **Count** - The number of trades that took place this minute.
4. **Open** - The USD price at the beginning of the minute.
5. **High** - The highest USD price during the minute.
6. **Low** - The lowest USD price during the minute.
7. **Close** - The USD price at the end of the minute.
8. **Volume** - The number of cryptoasset u units traded during the minute.
9. **VWAP** - The volume-weighted average price for the minute.
10. **Target** - 15 minute residualized returns. See the 'Prediction and Evaluation section of this notebook for details of how the target is calculated.
11. **Weight** - Weight, defined by the competition hosts [here](https://www.kaggle.com/cstein06/tutorial-to-the-g-research-crypto-competition)
12. **Asset_Name** - Human readable Asset name.

Indexing

The dataframe is indexed by timestamp and sorted from oldest to newest. The first row starts at the first timestamp available on the exchange, which is July 2017 for the longest-running pairs.

Usage Example

The following is a collection of simple starter notebooks for Kaggle's Crypto Comp showing PurgedTimeSeries in use with the collected dataset. Purged TimesSeries is explained here. There are many configuration variables below to allow you to experiment. Use either GPU or TPU. You can control which years are loaded, which neural networks are used, and whether to use feature engineering. You can experiment with different data preprocessing, model architecture, loss, optimizers, and learning rate schedules. The extra datasets contain the full history of the assets in the same format as the competition, so you can input that into your model too.

Baseline Example Notebooks:

• Neural Network Starter
• LightGBM Starter
• Catboost Starter
• XGBoost Starter
• TabNet Starter
• Reinforcement Learning (PPO) Starter

These notebooks follow the ideas presented in my "Initial Thoughts" here. Some code sections have been reused from Chris' great (great) notebook series on SIIM ISIC melanoma detection competition here

Loose-ends:

This is a work in progress and will be updated constantly throughout the competition. At the moment, there are some known issues that still needed to be addressed:

• VWAP: - At the moment VWAP calculation formula is still unclear. Currently the dataset uses an approximation calculated from the Open, High, Low, Close, Volume candlesticks. [Waiting for competition hosts input]
• Target Labeling: There exist some mismatches to the original target provided by the hosts at some time intervals. On all the others - it is the same. The labeling code can be seen here. [Waiting for competition hosts] input]
• Filtering: No filtration of 0 volume data is taken place.

Example Visualisations

Opening price with an added indicator (MA50): https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F2234678%2Fb8664e6f26dc84e9a40d5a3d915c9640%2Fdownload.png?generation=1582053879538546&alt=media" alt="">

Volume and number of trades: https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F2234678%2Fcd04ed586b08c1576a7b67d163ad9889%2Fdownload-1.png?generation=1582053899082078&alt=media" alt="">

License

This data is being collected automatically from the crypto exchange Binance.

Authors, through Twitter API, collected this database over eight months. These data are tweets of over 50 experts regarding market analysis of 40 cryptocurrencies. These experts are known as influencers on social networks such as Twitter. The theory of Behavioral economics shows that the opinions of people, especially experts, can impact the stock market trend (here, cryptocurrencies). Existing databases often cover tweets related to one or more cryptocurrencies. Also, in these databases, no attention is paid to the user's expertise, and most of the data is extracted using hashtags. Failure to pay attention to the user's expertise causes the irrelevant volume to increase and the neutral polarity to increase considerably. This database has a main table named "Tweets1" with 11 columns and 40 tables to separate comments related to each cryptocurrency. The columns of the main table and the cryptocurrency tables are explained in the attached document. Researchers can use this dataset in various machine learning tasks, such as sentiment analysis and deep transfer learning with sentiment analysis. Also, this data can be used to check the impact of influencers' opinions on the cryptocurrency market trend. The use of this database is allowed by mentioning the source. Also, in this version, we have added the excel version of the database and Python code to extract the names of influencers and tweets. in Version(3): In the new version, three datasets related to historical prices and sentiments related to Bitcoin, Ethereum, and Binance have been added as Excel files from January 1, 2023, to June 12, 2023. Also, two datasets of 52 influential tweets in cryptocurrencies have been published, along with the score and polarity of sentiments regarding more than 300 cryptocurrencies from February 2021 to June 2023. Also, two Python codes related to the sentiment analysis algorithm of tweets with Python have been published. This algorithm combines RoBERTa pre-trained deep neural network and BiGRU deep neural network with an attention layer (see code Preprocessing_and_sentiment_analysis with python).

Cryptocurrency options markets have grown increasingly sophisticated, requiring reliable data infrastructure to support trading and analysis. Our platform gives you direct access to comprehensive crypto options data through straightforward API connections.

We capture the complete options chain across major crypto derivatives exchanges, delivering real-time and historical cryptocurrency market data that shows exactly what's happening in these complex markets. Each options contract is tracked with precision - strikes, expiration dates, premiums, open interest, and volume metrics all accessible through our standardized data feeds.

The data is available through multiple integration methods depending on your needs. Use our REST API for flexible queries and historical analysis, WebSocket for real-time market monitoring, or FIX protocol for institutional-grade connectivity with minimal latency.

Why work with us?

Market Coverage & Data Types: - Real-time and historical data since 2010 (for chosen assets) - Full order book depth (L2/L3) - Tick-by-tick data - OHLCV across multiple timeframes - Market indexes (VWAP, PRIMKT) - Exchange rates with fiat pairs - Spot, futures, options, and perpetual contracts - Coverage of 90%+ global trading volume

Technical Excellence: - 99% uptime guarantee - Multiple delivery methods: REST, WebSocket, FIX, S3 - Standardized data format across exchanges - Ultra-low latency data streaming - Detailed documentation - Custom integration assistance

When options traders need reliable market intelligence, they don't leave it to chance. That's why trading desks across five continents, quantitative hedge funds managing billions, and fintech innovators building tomorrow's trading platforms all rely on our data infrastructure. We've established ourselves as a dependable source in a market where accuracy isn't just preferred - it's essential. While others promise comprehensive coverage, we deliver it consistently, trade after trade, day after day.

Through Telegram API, the authors collected this database over four months ago. These data are Telegram's comments of over eight professional Telegram channels about cryptocurrencies from December 2023 to March 2024. The theory of Behavioral economics shows that the opinions of people, especially experts, can impact the stock market trend (here, cryptocurrencies). Existing databases often cover tweets or Telegram's comments on one or more cryptocurrencies. Also, in these databases, no attention is paid to the user's expertise, and most of the data is extracted using hashtags. Failure to pay attention to the user's expertise causes the irrelevant volume to increase and the neutral polarity considerably. This database has a main table with eight columns. The columns of the main table are explained in the attached document. Researchers can use this dataset in various machine learning tasks, such as sentiment analysis and deep transfer learning with sentiment analysis. Also, this data can be used to check the impact of influencers' opinions on the cryptocurrency market trend. The use of this database is allowed by mentioning the source. Furthermore, we have added Python code to extract Telegram's comments. We used the RoBERTa pre-trained deep neural network and BiGRU deep neural network with an attention layer-based HDRB model(https://ieeexplore.ieee.org/document/10292644) for sentiment analysis.

CoinAPI delivers institutional-grade Historical Crypto Data for backtesting and analysis. Our cryptocurrency archive powers research across Bitcoin, Ethereum and all markets through one reliable API—transforming strategies with precision data that matters.

Context:

This dataset is an extra updating dataset for the G-Research Crypto Forecasting competition.

Introduction

This is a daily updated dataset, automaticlly collecting market data for G-Research crypto forecasting competition. The data is of the 1-minute resolution, collected for all competition assets and both retrieval and uploading are fully automated. see discussion topic.

The Data

For every asset in the competition, the following fields from Binance's official API endpoint for historical candlestick data are collected, saved, and processed.

1. **timestamp** - A timestamp for the minute covered by the row.
2. **Asset_ID** - An ID code for the cryptoasset.
3. **Count** - The number of trades that took place this minute.
4. **Open** - The USD price at the beginning of the minute.
5. **High** - The highest USD price during the minute.
6. **Low** - The lowest USD price during the minute.
7. **Close** - The USD price at the end of the minute.
8. **Volume** - The number of cryptoasset u units traded during the minute.
9. **VWAP** - The volume-weighted average price for the minute.
10. **Target** - 15 minute residualized returns. See the 'Prediction and Evaluation section of this notebook for details of how the target is calculated.
11. **Weight** - Weight, defined by the competition hosts [here](https://www.kaggle.com/cstein06/tutorial-to-the-g-research-crypto-competition)
12. **Asset_Name** - Human readable Asset name.

Indexing

The dataframe is indexed by timestamp and sorted from oldest to newest. The first row starts at the first timestamp available on the exchange, which is July 2017 for the longest-running pairs.

Usage Example

The following is a collection of simple starter notebooks for Kaggle's Crypto Comp showing PurgedTimeSeries in use with the collected dataset. Purged TimesSeries is explained here. There are many configuration variables below to allow you to experiment. Use either GPU or TPU. You can control which years are loaded, which neural networks are used, and whether to use feature engineering. You can experiment with different data preprocessing, model architecture, loss, optimizers, and learning rate schedules. The extra datasets contain the full history of the assets in the same format as the competition, so you can input that into your model too.

Baseline Example Notebooks:

• Neural Network Starter
• LightGBM Starter
• Catboost Starter
• XGBoost Starter
• TabNet Starter
• Reinforcement Learning (PPO) Starter

These notebooks follow the ideas presented in my "Initial Thoughts" here. Some code sections have been reused from Chris' great (great) notebook series on SIIM ISIC melanoma detection competition here

Loose-ends:

This is a work in progress and will be updated constantly throughout the competition. At the moment, there are some known issues that still needed to be addressed:

• VWAP: - At the moment VWAP calculation formula is still unclear. Currently the dataset uses an approximation calculated from the Open, High, Low, Close, Volume candlesticks. [Waiting for competition hosts input]
• Target Labeling: There exist some mismatches to the original target provided by the hosts at some time intervals. On all the others - it is the same. The labeling code can be seen here. [Waiting for competition hosts] input]
• Filtering: No filtration of 0 volume data is taken place.

Example Visualisations

Opening price with an added indicator (MA50): https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F2234678%2Fb8664e6f26dc84e9a40d5a3d915c9640%2Fdownload.png?generation=1582053879538546&alt=media" alt="">

Volume and number of trades: https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F2234678%2Fcd04ed586b08c1576a7b67d163ad9889%2Fdownload-1.png?generation=1582053899082078&alt=media" alt="">

License

This data is being collected automatically from the crypto exchange Binance.

Get complete crypto market data from CoinAPI, featuring comprehensive price history, detailed trading volumes, and powerful analytics tools. Our service combines real-time feeds with extensive historical crypto data, giving you a complete view of market movements over time.

CoinAPI's Crypto Market Data delivers quotes, orderbooks, trades and OHLCV data through flexible delivery methods including API, WebSocket, and convenient S3 exports. This institutional-grade data is clean, normalized, and ready for immediate use in your trading platforms or analysis tools.

Whether you're tracking market trends, analyzing volume patterns, developing trading strategies, or conducting risk assessments, our reliable data foundation gives you the insights you need to make informed decisions in the crypto market.

Why work with us?

Market Coverage & Data Types: - Real-time and historical data since 2010 (for chosen assets) - Full order book depth (L2/L3) - Trading Data - Tick-by-tick data - OHLCV across multiple timeframes - Market indexes (VWAP, PRIMKT) - Exchange rates with fiat pairs - Spot, futures, options, and perpetual contracts - Coverage of 90%+ global trading volume

Technical Excellence: - 99,9% uptime guarantee - Multiple delivery methods: REST, WebSocket, FIX, S3 - Standardized data format across exchanges - Ultra-low latency data streaming - Detailed documentation - Custom integration assistance

CoinAPI works with hundreds of companies around the world - from trading desks and hedge funds to research teams and tech developers. We've built our reputation on reliable data and solid technical performance. That's why so many businesses turn to us when they need dependable crypto market information.

The dataset was collected for the period spanning between 01/07/2019 and 31/12/2022.The historical Twitter volume were retrieved using ‘‘Bitcoin’’ (case insensitive) as the keyword from bitinfocharts.com. Google search volume was retrieved using library Gtrends. 2000 tweets per day using 4 times interval were crawled by employing Twitter API with the keyword “Bitcoin. The daily closing prices of Bitcoin, oil price, gold price, and U.S stock market indexes (S&P 500, NASDAQ, and Dow Jones Industrial Average) were collected using R libraries either Quantmod or Quandl.

Context:

This dataset is an extra updating dataset for the G-Research Crypto Forecasting competition.

Introduction

This is a daily updated dataset, automaticlly collecting market data for G-Research crypto forecasting competition. The data is of the 1-minute resolution, collected for all competition assets and both retrieval and uploading are fully automated. see discussion topic.

The Data

For every asset in the competition, the following fields from Binance's official API endpoint for historical candlestick data are collected, saved, and processed.

1. **timestamp** - A timestamp for the minute covered by the row.
2. **Asset_ID** - An ID code for the cryptoasset.
3. **Count** - The number of trades that took place this minute.
4. **Open** - The USD price at the beginning of the minute.
5. **High** - The highest USD price during the minute.
6. **Low** - The lowest USD price during the minute.
7. **Close** - The USD price at the end of the minute.
8. **Volume** - The number of cryptoasset u units traded during the minute.
9. **VWAP** - The volume-weighted average price for the minute.
10. **Target** - 15 minute residualized returns. See the 'Prediction and Evaluation section of this notebook for details of how the target is calculated.
11. **Weight** - Weight, defined by the competition hosts [here](https://www.kaggle.com/cstein06/tutorial-to-the-g-research-crypto-competition)
12. **Asset_Name** - Human readable Asset name.

Indexing

The dataframe is indexed by timestamp and sorted from oldest to newest. The first row starts at the first timestamp available on the exchange, which is July 2017 for the longest-running pairs.

Usage Example

The following is a collection of simple starter notebooks for Kaggle's Crypto Comp showing PurgedTimeSeries in use with the collected dataset. Purged TimesSeries is explained here. There are many configuration variables below to allow you to experiment. Use either GPU or TPU. You can control which years are loaded, which neural networks are used, and whether to use feature engineering. You can experiment with different data preprocessing, model architecture, loss, optimizers, and learning rate schedules. The extra datasets contain the full history of the assets in the same format as the competition, so you can input that into your model too.

Baseline Example Notebooks:

• Neural Network Starter
• LightGBM Starter
• Catboost Starter
• XGBoost Starter
• TabNet Starter
• Reinforcement Learning (PPO) Starter

These notebooks follow the ideas presented in my "Initial Thoughts" here. Some code sections have been reused from Chris' great (great) notebook series on SIIM ISIC melanoma detection competition here

Loose-ends:

This is a work in progress and will be updated constantly throughout the competition. At the moment, there are some known issues that still needed to be addressed:

• VWAP: - At the moment VWAP calculation formula is still unclear. Currently the dataset uses an approximation calculated from the Open, High, Low, Close, Volume candlesticks. [Waiting for competition hosts input]
• Target Labeling: There exist some mismatches to the original target provided by the hosts at some time intervals. On all the others - it is the same. The labeling code can be seen here. [Waiting for competition hosts] input]
• Filtering: No filtration of 0 volume data is taken place.

Example Visualisations

Opening price with an added indicator (MA50): https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F2234678%2Fb8664e6f26dc84e9a40d5a3d915c9640%2Fdownload.png?generation=1582053879538546&alt=media" alt="">

Volume and number of trades: https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F2234678%2Fcd04ed586b08c1576a7b67d163ad9889%2Fdownload-1.png?generation=1582053899082078&alt=media" alt="">

License

This data is being collected automatically from the crypto exchange Binance.

Graph and download economic data for Coinbase Bitcoin (CBBTCUSD) from 2014-12-01 to 2025-07-12 about cryptocurrency and USA.

Context:

This dataset is an extra updating dataset for the G-Research Crypto Forecasting competition.

Introduction

This is a daily updated dataset, automaticlly collecting market data for G-Research crypto forecasting competition. The data is of the 1-minute resolution, collected for all competition assets and both retrieval and uploading are fully automated. see discussion topic.

The Data

For every asset in the competition, the following fields from Binance's official API endpoint for historical candlestick data are collected, saved, and processed.

1. **timestamp** - A timestamp for the minute covered by the row.
2. **Asset_ID** - An ID code for the cryptoasset.
3. **Count** - The number of trades that took place this minute.
4. **Open** - The USD price at the beginning of the minute.
5. **High** - The highest USD price during the minute.
6. **Low** - The lowest USD price during the minute.
7. **Close** - The USD price at the end of the minute.
8. **Volume** - The number of cryptoasset u units traded during the minute.
9. **VWAP** - The volume-weighted average price for the minute.
10. **Target** - 15 minute residualized returns. See the 'Prediction and Evaluation section of this notebook for details of how the target is calculated.
11. **Weight** - Weight, defined by the competition hosts [here](https://www.kaggle.com/cstein06/tutorial-to-the-g-research-crypto-competition)
12. **Asset_Name** - Human readable Asset name.

Indexing

The dataframe is indexed by timestamp and sorted from oldest to newest. The first row starts at the first timestamp available on the exchange, which is July 2017 for the longest-running pairs.

Usage Example

The following is a collection of simple starter notebooks for Kaggle's Crypto Comp showing PurgedTimeSeries in use with the collected dataset. Purged TimesSeries is explained here. There are many configuration variables below to allow you to experiment. Use either GPU or TPU. You can control which years are loaded, which neural networks are used, and whether to use feature engineering. You can experiment with different data preprocessing, model architecture, loss, optimizers, and learning rate schedules. The extra datasets contain the full history of the assets in the same format as the competition, so you can input that into your model too.

Baseline Example Notebooks:

• Neural Network Starter
• LightGBM Starter
• Catboost Starter
• XGBoost Starter
• TabNet Starter
• Reinforcement Learning (PPO) Starter

These notebooks follow the ideas presented in my "Initial Thoughts" here. Some code sections have been reused from Chris' great (great) notebook series on SIIM ISIC melanoma detection competition here

Loose-ends:

This is a work in progress and will be updated constantly throughout the competition. At the moment, there are some known issues that still needed to be addressed:

• VWAP: - At the moment VWAP calculation formula is still unclear. Currently the dataset uses an approximation calculated from the Open, High, Low, Close, Volume candlesticks. [Waiting for competition hosts input]
• Target Labeling: There exist some mismatches to the original target provided by the hosts at some time intervals. On all the others - it is the same. The labeling code can be seen here. [Waiting for competition hosts] input]
• Filtering: No filtration of 0 volume data is taken place.

Example Visualisations

Opening price with an added indicator (MA50): https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F2234678%2Fb8664e6f26dc84e9a40d5a3d915c9640%2Fdownload.png?generation=1582053879538546&alt=media" alt="">

Volume and number of trades: https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F2234678%2Fcd04ed586b08c1576a7b67d163ad9889%2Fdownload-1.png?generation=1582053899082078&alt=media" alt="">

License

This data is being collected automatically from the crypto exchange Binance.