2 releases
0.0.3 | Aug 27, 2024 |
---|---|
0.0.2 | Aug 27, 2024 |
#36 in #data-exchange
Used in ohlcv-ctl
69KB
1K
SLoC
OHLCV
OHLCV is a library and a command line tool for downloading historical OHLCV (Open, High, Low, Close, Volume) data from various cryptocurrency exchanges. The data is stored in a database and can be easily queried using SQL.
Price values are stored as Decimal
values. This ensures that the data is processed accurately and without rounding
errors.
The library uses the SQLx crate for database access.
Status
This project is in the early stages of development and is not yet ready for production use.
- Data model and base types
- Initialize and drop schema
- Download historical OHLCV data
- Export/import OHLCV data as CSV or JSON
Data model
The data model mainly consists of the following types:
Candle
: Represents a candlestick in a trading pair.Coin
: Represents a cryptocurrency and the quote currency.Currency
: Represents a currency.Timeframe
: Represents a timeframe of a candlestick.
The data model is designed to be simple and easy to use. For every trading pair
(Candle
) consisting of a base currency and a quote currency, there is a
corresponding table in the database. The table name is constructed from the base
currency, the quote currency, and a prefix. The prefix is the same for all
tables and is used to group the tables together.
In the table there the candles are aggregated for all timeframes. The columns of the table are the following:
time_stamp
: The start time of the candle.time_frame
: The timeframe of the candle.sources
: The number of sources the candle was downloaded from.open
: The opening price of the candle.high
: The highest price of the candle.low
: The lowest price of the candle.close
: The closing price of the candle.volume
: The volume of the candle.
The primary key of the table is the combination of time_stamp
and
time_frame
.
Database access
The library supports the following databases:
- SQLite
- PostgreSQL
- MySQL/MariaDB
The database can be accessed using the DbType
type. The tables defining the
candles can be initialized and dropped using the init_schema
and drop_schema
methods. All data definition is done by the root
user. The normal user only
has access to the data. Exception to this is SQLite, where no user management is
needed.
See the implementation of the database configuration for more details.
The Database
trait provides methods to interact with the database. The trait
is implemented for the DbType
type.
Download historical OHLCV data
This feature is not yet implemented.
The library can download historical OHLCV data from various cryptocurrency exchanges. The data is stored in a database and can be queried using SQL. The data will be downloaded in a 5-minute interval of the previous day, resulting in 288 candles per day. The candles will be aggregated in the database to form larger candles, such as 15-minute, 1-hour, 4-hour, and 1-day candles.
It is possible to download data for multiple trading pairs and multiple exchanges at the same time. In order to collect the data, the library will use the exchange's public REST API. The data will be downloaded in parallel to speed up the process. Care will be taken to avoid rate limiting and to handle errors gracefully.
To get a consistent time-series of the data, the command line tool must be run at least once a day. The tool will download the data for the previous day and aggregate it in the database. If the tool is run more than once a day, it will only download the missing trading pairs. All times are in UTC only.
Supported exchanges include:
- Binance
- KuCoin
More exchanges will be added in the future.
The databases supported include:
- SQLite
- PostgreSQL
- MySQL
- MariaDB
The downloaded data can be exported to a CSV file.
There will be methods implemented to handle gaps in the data. Gaps will be classified as:
- Short gaps: A gap of one or two 5-minute candles.
- Moderate gaps: A gap of three to five 5-minute candles.
The library provides methods to fill the gaps in the data. The gaps will be filled by interpolating the missing candles. Short gaps will be filled by linear interpolation, while moderate gaps will be filled by cubic spline interpolation. Special care is taken, if the gap is at the beginning or end of the data.
The download will fail for a trading pair for an exchange if:
- there is a gap of more than five candles,
- next gap is less than 5 candles away,
- more than 5% of the data is missing.
If dowloading a trading pair from more than one exchange a gap may not be across two or more exchanges at the same time. In this case the candles of the first exchange will be kept and the candles of the other exchanges will be downloaded again. The order of the exchanges is determined by the order of the exchanges in the configuration file for the trading pair.
There will be three attempts to download the data with increasing time between attempts for a trading pair for an exchange.
If downloading a trading pair from more than one exchange, the prive values of the candles will be averaged by a volume-weighted average price (VWAP). The volume of the candles will be summed. In the candle the number of sources will be stored.
Dependencies
~6–21MB
~344K SLoC