racing_data
This project aims to provide horse racing data management and analysis services in Python.
Installation
Prior to using racing_data, the package must be installed in your current Python environment. In most cases, an automated installation via PyPI and pip will suffice, as follows:
pip install racing_data
If you would prefer to gain access to new (unstable) features via a pre-release version of the package, specify the 'pre' option when calling pip, as follows:
pip install --pre racing_data
To gain access to bleeding edge developments, the package can be installed from a source distribution. To do so, you will need to clone the git repository and execute the setup.py script from the root directory of the source tree, as follows:
git clone https://github.com/justjasongreen/racing_data.git cd racing_data python setup.py install
If you would prefer to install the package as a symlink to the source distribution (for development purposes), execute the setup.py script with the 'develop' option instead, as follows:
python setup.py develop
Basic Usage
To access the functionality described below, you must first create an instance of the racing_data.Provider class. To do so, you will need to provide a compatible web scraper and a database connection. The web scraper can be any object that implements the punters_client.Scraper API, support calls such as the following:
meets = scraper.scrape_meets(date) races = scraper.scrape_races(meet) runners = scraper.scrape_runners(race) horse = scraper.scrape_horse(runner) jockey = scraper.scrape_jockey(runner) trainer = scraper.scrape_trainer(runner) performances = scraper.scrape_performances(horse)
The database connection can be any object that implements the pymongo.Database API, supporting calls such as the following:
documents = database[collection_name].find(query)
document['_id'] = database[collection_name].insert_one(document).inserted_id
database[collection_name].replace_one({'_id': document['_id']}, document)
racing_data has only been tested with punters_client.Scraper as the web scraper and pymongo.Database as the database connection. To set up the required dependencies in your own project using the same packages, execute the following code in your Python interpreter:
>>> import pymongo >>> database_uri = 'mongodb://localhost:27017/racing_data' >>> database_client = pymongo.MongoClient(database_uri) >>> database = database_client.get_default_database() >>> import cache_requests >>> http_client = cache_requests.Session() >>> from lxml import html >>> html_parser = html.fromstring >>> import punters_client >>> scraper = punters_client.Scraper(http_client, html_parser)
With these dependencies in place, you can now create an instance of the racing_data.Provider class as follows:
>>> import racing_data >>> provider = racing_data.Provider(database, scraper)
The provider instance can now be used to scrape, store and access a range of racing data, as illustrated in the following sections...
Getting Meets
Meets represent a collection of races occurring at a given track on a given date. To get a list of meets occurring on a given date, execute the following code in your Python interpreter:
>>> from datetime import datetime >>> date = datetime(2016, 2, 1) >>> meets = provider.get_meets_by_date(date)
The get_meets_by_date method will return a list of Meet objects. Meet objects are derived from Python's built-in dict type, so a meet's details can be accessed as follows:
>>> meet = meets[index] >>> track = meet['track']
Getting Races
Races represent a collection of runners competing in a single event at a meet. To get a list of races occurring at a given meet, execute the following code in your Python interpreter (where meet is an existing Meet object obtained via the provider.get_meets_by_date method):
>>> races = meet.races
The meet.races property will return a list of Race objects. Race objects are derived from Python's built-in dict type, so a race's details can be accessed as follows:
>>> race = races[index] >>> number = race['number']
In addition to the dictionary values, Race objects also provide a 'meet' property that can be used to get the meet at which the race occurs:
>>> meet = race.meet
Getting Runners
Runners represent a single combination of horse, jockey and trainer competing in a race. To get a list of runners competing in a given race, execute the following code in your Python interpreter (where race is an existing Race object obtained via the Meet.races property):
>>> runners = race.runners
The race.runners property will return a list of Runner objects. Runner objects are derived from Python's built-in dict type, so a runner's details can be accessed as follows:
>>> runner = runners[index] >>> number = runner['number']
In addition to the dictionary values, Runner objects also provide a 'race' property that can be used to get the race in which the runner occurs:
>>> race = runner.race
Furthermore, Runner objects also offer the following calculated values as properties that can be accessed using dot-notation:
| Property | Description |
|---|---|
| actual_distance | The race distance adjusted for the runner's barrier and race track circ/straight values |
| actual_weight | The average weight of a racehorse plus the actual weight being carried by the horse |
| age | The age of the horse as at the date of the race |
| carrying | The weight being carried by the horse after allowances |
| result | The final result for the runner (if the race has already been run) |
| spell | The number of days since the horse's previous race |
| starting_price | The starting price for the runner (if the race has already been run) |
| up | The number of races run by the horse (including this one) since a rest period of 90 days or more |
In addition to the properties listed above, Runner objects also offer the following performance lists (see below) as properties that can also be accessed using dot-notation:
| Property | Description |
|---|---|
| at_distance | All prior performances for the horse within 100m of the current race distance |
| at_distance_on_track | All prior performances for the horse within 100m of the current race distance on the same track |
| at_up | All prior performances for the horse with the same UP number as the current run |
| career | All performances for the horse prior to the current race date |
| last_10 | The last 10 performances for the horse prior to the current race |
| last_12_months | All performances for the horse within 12 months prior to the current race date |
| on_firm | All prior performances for the horse on FIRM tracks |
| on_good | All prior performances for the horse on GOOD tracks |
| on_heavy | All prior performances for the horse on HEAVY tracks |
| on_soft | All prior performances for the horse on SOFT tracks |
| on_synthetic | All prior performances for the horse on SYNTHETIC tracks |
| on_track | All prior performances for the horse on the same track as the current race |
| on_turf | All prior performances for the horse on turf tracks (that is, NOT synthetic tracks) |
| since_rest | All prior performances for the horse since its last rest period of 90 days or more |
| with_jockey | All prior performances for the horse with the same jockey |
Performance Lists
The performance list properties described above return PerformanceList objects. The PerformanceList class is derived from Python's built-in list type, allowing easy access to the individual performances contained in the list. In addition to the built-in list functionality, PerformanceList objects also offer the following calculated values as properties that can be accessed via dot-notation:
| Property | Description |
|---|---|
| earnings | The total amount earned by the horse and connections in the list |
| earnings_potential | The total earnings as a percentage of the total prize pools in the list |
| fourths | The number of fourth placed performances in the list |
| fourth_pct | The percentage of fourth placed performances in the list |
| momentums | Returns a tuple containing minimum, maximum and average momentum for the list |
| places | The number of placing (first/second/third) performances in the list |
| place_pct | The percentage of placing (first/second/third) performances in the list |
| result_potential | Returns 1.0 - (the sum of all results / the sum of all starters) in the list |
| roi | The total return on investment if a $1 WIN bet were placed on all performances in the list |
| seconds | The number of second placed performances in the list |
| second_pct | The percentage of second placed performances in the list |
| starting_prices | Returns a tuple containing minimum, maximum and average starting prices for the list |
| starts | The total number of starts in the list |
| thirds | The total number of third placed performances in the list |
| third_pct | The percentage of third placed performances in the list |
| wins | The total number of winning performances in the list |
| win_pct | The percentage of winning performances in the list |
Getting Horses, Jockeys and Trainers
To get the horse, jockey or trainer associated with a given runner, execute the following code in your Python interpreter (where runner is an existing Runner object obtained via the Race.runners property):
>>> horse = runner.horse >>> jockey = runner.jockey >>> trainer = runner.trainer
The runner.horse, runner.jockey and runner.trainer properties will return Horse, Jockey and Trainer objects respectively. Horse, Jockey and Trainer objects are derived from Python's built-in dict type, so a horse/jockey/trainer's details can be accessed as follows:
>>> name = horse['name'] >>> name = jockey['name'] >>> name = trainer['name']
Getting Performances
Performances represent the results of completed runs by horses and jockeys. To get a list of performances for a given horse, execute the following code in your Python interpreter (where horse is an existing Horse object obtained via the Runner.horse property):
>>> performances = horse.performances
The horse.performances property will return a list of Performance objects. Performance objects are derived from Python's built-in dict type, so a performance's details can be accessed as follows:
>>> performance = performances[index] >>> result = performance['result']
In addition to the dictionary values, Performance objects also provide 'horse' and 'jockey' properties that can be used to get the horse/jockey associated with the performance:
>>> horse = performance.horse >>> jockey = performance.jockey
Furthermore, Performance objects also offer the following calculated values as properties that can be accessed using dot-notation:
| Property | Description |
|---|---|
| actual_distance | The actual distance covered by the horse in the winning time |
| actual_weight | The average weight of a racehorse plus the actual weight being carried by the horse |
| momentum | The average momentum achieved by the horse/jockey during the run |
| profit | The profit/loss on a $1 bet on this performance |
| result | The final result for the runner (if the race has already been run) |
| speed | The average speed achieved by the horse/jockey during the run |
| spell | The number of days since the horse's previous race |
| up | The number of races run by the horse (including this one) since a rest period of 90 days or more |
(NOTE: Jockey objects also provide a 'performances' property that can be used to access a list of performances associated with that jockey. Unlike the Horse.performances property though, the Jockey.performances property will not scrape the web in search of relevant data. Instead, the Jockey.performances property will only return relevant performances that already exist in the database. This is due to the vast number of past performances associated with any given jockey, and the inherent difficulty in scraping such a vast amount of data in a timely fashion from most data providers.)
Development and Testing
The source distribution includes a test suite based on pytest. To ensure compatibility with all supported versions of Python, it is recommended that the test suite be run via tox.
To install all development and test requirements into your current Python environment, execute the following command from the root directory of the source tree:
pip install -e .[dev,test]
To run the test suite included in the source distribution, execute the tox command from the root directory of the source tree as follows:
tox
