Gerardo Salazar
eb1181f5f7
* Adds preliminary universe selection for Future Options
* Fixes scaling issues with Future Options
* Fixes scaling multiplying by 10000x instead of using _scaleFactor
* Fixes scaling for Tick
* Revert changes to Tick since it divides the scaling factor
* Changes stale method name to new method name after rebase
* Fixes selection bugs, adds new methods, and adds unit tests
* Fixes bug where Equity Symbol was created for an underlying
non-equity Symbol, resulting in equity data trying to be loaded
* Adds unit tests covering changes to Tick, QuoteBar, TradeBar and
LeanData
* Adds regression test for AddUniverseOption filter contract selection
for Future Options
* Addresses review - modifies the AddFutureOption signature
* Adds new AddUniverseOptions method overload
* Removes and adds a new unit test
* Misc. modifications to account for new changes
* Fixes bug where futures were loaded using default SID Date
* Refactors and removes unnecessary work
* Fixes regression algorithm, which previously made no trades
* Adds future option data
* Adds the corresponding underlying data, in this case, futures data
to enable usage of future options data
* Replaces data with new data (ES18Z20)
* Improves Future chain filtering and updates regression stats
* Add AddFutureOptionContract API
* Expands regression and unit tests to test in finer detail
* Adds Python regression algorithms for AddFutureOption[Contract] methods
* Adds new unit test for BacktestingOptionChainProvider
* Fixes bug with BacktesingOptionChainProvider where we
attempted to load the Trades option chain first, resulting
in breakage of backwards compatibility and limitation of the
option chain.
* Adds new regression algorithms (Py) to Algorithm.Python project
* Adds FutureOptionMarginBuyingPowerModel
* Modifies code paths used to select margin model
* Adds related unit tests for margin model
* Fixes issue with unit test and MHDB/SPDB lookup for Future Options
* Preliminary regression algorithm testing ITM call/put option buying
* Fixes bug where fee model used did not find non-US market
options fee model. We now use the futures fee model for future
options because IB charges the same commissions per contract
between futures and futures options
* Adds proper regression algorithm for ITM future options expiration
* Pushing broken algorithm for review
* Currently, algorithm does not fill forward, causing
a single future option to not get exercised when it is delisted.
* Adds FutureOptionPutITMExpiryRegressionAlgorithm
* Improves existing regression algorithm for call side
* Fixes bug in existing regression algorithm
* Adds AAPL daily data to advance enumerator for ^^^ fix
* Adds additional future option regression algorithms
* Adds Buy OTM expiration regression algorithms
* Adds Sell ITM/OTM expiration regression algorithms
* Adds missing Python regression algorithms
* Adds remaining Python regression algorithms and fixes issues
* Fixes naming issues and statistics
* Adds short option OTM regression algorithms (Py)
* Add license header and class comments to python algorithms
* Cleans up comments and docstrings
* Create Buy/Sell call intraday regression algo
* Redirects future options symbol properties to futures symbol properties
* Asserts exercise/assignment price and updates stats in regression algos
* Adds new unit test covering changes to SecurityService
* Adds comments and fixes failing test
* Partially fixes future option mis-calculated profit/loss
* Adjusts portfolio model to calculate FOP as a no upfront pay asset class
* Updates regression algorithm statistics
* Begin IB FOP support
* Initial support for FOP IB data streaming, live í¾
* Adds additional functionality to LiveOptionChainProvider
- Allows querying CME API to retrieve option chains for CME products
- Ultimately, it's also the groundwork for the CME
LiveFutureChainProvider
* Edits IDataQueueUniverseProvider interface to provide greater
control to implementors of it
* Misc. bug fixes required to get FOP data streaming through IB
* Adds comments, adds missing rategate call, and cleans up code
* Force exchange for FOP and Futures when no exchange is provided
* Fixes bug with Portfolio modeling across all asset classes
* Adds LiveOptionChainProvider tests for Future Options
* IB brokerage option symbol bug fixes and improvements
* Fixes contract multiplier lookup bug
* Fixes issue where we attempted to subscribe to IB data feed with canonical security
* Adds ES MHDB entry
* Reverts portfolio modeling changes for Futures Options
* Since IB eats into our account's cash balance when
a new FOP contract is purchased, we must model by applying funds
to our cash whenever a new purchase/sell occurs.
If we choose to model FOPs exactly as we do with futures, we
will end up with an invalid TotalPortfolioValue on algorithm
restart. By all means and purposes, FOPs are modeled exactly
the same as equity options with respect to the portfolio.
* Adds comments clarifying portfolio modeling and clarifies
existing portfolio modeling comments with additional context.
* Fixes IB symbol lookup for future options
* Fixes LiveOptionChainProvider looping 5 times per option chain
request, even on success
* Sets OptionChainedUniverseSelectionModel to produce a canonical
future/future option/option Symbol to avoid creating two Symbols
* Adds GLOBEX future option symbol mapping from future -> fop
* Fixes LiveOptionChainProvider loading wrong contract option chains
* Fixes loading of futures options ZIP files when backtesting
* Adds a string -> decimal JSON converter
* Additional fixes/refactoring to the LiveOptionChainProvider
* Adds tests for changes to Symbol and LeanData
* Reverts changes to IB-symbol-map
* Fixes Value for mapped future options tickers
* Fixes Symbol test
* Changes path of future options to future's expiry date
* Extra changes made to remove scaling from writing CSV
* Added method to map from FOP Globex -> FUT Globex
* Fixes MOO and MOC orders for future options
* Note: this order type might not be supported by IB or CME.
* Bug fixes and updates unit tests
* Update regression tests and data format
* Rebase changes
* 1. Multiple bug fixes for LiveOptionChainProvider, reverts IQFeed changes
2. Address review (partial): Code reuse and cleanup
1.
* Modifies check in
`AddFutureOptionShort(Call|Put)ITMExpiryRegressionAlgorithm`
to ensure no buys have negative quantity
* Code reuse changes in IB brokerage
* Bug fix in IB brokerage where we assigned the FOP expiry
as the futures expiry (requires verification)
* Doc changes and adds missing summaries/license banners
* Disposes of HTTP client resources in LiveOptionChainProvider
* Renames classes and adds FutureOption folder in Common/Securities
2.
* We revert back to the quotes API for the option chain,
since the settlement API sometimes had missing strikes.
* Fixes future option expiry being set as future's expiry
in LiveOptionChainProvider
* Fixes bug where wrong option chain was selected because of bad
expiry lookup in the futures expiries returned from CME
* Fixes multiple looping bug in LiveOptionChainProvider
* Adds strike price scaling for LiveOptionChainProvider
* Reverts IQFeed changes and simplifies interface upgrade changes
Some additional challenges we'll have to solve as part of FOPs:
- The `OptionSymbol.IsStandard` method makes the assumption that
weeklies contracts follow the pattern equities follows, which
does not apply to Futures Options
- The Subscription created in:
`OptionChainUniverseSubscriptionEnumeratorFactory`
...adds a Trade config. For illiquid contracts, this
will delay universe selection for the option symbol
until we get a trade. However, if we add a quote config,
the data would instead be loaded based on the first quote
we received from the brokerage.
But since we're currently using a trade config, illiquid
contracts won't start streaming data until it receives a trade.
NOTE: this commit is a WIP to addressing the reviews received in the PR,
but has been committed early for efficiency in the review process
* Fixes regression algorithms and misc. bugs
* Fixes map file lookup for non-equity options
* Adds extra assertion at end of algorithm to ensure no holdings are
left when the algorithm ends.
* Adds FutureOptionSymbol, allowing all contracts through as standard
* Changes SPDB to allow defaulting to underlying future symbol
properties if no entry is found for the given FOP
* Fixes calls to SPDB in SecurityService, IBBrokerage
* Reverts AAPL daily ZIP file to fix majority of regression algorithms
* Adds FOPs symbol properties
* Fixes existing symbol properties for a few futures
* Adds tests for changes to Symbol Properties Database
* Removes string SPDB lookup method
* Updates tests and misc callees of previous method
* Updates all regression tests to use data of already expired contracts
* Adds Futures Options Expiry Functions tests
* Adds required futures data for 2020-01-05
* Address review (partial): Expands test coverage and fixes tests
* Set option chain tests parallelism to fixture only
* Fixes broken test for contract month delta for FuturesOptionsExpiryFunctions
* Changes delisting date logic for Futures Options
* Address review: removes duplicate code, misc code fixes
* Bug fix in MarketHoursDatabase.GetDatabaseSymbolKey() where
we would use the underlying's Symbol for lookup in the MHDB
* Adds missing license banner
* Removes Futures Options entries from MHDB
* Adds new tests
* Adds SecurityType.FutureOption
* Converts any underlying comparisons and uses SecurityType directly
instead for FOP specific behavior
* Extra code modifications to acommodate new SecurityType
* Addresses review: fixes order fee bug on exercise
* Additional bug fixes and adding of SecurityType.FutureOption
* Updates regression algorithms OrderListHash
* Fixes various bugs in IB live implementation
* Fixes bug setting the right contract expiration date for FOP
generated by LiveOptionChainProvider
* Adds new function to FuturesOptionsExpiryFunctions
* Clarifies parameter names better in some functions/methods
* Fixes bugs in IB brokerage for FOPs
* Address review - code cleanup and refactor
* Remove MappingEventProvider, SplitEventProvider, and
DividendEventProvider for Futures Options in
CorporateEventEnumeratorFactory
* Address review: Use MHDB key resolver in SPDB
* Makes regression tests pass and adds comment for expiry issue
* Fixes MHDB lookup on string symbol method
* Adds Futures Options greeks regression algorithm (C# only)
* Adds explanitory comment on MHDB FOP lookup
* Remove python from FutureOptionCallITMGreeksExpiryRegressionAlgorithm
QuantConnect Research Project
Currently we have two ways to use QuantConnect research notebooks, you can either install and run locally or just use our docker image (Recommended).
The up to date docker image is available at quantconnect/research. You can pull this image with docker pull quantconnect/research.
Using the Docker Image
Starting the Container
The docker image we created can be started using the included .bat/.sh file in this directory (Lean/Research). These scripts take care of all the work required to get the notebook container setup and started for use. Including launching a browser to the notebook lab environment for you.
From a terminal launch the run_docker_notebook.bat/.sh script; there are a few options on how to launch this:
-
Launch with no parameters and answer the questions regarding configuration (Press enter for defaults) ex:
./run_docker_notebook.bat* Enter docker image [default: quantconnect/research:latest]: * Enter absolute path to Data folder [default: ~yourpathtolean~\Lean\Data\]: * Enter absolute path to store notebooks [default: ~yourpathtolean~\Lean\Research\Notebooks]: -
Using the docker.cfg to store args for repeated use; any blank entries will resort to default values! ex:
./run_docker_notebook.bat docker.cfgIMAGE=quantconnect/research:latest DATA_DIR= NOTEBOOK_DIR= -
Inline arguments; anything you don't enter will use the default args! ex:
./run_docker.bat IMAGE=quantconnect/research:latest- Accepted args for inline include all listed in the file docker.cfg
Once the docker image starts, the script will attempt to open your browser to the Jupyter notebook web app, if this fails open your browser and go to localhost:8888
C# Notebook
When using C# for research notebooks it requires that you load our setup script CSX file QuantConnect.csx into your notebook. This will load our QuantConnect libraries into your C# Kernel. In this setup, the file is one directory above the notebooks dir. Be sure to use the following line in your first cell to load in this csx file:
load "../QuantConnect.csx"
After this the environment is ready to use; take a look at our reference notebook KitchenSinkCSharpQuantBookTemplate.ipynb for an example of how to use our QuantBook interface!
Python Notebook
With Python we have a setup script that will automatically load QuantBooks libraries into the Python kernel so there is no need to import them.
You notebook is ready to use; take a look at our reference notebook KitchenSinkQuantBookTemplate.ipynb for an example of how to use our QuantBook interface!
Using the Web Api from Notebook
Both of our setup scripts for Python & C# include a instantiated Api object under the variable name api. Before you can use this api object to interact with the cloud you must edit your config in the root of your Notebook directory. Once this has been done once, it does not need to be done again.
In config.json add the following entries with your respective values
job-user-id: 12345, // Your id here
api-access-token: "token13432", // Your api token here
Once this has been done, you may restart your kernel and begin to use the api variable.
Reference our examples mentioned above for practical uses of this object.
Shutting Down the Notebook Lab
When you are done with the research environment be sure to stop the container with either Docker's dashboard or through the Docker CLI with docker kill LeanResearch.
Build a new image
For most users this will not be necessary, simply use docker pull quantconnect/research to get the latest image.
docker build -t quantconnect/research - < DockerfileJupyter will build a new docker image using the latest version of lean. To build from particular tag of lean a build arg can be provided, for example --build-arg LEAN_TAG=8631.
Running Jupyter Locally
Note: we recommend using the above approach with our Docker container, where the setup and evironment is tested and stable.
Before we enable Jupyter support, follow Lean installation and Python installation to get LEAN running Python algorithms in your machine.
1. Installation:
- Install JupyterLab:
pip install jupyterlab
- Install QuantConnect Python API
pip install quantconnect
- Linux and macOS: Copy pythonnet binaries for jupyter
cp Lean/Launcher/bin/Debug/jupyter/* Lean/Launcher/bin/Debug
2. Run Jupyter:
- Update the
config.jsonfile inLean/Launcher/bin/Debug/folder
"composer-dll-directory": ".",
- Run Jupyter from the command line
cd Lean/Launcher/bin/Debug
jupyter lab