Introduction
Blackbird Bitcoin Arbitrage is a C++ trading system that does automatic long/short arbitrage between Bitcoin exchanges.
How It Works
Bitcoin is still a new and inefficient market. Several Bitcoin exchanges exist around the world and the bid/ask prices they propose can be briefly different from an exchange to another. The purpose of Blackbird is to automatically profit from these temporary price differences while being market-neutral.
Here is a real example where an arbitrage opportunity exists between Bitstamp (long) and Bitfinex (short):
At the first vertical line, the spread between the exchanges is high so Blackbird buys Bitstamp and short sells Bitfinex. Then, when the spread closes (second vertical line), Blackbird exits the market by selling Bitstamp and buying Bitfinex back.
Advantages
Unlike other Bitcoin arbitrage systems, Blackbird doesn't sell but actually short sells Bitcoin on the short exchange. This feature offers two important advantages:
-
The strategy is always market-neutral: the Bitcoin market's moves (up or down) don't impact the strategy returns. This removes a huge risk from the strategy. The Bitcoin market could suddenly lose half its value that this won't make any difference in the strategy returns.
-
The strategy doesn't need to transfer funds (USD or BTC) between Bitcoin exchanges. The buy/sell and sell/buy trading activities are done in parallel on two different exchanges, independently. Advantage: no need to deal with transfer latency issues.
More details about short selling and market neutrality can be found on issue #100.
Disclaimer
USE THE SOFTWARE AT YOUR OWN RISK. YOU ARE RESPONSIBLE FOR YOUR OWN MONEY. PAST PERFORMANCE IS NOT NECESSARILY INDICATIVE OF FUTURE RESULTS.
THE AUTHORS AND ALL AFFILIATES ASSUME NO RESPONSIBILITY FOR YOUR TRADING RESULTS.
The trade results are stored in CSV files and the detailed activity is stored in log files. New files are created every time Blackbird is started.
It is possible to automatically stop Blackbird after the next trade has closed by creating, at any time, an empty file named stop_after_notrade.
Blackbird uses functions written by René Nyffenegger to encode and decode base64.
How To Test Blackbird
Please make sure that you understand the disclaimer above if you want to test Blackbird with real money, and start with a small amount of money.
IMPORTANT: all your BTC accounts must be empty before starting Blackbird. Make sure that you only have USD on your accounts and no BTC.
It is never entirely safe to just tell Blackbird to use, say, $25 per exchange. You also need to only have $25 available on each of your trading accounts as well as 0 BTC. In this case, you are sure that even with a bug your maximum loss on an exchange won't be greater than $25 no matter what.
Note: on Bitfinex, your money has to be available on the Margin account.
If DemoMode=false, only the exchanges for which the credentials exist in blackbird.conf are used.
Credentials
For each of your exchange accounts, you need to create the API authentication keys. This is usually done in the Settings section of your accounts.
Then, you need to add your API keys into the file blackbird.conf. You need at least two exchanges and one of them should allow short selling. Never share this file as it will contain your personal exchange credentials!
Blackbird Parameters
Parameter, Default Value, Description, ------------, -------------------, -------------, DemoMode, true, The demo mode will show the spreads but won't actually trade anything, Leg1, BTC, The first leg of the traded pair. This leg is hedged against market risk, Leg2, USD, The second leg of the traded pair. This leg is not hedged against market risk, UseFullExposure, false, When true, all the Leg2 exposure available on your accounts will be used. Otherwise, the amount defined by TestedExposure will be used. Note: the cash used for a trade will be the minimum of the two exchanges, minus 1.00% as a small margin: if there is $1,000 on the first account and $1,100 on the second one, $990 will be used for each exchange, i.e. $1,000 - (1% * $1,000). The exposure is $1,980, TestedExposure, 25, If UseFullExposure is false, that parameter defines the USD amount that will be used. The minimum has to be $10 otherwise some exchanges might reject the orders, MaxExposure, 25,000, Maximum exposure per exchange. If the limit is $25,000 then Blackbird won't send any order larger than that on each exchange, MaxLength, 5,184,000, The maximum length of a trade in number of iterations. If this value is reached then Blackbird will exit the market regardless of the spread. Warning: with this value, the system can exit with a loss so It's recommended to use a large value. The default is 180 days with GapSec at 3 seconds, DebugMaxIteration, 3,200,000, The maximum number of iteration. Once DebugMaxIteration is reached Blackbird is terminated with return=0. Useful for troubleshooting the software, Verbose, true, Write the bid/ask and then spreads to the log file at every iteration. The log file size will be larger but it will show how Blackbird analyses the spreads, Interval, 3 sec., Timelapse in seconds of an iteration. By default, the quotes download and the spreads analysis for all the exchanges are done every 3 seconds, SpreadEntry, 0.0080, The spread threshold above which the trailing spreads are generated to capture an arbitrage opportunity, SpreadTarget, 0.0050, This is the targeted profit. It represents the net profit and takes the exchange fees into account. If SpreadEntry is at 0.80% and trades are generated at that level on two exchanges with 0.25% fees each, Blackbird will set the exit threshold at -0.70% (0.80% spread entry - 4x0.25% fees - 0.50% target = -0.70%), PriceDeltaLimit, 0.10, The maximum difference between the target limit price and the computed limit price of an order. That is the price generated by looking at the current liquidity in the order books. If the difference is greater than PriceDeltaLimit then no trades will be generated because there is not enough liquidity (risk of slippage), TrailingSpreadLim, 0.0008, The limit under which the trailing spread is generated. If the current spread is above SpreadTarget and at 0.70%, then by default, the trailing spread will be generated at 0.62%, TrailingSpreadCount, 1, The number of times the spread must be between SpreadTarget and the trailing spread before sending the orders to the market, OrderBookFactor, 3.0, In order to be executed as fast as possible and avoid slippage, Blackbird checks the liquidity in the order books of the exchanges and makes sure there are at least 3.0 times the needed liquidity before executing the order, UseVolatility, false, If true, display the spreads volatility information in the log file. This is not used for the moment and only displayed as information, VolatilityPeriod, 600, The period length of the volatility in number of iterations. This is not used for the moment and only displayed as information, SendEmail, false, When true, an e-mail will be sent every time an arbitrage trade is completed, with information such as the names of the exchanges and the trade return, DBFile, 'blackbird.db', SQLite3 database file to use for storing the bid/ask information of the exchanges for reference. Blackbird will create this file if it doesn't already exist, #### Getting and building the software
You need the following libraries: OpenSSL, Jansson (v2.7 minimum), cURL, SQLite3 and sendEmail. Usually this is what you need to install:
libssl-dev
libjansson-dev
libcurl4-openssl-dev
libsqlite3-dev (available as a Blackbird submodule)
sendemail
Download the source from GitHub with:
mkdir blackbird
cd blackbird
git clone --recursive git://github.com/butor/blackbird.git .
Alternatively, if you already have the existing source tree use:
git submodule update --init
to sync the submodules.
Once you have downloaded the source code, build Blackbird by typing:
cmake -B./build -H. -DCMAKE_BUILD_TYPE=Debug
or
cmake -B./build -H. -DCMAKE_BUILD_TYPE=Release
then
cmake --build ./build -- install
If all goes well this produces a Blackbird executable in the project directory.
Ubuntu (Amazon EC2 compatible)
- Run the following commands:
sudo apt-get install libssl-dev libjansson-dev libcurl4-openssl-dev libsqlite3-dev sendemail make gcc g++
mkdir blackbird
cd blackbird
git clone --recursive git://github.com/butor/blackbird.git .
cmake -B./build -H. -DCMAKE_BUILD_TYPE=Release
cmake --build ./build -- install
- Run the software, by typing:
./blackbird
Docker
-
Download and install Docker (with Docker Compose) here.
-
Download the source from GitHub with
mkdir blackbird
cd blackbird
git clone --recursive git://github.com/butor/blackbird.git .
Alternatively, if you already have the existing source tree use:
git submodule update --init
- Build the container:
docker build -t blackbird .
- Spin up the entire stack with docker-compose (OSX/Linux):
docker-compose up
Mac OS X
-
Install Homebrew
-
Run the following commands:
xcode-select --install
brew install cmake openssl jansson curl sqlite3 sendemail
mkdir blackbird
cd blackbird
git clone --recursive git://github.com/butor/blackbird.git .
cmake -DOPENSSL_ROOT_DIR=/usr/local/opt/openssl -B./build -H. -DCMAKE_BUILD_TYPE=Release
cmake --build ./build -- install
- Run the software, by typing:
./blackbird
Understanding and debugging the software
Step-through debugging is a helpful method to understand how any software application works. Various scripts and metadata files are included alongside the source code to facilitate building the application so you can use GDB and VisualStudio Code to debug the application while it executes in a Docker container.
Once you have successfully launched Blackbird following the instructions in the "Docker" section, you need to do the following:
-
Download and install Visual Studio Code
-
Setup the C/C++ extension for Visual Studio Code
-
You need to have GDB installed on your machine (this is described in the "Debugging" section of Setup the C/C++ extension for Visual Studio Code).
-
From VS Code, File > Open Workspace... and select the "blackbird" workspace (this is in the root directory of your blackbird clone; the one you created this during "Download source code" step)
-
Press F5 to start the debugger. This will build and launch the Docker container, Blackbird w/ debug symbols, and a GDB Server running Blackbird on the Docker container. Since there is a delay between when the Docker container gets launched and the GDB Server is running, you may need to start the debugger a few times before it attaches to the GDB server. (If you can figure out how to create a "pause" between when the container launches and VS Code attempts to attach to GDB Server, that will fix it).
- If you found a bug, please open a new issue with the label bug
- If you have a general question or have troubles running Blackbird, you can open a new issue with the label question or help wanted
- For anything else you can contact the author at julien.hamilton@gmail.com
Log Output Example
This is what the log file looks like when Blackbird is started:
Blackbird Bitcoin Arbitrage
DISCLAIMER: USE THE SOFTWARE AT YOUR OWN RISK.
[ Targets ]
Spread Entry: 0.80%
Spread Target: 0.30%
[ Current balances ]
Bitfinex: 1,857.79 USD 0.000000 BTC
OKCoin: 1,801.38 USD 0.000436 BTC
Bitstamp: 1,694.15 USD 0.000000 BTC
Gemini: 1,720.38 USD 0.000000 BTC
[ Cash exposure ]
FULL cash used!
[ 10/31/2015 08:32:45 ]
Bitfinex: 325.21 / 325.58
OKCoin: 326.04 / 326.10
Bitstamp: 325.37 / 325.82
Gemini: 325.50 / 328.74
----------------------------
OKCoin/Bitfinex: -0.27% [target 0.80%, min -0.27%, max -0.27%]
Bitstamp/Bitfinex: -0.19% [target 0.80%, min -0.19%, max -0.19%]
Gemini/Bitfinex: -1.07% [target 0.80%, min -1.07%, max -1.07%]
[ 10/31/2015 08:32:48 ]
Bitfinex: 325.21 / 325.58
OKCoin: 326.04 / 326.10
Bitstamp: 325.39 / 325.68
Gemini: 325.50 / 328.67
----------------------------
OKCoin/Bitfinex: -0.27% [target 0.80%, min -0.27%, max -0.27%]
Bitstamp/Bitfinex: -0.14% [target 0.80%, min -0.19%, max -0.14%]
Gemini/Bitfinex: -1.05% [target 0.80%, min -1.07%, max -1.05%]
