Wave arbitrage is a trading algorithm that takes advantage of waves in prices. The basic idea is to continuously rebalance a portfolio to a 50-50 ratio. This algorithm has a higher expected value than a buy-and-hold strategy given that the market is a "fair game", prices don't go to zero, and market prices fluctuate. This algorithm is a disproof of the efficient market hypothesis.
The current version of the paper is at paper/paper.pdf.
bazelclangpython3- Probably others.
git clone https://github.com/LoganEvans/WaveArbitrage.git
git submodule update --init --recursive
The scraper.py script fetches data from IEX and converts it into protobufs
It stores data in $HOMEDIR/iex_data. It also takes a long time. In order to
speed things up, you can use script-level parallelism. I'm not proud of how
this works -- well, maybe I am a little.
bazel build :scraper
for _ in `seq 10`; do bazel-bin/scraper & done
This takes quite a while, although not as long as collecting the data.
bazel run -c opt :backtest
The backtest binary prints out json representations of histograms. The
show_histogram.py script can render multiple histograms together as long as
they are all piped through stdin. As an example:
lib/DynamicHistogram/scripts/show_histogram.py <<< `cat results.txt`
The script flip.py and the code in simulate.cpp are simulations that run
wave arbitrage and buy-and-hold against some basic price models. flip.py
appears to be using a submartingale. simulate.cpp is using a martingale.
python3 flip.py --help
bazel run -c opt :simulate