Viper, Automated Bitfinex Swap Trading
Bitfinex is was one of the first bitcoin exchanges to offer bitcoin trading on margin. This opened
up a whole new opportunity for traders as they could now use margin to create significant gains.
This also increased the liquidity on the exchanges substantially. Bitfinex has a side market of
margin swap contract trading. This allowed traders to either lend bitcoin or USD at a preferred
interest rate, or loan bitcoin to increase there own liquidity and take out larger trades at
a preferred interest rate.
How can one loan out the coin while getting the best interest rate possible? are there data feeds and charts? How can you trade this? This is the questions we had when first looking into this market as a possible investment opportunity. Being the market that drives quite a bit of the bitcoin market on Bitfinex, we were surprised to find a lack of data. No charts, no high speed data feeds, and no analysis. After realizing this market could be changed in a big way, and statistically modeled we started researching and running simulations.
This has evolved to us creating the Viper software and the Viper dark pool. Viper is a program that provides advanced statistical analysis on the swap market, and is able to predict the probability of a specific interest rate being taken in the next hour, and make swap contract trades based on this. The dark pool allows us to provide a large amount of liquidity to the margin market, at better rates.
We are extremely excited to announce this, and already have started development of the software. The algorithm has been finalized and is being implemented within the software. We expect to have a production version running within the next few months.
Bitcoin Volatility Index (BVX)
Bitcoin has been known to be a very volatile asset, and tends to whipsaw traders out of there trades.
this is one of the reasons why some traders hate trading it so much. We aim to create a volatility
index across the whole sector. Getting prices from every exchange and calculating the standard
deviation in price.
We are currently building software to do this, and to do this very fast. Using python, we aim to get tick by tick, millisecond exchange data, and archive it in mass quantities. This allows economists to perform research on mass amounts of price, swap, and spread data. It also allows us to perform our own research and develop high frequency trading solutions for the market.
Deep Neural Networks with Stacked RBM
Neural networks are a rather amazing part of science, Taking software, math, and modeling data processing
in the same way as the human brain. This allows us to do some pretty profound things, including predicting
price movements in socks, forex, and commodities.
Right now we are testing our models within python, and they are still being developed and researched upon. This is a rather large task, and lots of research and development will have to be done. When we have a working model, that is back tested, we will port it over to MQL4. This will be available for license on the store.
Raspberry PI Compute Cluster For Machine Learning
Because of the high data processing requirements of neural networks, standard computers can not accurately
test the true parallel processing power of the networks. Large data centers of high density computers are
very expensive and not available to the normal trader. Neural networks can be made to run on MPI clusters,
greatly expanding the capability of the program, and accuracy.
Once trial runs are complete, and some actionable data is obtained, we can start porting the platform to the Parallella Board. The Parallella Board is a credit card sized computer, like the Raspberry Pi, but with an 16-core Epiphany Coprocessor, and 1 GB of ram. This platform will be available completely pre set up. with software already installed to make predictions on data sets, run simulations, and perform automated trading.