Hakim Khalafi Fall 2013 Project Class: Peer -to-peer - - PowerPoint PPT Presentation

Hakim Khalafi

Fall 2013 Project Class: “Peer-to-peer networking" Instructor: Dario Rossi

Bitcoin becoming very popular!

 Most studies focus on either

• Bitcoin transaction graph (Harrigan et. al)
• Bitcoin mining weaknesses (Barber et. al)

 But there is very interesting financial activity!  This study

• Trading strategies – algorithmic trading
• How to use time series other than the actual

financial data, to predict buy and sell signals

Moat, Helen Susannah, et al. "Quantifying wikipedia usage patterns before stock market moves." Scientific reports 3 (2013).

 Data mining

• Four time series, three sources, a lot of python.

 Bitcoin to euro conversion: Bitcoincharts.com/charts  Wikipedia search volume: stats.grok.se  Transaction traffic: Blockchain.info/charts  Transaction volume: Blockchain.info/charts

 Spike detection

• Simple algorithm to detect spikes in time series in matlab

 Hypothetical trading based on historical data

• Toy-trading with 100 euro initial investment in matlab

 Implications

• 2011- 2013: Trade based on transaction volume!
• 2013-Future: Trade based on Wikipedia search volume
• Change of regime in-between, why?

 Limitations of study

• Buying and selling transaction fee not included
• Haven’t considered data for november and december
• Time series - daily resolution. Why not use hourly?

 We focus on strategies viable for average investor, not high frequency traders!

 Future studies

• Trading based on wikipedia searches for other search

terms and assets

• Trading based on time series correlations

 What I initially considered but did not succeed with

 Conclusions

• Before 2013, trading bitcoin based on transaction

volume might theoretically give you higher profit than buy and hold

• Now, trading bitcoin based on wikipedia search volume

might theoretically give you higher profit than just buy and hold

[1] Nakamoto, Satoshi. ”Bitcoin: A peer-to-peer electronic cash system.” [2] Moat, Helen Susannah, et al. ”Quantifying wikipedia usage patterns before stock market moves.” Scientific reports 3 (2013) [3] Bordino, I. et al. Web Search Queries Can Predict Stock Market

• Volumes. PLoS One 7, e40014 (2012).

[4] Barber, Simon, et al. ”Bitter to Better - How to Make Bitcoin a Better Currency.” Financial Cryptography and Data Security. Springer Berlin Heidelberg, 2012. 399-414.

Questions?