Snakdd2013 wei-20130810

183 views

Published on

My slides for paper "Twitter Volume Spikes: Analysis and Application in Stock Trading".

Published in: Business, Economy & Finance
0 Comments
0 Likes
Statistics
Notes
  • Be the first to comment

  • Be the first to like this

No Downloads
Views
Total views
183
On SlideShare
0
From Embeds
0
Number of Embeds
1
Actions
Shares
0
Downloads
4
Comments
0
Likes
0
Embeds 0
No embeds

No notes for slide

Snakdd2013 wei-20130810

  1. 1. Twitter Volume Spikes: Analysis and Application in Stock Trading Yuexin Mao, Wei Wei, Bing Wang Founder & CEO FinStats.com University of Connecticut
  2. 2. Motivation • Challenge Efficient Market Hypothesis • Stock related tweets can lead to market insight – S&P 500 stocks mentioned more than 11,000 times daily on Twitter • Goal – Demonstrate potentials of analyzing stock related Twitter data
  3. 3. Overview • Twitter volume of S&P 500 stocks • Twitter volume has spikes • interest in stock fluctuates • When? • Surprise? • Causes? • Trading Signals?
  4. 4. Twitter Volume Spikes: Definition • Twitter volume of a stock – # of times stock ticker mentioned in tweets (e.g., # of times $AAPL mentioned) • Twitter volume spike – Daily volume is over K times average of previous 70 days • K=2, 3, 4
  5. 5. Twitter Volume Spikes: When? 46.4% of spikes within one day of earnings day
  6. 6. Twitter Volume Spikes: Surprises? • Implied volatility of short term options • increases dramatically before twitter volume spikes • collapses afterwards • Options more expensive when market events are expected • Twitter volume spikes not unexpected
  7. 7. Twitter Volume Spikes: Causes? Median Correlation Coefficient • Earnings day 0.37 • Implied volatility 0.14 Correlation with Twitter volume spikes • Earnings day • Implied volatility • Interday price change • Intraday price change • Breakout
  8. 8. Twitter Volume Spikes: Trading Signals? • Strategy: Bottom picking using twitter volume spikes – Price retreated at least λ% from recent high • Classification – success: price higher at 10-th day after Twitter volume spike – failure: otherwise • Features – Interday price change rate/average of last 70 days – Intraday price change rate/average of last 70 days – Breakout: price higher/low than previous 70 days – Earnings day: earning report within one day • Bayes Classifier (Naïve Bayes: independent features) – Trading signal: success probability above 0.7 – Training set: 2/21/2012 – 10/19/2012 – Test set: 10/22/2012 – 3/28/2013
  9. 9. Twitter Volume Spikes: Trading Signals? • 14/17 winning trades • Average 15% gain • Best trade: 93% • Worst trade: -11%
  10. 10. Trading Signals: An Example
  11. 11. Future Work • More robust Twitter spike metrics – Total number of followers • Options market and Twitter • Fundamental Analysis • Technical Analysis • More research results at FinStats.com – Put selling strategy – Picks and performance • > 92% winning trades • Every 1 dollar of premium received, 74c kept

×