AIQuant
Smarter Investment Decisions
Who We Are
Team
Products
Opportunities
Contact Us
AIQuant Research

Feb. 16 ~ Current

  • Study OpenQuant. Use it to test strategies.
  • Study RapidMiner. Use it to build models

 

Feb. 9 ~ 15, 2009

  • New recruiting process initialized
  • New code base updated and moved to private source control
  • "One-day Two-tades" strategy is finalized
Feb. 8, 2009: AIQuant Research Kick-off meeting minutes
 
Today the AIQuant Research team had the kick-off meeting. We mainly followed the agenda as planned.
 
Meeting Agenda:
 
1. Meeting Introduction (Charley, 3 minutes)
2. Self-Introduction (each people 3 minutes).
3. AIQuant Research Mission discussion: (1). Conducting market research, (2). Pursuing winning trading strategy, and (3). Building automated trading systems, by a group and for a group.
4. Research Tools discussion (use tools to save our time on research) : (1) TA indicator tools (ta-lib); (2) Analysis tools (quant lib), (3) Backtesting & charting tools (NinjaTrader), (4) statistics tools (R, matlab, RapidMiner, Clementine, to be determined and not limited to one.), (5) Automated Trading System Platform (C#, .net); (6) UI platform: web-based (ASP.net)
5. Research Topics discussion: (1) Normalized indicators and stock rating re-analysis; (2) Correlation (stock, sector) and pair trading; (3) Portfolio optimization and risk management; (4) Money flow analysis (Market Maker analysis);
6. Logistics discussion: (1) research by a group and for a group, an agreement shall be signed; (2) actively involved; (3) weekly meeting is mainly for operation discussion, person-to-person discussion, forum discussion will be major communication channel; (4) choose or build a private forum;
7. Book study: (1) Trading for a living; (2) Trading systems and methods.  (Kaufman Perry); (3) "Pairs Trading - Quantitative
Methods and Analysis; (4) Analysis of Financial Time Series.
8. Questions and other issues
 
Most discussion was focused the spin-off of AIQuant Research. It seems most member agreed that this is right choice such that product team will be focused on the product and the research team will be focused on the research. 
 
For next week:

1. Charley will build the new code base for our own simple strategy test. It will be checked into our private SVN host in about 2~3 days.
2. Charley will issue a new thread at aiquant-research@googlegroups.com to discuss our future "Research Topics". If you are not on this list and wish to join the discussion, please send Charley email.
3. The whole team is encouraged to download the current Toydownloader at: http://aiquant.com/Products.aspx. Run it and download all the stock daily data. For coding convenience, please use the default data path setting: D:\stockdata, and D:\portfolio when installing the ToyDownloader.
4. The whole team is encouraged to learn NinjaTrader http://www.ninjatrader.com/webnew/index.htm. Try to realize some of your simple trading strategy with it.
5. The whole team is encouraged to download and learn TA-lib from http://ta-lib.org/
6. The whole team is encouraged to download and learn Quant-lib from http://quantlib.org/index.shtml
7. The whole team is encouraged to download and read the following books: (1) Trading for a living; (2) Trading systems and methods.  (Kaufman Perry); (3) "Pairs Trading - Quantitative Methods and Analysis; (4) Analysis of Financial Time Series and other books from Charley's personal website.
8. Next week we will use conference call plus skype.
Feb. 1, 2009 ~ Feb. 7, 2009
  • Build the team
  • Choose the platform for our automated system, which including 5 parts: data source, market analysis, backtesting, real-time trading and UI.
  • Decide the trading time frame: Day-Trading, Swing Trade, Buy&Hold
  • Discuss the research areas: TA indicators, Candle patterns, Equity correlations, Sector correlations, etc.

Dec. 1, 2008 ~ Jan. 31, 2009
  • Seasonal Strategy on daily data
    . Very interesting strategy called "One Year One Trade" was achieved. This strategy discloses that if every year if buy at the end of October and sell in the beginning of September, the total gain on Dow Jones Index in the past 80 years was 32,602%., while Dow Jones Index gained 860% during this period. The detail is at "One Year One Trade".
  •  
    Research on unemployment rate and the market. The conclusion is that in the past 7 market correction, the unemployment rate reached its high 8~9 months after the market bottom. The details is at "How long does unemployment rate lag the Market?".
  • Research on "How is January correlated to the year?" The observation is that in the past 80 years, for those years with negative Januanry gain (totally 30 ), the averaged yearly gain is -3.93%. For those years with positive January gain (50 of 80). It is 12.9%, which is significant different.