Seeking Alpha

Henry Bee

 
View as an RSS Feed
View Henry Bee's Comments BY TICKER:
Latest  |  Highest rated
  • Creating Robust, High-Performance Stock Portfolios [View article]
    "And since these are all limit orders, only some of them will be filled"

    Limit orders have capacity issues. Therefore, using limit orders for backtesting only validates the idea. For it to be actually useful you will still need to have a good execution system to get prices close to, if not, better than the limit price. The drop off in Sharpe isn't linear either, as the amount of capital that uses the system increases, the Sharpe and CAGR drops off substantially if you can't get a good fill.
    Aug 3, 2014. 05:05 PM | 1 Like Like |Link to Comment
  • Creating Robust, High-Performance Stock Portfolios [View article]
    What is the strategy's returns if you assume a fill at next day, next week, next month's closing prices? This is a crucial test for robustness of a strategy. If the system relies on out-trading everyone else, and the quality of high frequency execution (which most people don't have access to), then the decline in Sharpe will be substantial as you use less aggressive execution assumptions.

    Most people don't have access to any type of automated execution.
    Aug 2, 2014. 02:46 PM | Likes Like |Link to Comment
  • Creating Robust, High-Performance Stock Portfolios [View article]
    Jonathan, what is the turnover in the underlying strategies? I think the problem is that most advisors and individuals cannot and do not wish to trade so frequently. Lower frequency strategies will have lower Sharpe, and the improvement has to come from improving the estimators, instead of trading around the underlying.
    Aug 2, 2014. 05:33 AM | Likes Like |Link to Comment
  • Best Hedge Funds: Jim Simons' Medallion Fund [View article]
    hi ian,

    david varadi and I here at css analytics have come pretty close to reverse engineering the medallion fund's returns. a rolling 5-year correlation analysis shows that our systems returns was 100% correlated to medallion's returns from 2002-2007.

    [slide 1] dl.dropbox.com/u/12156...

    A very simple ensemble performs surprisingly well in walk-forward testing.

    Only has 3 components
    short-term mean reversion
    Intermediate mean reversion
    Long-term trend

    Transaction costs around 2% annually.

    [slide 2] dl.dropbox.com/u/12156...

    By looking at the returns of the Medallion fund, we can get a rough idea what’s driving the returns and the type of strategies being utilized.

    Keep in mind that our simple little ensemble only consists of 3 basic components applied to a single market (S&P 500). Returns would be even better if improvements were included and applied to multiple markets.

    [slide 3] dl.dropbox.com/u/12156...

    Medallion’s returns since 2003 have become highly correlated with our simple ensemble model.

    Because the ensemble has increasingly allocated to short-term mean-reversion strategy since 2003, we can deduce that mean-reversion type strategies have been driving Medallion Fund’s returns in recently years.

    cheers,

    henry
    Jan 2, 2011. 06:24 PM | Likes Like |Link to Comment
  • Why to Avoid Heavily Promoted Investments [View article]
    Welcome to the community. I completely and utterly agree with your view.
    Jun 4, 2009. 11:53 PM | Likes Like |Link to Comment
  • China Sky One Medical: Response to Cabeza Howe [View article]
    "The hospital’s stamp was included on the document but could not be transferred to this post."

    This raises SERIOUS question as to the validity of the trial. Even if we assume the source is reliable, there are still many weaknesses with the trial:

    1. Sample size is too small.
    2. No explanations of the recruitment method for subjects.
    3. There were no follow up studies. Do subjects maintain their weight loss after 1 month? 3 months? 6 months?
    4. Subjects in the control group and the test group were not matched based on activity level and lifestyle.
    May 20, 2009. 03:49 AM | Likes Like |Link to Comment
COMMENTS STATS
6 Comments
1 Like