Sectors to exclude

Discussion in 'Strategy Development' started by markd01, May 25, 2011.

  1. markd01


    For those of you swing trading mean reversion, do you trade all stocks or choose to exclude certain sectors, such as:
    1) OTC / Pinksheets (less liquidity, higher chance of fraud)
    2) Biotechs (extreme overnight gaps if FDA approval/disapproval)
    3) Chinese small caps (extreme moves when accused of fraud)
    4) Commodity linked ETFs (tend to trend)
    5) Commodity related stocks (tend to trend)
    6) Low volatility large caps (not enough profit potential)?

    Thank you,

  2. lindq


    That's a good list. In addition, you might want to add (1) Companies that have just warned or issued poor guidance, (2) Any pharma or medical device company that can be impacted by regulatory news, (3) Companies with seriously negative EPS. Over time, you will find in most cases that if you are buying into a pullback those companies with positive numbers will attract more buyers on weakness.

    And a good rule-of-thumb with the strategy once you've honed your instincts is: If in doubt, stay out.
  3. markd01


    Thank you, lindq.

    Are you a discrecionary or a mechanical trader?

    Would you be able to share some backtest or live trading metrics you got form excluding a certain group of stocks? For example, by excluding pharma stocks, reward (CAGR, Profit Factor) decreased by only 5%, while risk (Ulcer Index, Max Drawdown) decreased by a whopping 15%..

    When it comes to earnings or guidance, for how many days from that event do you keep on excluding those signals? I'm trying to define PEAD (Post Earnings Announcement Drift) length..
  4. lindq


    I 'swing traded' for years but moved my trading to intraday only. Same system, much shorter timeframe. I am completely systematic, but I don't shut off my brain if something smells bad.

    I don't have metrics to share. IMHO, you may be getting farther into the weeds on that than you need to be.

    Regarding warnings, etc., it is often dead money for longer than you want to tie up capital. I don't generally have a problem trading an earnings miss or a downgrade if it falls into my system. But a warning...foggataboutit.
  5. markd01


    Do you find it valid to exclude individual stocks because they have not performed well in a backtest? How many trades would it take and over how many years or market cycles to make it statistically valid?

    For example, I got burned on silver etf's lately. Backtesting AGQ, my main strategy would have traded it only 12 times staring in December 2008, ending up a net loser both going long and short.

    I would feel more confident that something is not suited well for my strategy, if I had 100+ trades over 20 year period.. that's why I normally look at larger groups of stocks as opposed to individual issues.
  6. #3 Exclude any ADR's from any country

    #4, #5 Maybe better off trading the commodity directly, not sure why "tend to trend" would mean exclusion

    #6 Exclude low volatility any cap, although realize there are many definitions of "volatility"
  7. lindq


    I don't exclude stocks that perform poorly in an otherwise good system run on a portfolio, unless the stock is in a sector that I would generally not trade at all. I am more interested in overall statistical probabilities of the entire portfoflio. That's just my personal preference.

    EFTs will not perform in the same way as an individual equity. They have different characteristics, and respond in different ways. They're just a completely different animal.
  8. markd01


    Thank you, Stoxtrader.

    Would you elaborate on why you excluded the above, especially with any metrics you have?

    For example, in regards to #6 excluding low volatility stocks.. I just ran a baseline test on my system and a second run where I looked for stocks with 20 day ATR of less than 3%, my definition of low volatility. My low volatility set had exposure adjusted returns almost 3 times lower, slightly better Win%, 3 times lower Maximum Trade Drawdown, Profit factor some 37% higher, average trade p/l some 35% lower, and slightly higher Sharpe ratio. In other words, low volatility mean reverting stocks are safer, but also offer lower return.

    RE #4 -- how sucessful are your mean reversion strategies trading commodities or commodity linked ETFs? I've developed my strategies focusing on equities.
  9. I'm curious to see why you think they are safer but offer lower return - wouldn't a higher PF and Sharpe indicate that you could provide better risk adjusted returns (meaning more return for less risk)?

    You'd have to leverage up the system to provide the same absolute amount of return, no? Thanks.
  10. markd01


    They are safer but offer lower return (in absolute terms). The trick is to choose a proper objective which takes into account (relative or risk adjusted) risk to reward ratio -- PF, Sharpe ratio, CAGR/MaxDD, Ulcer Performance Index, etc.

    Yes. Or look for other better opportunities (more trades) in place of the eliminated low risk low return trades.
    #10     May 26, 2011