Historical data for channeling stocks

Discussion in 'Data Sets and Feeds' started by markd01, Jul 23, 2010.

  1. markd01


    Does anyone have historical data for channeling stocks with at least 10% range? Especially for high volatility sideways markets?

    I do have a backtest for almost 2 years from one source that I'm optimizing right now... but my dataset contains very directional markets - end of 2008 to begining of 2010 - very steep bear market and a very steep bull market.
  2. There will be many others to chose from as well, but you can get several years of historical data for various instruments from www.pitrading.com.
  3. h1l0


    I think you may be asking the wrong question, since any stock can be in or out of a channel, you need to expand your symbol base not obtain the current stocks that are channeling.

    Asking for data on the currently channeling stocks will be of limited value. You need to code your system to identify when a stock is channeling and base your strategy off that condition.

    You didn't say how many stocks you are currently testing, but try the S&P 500; out of 500 stocks you will find many good candidates.

    P.S. I do the majority of my swing trading off of channels too.
  4. markd01


    abattia, thank you, I contacted pitrading.

    h1l0, I do mean stocks that had channeling pattern historically, and it doesn't matter if they still channel currently or not. What kind of backtest were you able to do and what type of markets did you cover?

    The dataset that I am working with is for recent 19 months, and has an average of 50 stocks per month that exhibited a possible channeling pattern.

    These are my conditions:
    1. Same support and resistance levels touched at least 3 times in the past 6 months and reversed
    2. Channel range at least 10%
    3. Stock price >$1 and <$40
    4. 90 day volume > 50,000
    5. Recent support (previous support less than 3 months)
    6. Narrow/quick resistance (channel length is less than 30 days)

    The inputs I'm looking for:
    Support price
    Resistance price
    Date when within +- 2% of support
    Date when within +- 2% of resistance
  5. h1l0


    When you refer to stocks that existed within a channel historically, that is just about every stock. At some point in a stock's history, it most likely was in a channel.

    For my current setup, I scan the entire market of stocks (~7000) from the NYSE, AMEX, and NASDAQ (including ETFs). I then filter for volume and price. This will narrow the universe down to around 1500 to 2000.

    From this filter list I then scan for channels. There are many methods to find and interpret if a stock is channeling. I use a linear regression method.

    1. What method do you use to calculate support and resistance? I use a linear regression line as the center of the trend/channel.
    2. I use a standard deviation of 2. You can also use an ATR look back.
    3. I use > $10 and < $90
    4. I use 20 day MA volume > 500,000
    5. Do you care if the stock is in a trend or do you need a flat channel?
    6. I use a 100 day channel
  6. markd01


    h1l0, I get my flat channel data from StockFetcher.com (several scans for channel and linear regression over 60, 90, 120 days) and services such as ChannelingStocks.com and WaveStocks.com
    I try to backtest in Excel using additional ciriteria in order to optimize my strategy.

    1) Do you have any data for stocks that traded in a channel during say 2000, 2004, or 2005, or any other period when we weren't in a clear raging bull or bear market?
    2) What have you found about descending and ascending channels vs flat channels?
  7. h1l0


    I do my own scans using local software and code I developed over the years. But the idea is similar to StockFetcher; just more customized.

    I do not use any third-party services, but WaveStocks.com and ChannelingStocks.com look very similiar from their samples. Looking at some of their examples from last year, many of the stocks broke their channels on their next leg. Coin toss results.

    In the quick peek I did of their sample stocks, I can beat their results any week. If I had the time (not) or could find a web designer to partner with, my signals could make a nice side business. Any web developers or marketers out their? :)

    Do you mean actual price data (OHLC) or stock results from a past date? In either case, the answer is yes. My testing software can be setup to run simulations on past dates.

    This is a matter of personal style. I prefer an up/down trending channel because they are easier to ride to length and you have a good way to identify a change in trend.

    Flat channels are more likely to break when you finally detect the channel; the shorter the time frame the worse the results. I will take an up/down trend on a longer channel than a short term flat channel. Just a higher probability of profit.
  8. markd01


    >> "In either case, the answer is yes. My testing software can be setup to run simulations on past dates."

    In my backtests, I collect data such as:
    Ag return per month
    Win/lose ratio
    Avg winner return
    Avg loser return
    # of positions per month
    Avg hold time in days for all positions
    Avg hold time in days for winners
    Avg hold time in days for losers
    on monthly basis.

    If it's not too much trouble, would you be able to give me stats for items like above to compare to mine? I have data for, for example, all of 2009, but would also want to see how it behaved in more sideways markets, such as 2005.