International Business Machines symbol IBM daily stock price mean reversion system test results. All tests show losses. Method is tested with 44.95 years of data.
Day trading and short term trading. International Business Machines symbol IBM 10 minute interval stock price mean reversion system test results. All tests show losses. Method is tested with 0.20 years of data. If anyone has more than 3 months of intraday price data then I am willing to test it.
Apple symbol AAPL daily stock price mean reversion system test results. All tests show losses. Method is tested with 22.38 years of data.
It depends on why the stocks went n days down. Find the ones that didn't work and go look up the news. Also determine how correlated the stock is to the S&P. Highly correlated stocks can't get that far away from the market. John
I have data that supports some of the TradingMarkets Research observations. TradingMarkets Research http://biz.yahoo.com/tm/070126/15384.html reports "Average 1-week return of stocks that close down 5+ consecutive days" is + 0.98 %. I am testing a computer program that buys 1000 shares of stock at the opening following five consecutive lower closing prices then sells at the opening five sessions later. Profit may be so small that my adjustment for slippage and commission completely offsets the gain. For example, trading 1000 shares of Federal Express stock (same data used earlier in the thread, 26.31 years, daily price data from 2 January 1980 to 28 April 2006) with no allowance for commission or slippage, no stop loss shows 114 trades and a profit of $ 34520. If I assume when I buy at the opening price the actual execution price is halfway between the opening price and the highest price of the day (and when I sell at the opening price the actual execution price is halfway between the opening price and the lowest price of the day) then overall the system shows a loss of $ 8099. It is possible that tests of some securities might show greater profit. Perhaps I should take this to a new thread since it is not a mean reversion technique as there is no mean.
You can test the article's thesis with NTRI. Today, assuming the stock closes down, marks the seventh consecutive "down day" for the stock. According to the article, we should see a nice move to the upside in the coming week.
Research is based on PowerRating -system, and based on simulation of ca. 6 million trades between 1995-2005. http://www.tradingmarkets.com/.site/powerratings/?hcode=menu I did plot the PowerRatings for NUE, KBH, FCX, OIH, RTP every day for 9 months. It shows ca. "7" on pullbacks and "3" on "push-ups".
It is an interesting topic. The idea is easy to prove using a proper TA tool set. Here is the result I got by feeding the test using 6435 NYSE and NASDAQ stocks EOD data from 1962 to the present. The result is not as good as expected.
In my opinion the research should be on pullbacks with the trend. Use ADX(10) > 30 or RSI(10) > 70 or something like that as a filter.
8.5 million trades now.... "The PowerRatings were created by Cesar Alvarez who is Director of Research for Connors Capital LLC, a money management and market research firm. Mr. Alvarez was previously a senior engineer for Excel at Microsoft. More than 8.5 million daily trades were looked at from January 1, 1995-September 30, 2006. Here are the simulated average performance for over 3000 big-cap, mid-cap, and small-cap stocks applying the TradingMarkets PowerRatings over a 5 trading day period. The performance shows the returns versus the S&P 500 for the equivalent 5-day hold. The stocks which have had a PowerRating of 8 have outperformed the S&P 500 index on average over the next 5 days by an 5.8-1 margin... " http://www.tradingmarkets.com/.site/powerrating?sec=aboutinfo&hcode=demopage