Buy The Close....Sell The Open $$$

Discussion in 'Trading' started by Rickshaw Man, Sep 28, 2020.

  1. trade2020

    trade2020

    https://www.bloomberg.com/news/arti...-buy-s-p-500-etf-at-market-open-sell-at-close

    There’s been a reversal of the usual state of play for trading patterns going back nearly three decades, in which the fund has largely outperformed outside of regular trading hours, according to Bespoke.

    Since 1993, buying SPY at the close and selling it at the next day’s open amounts to a rally of 500%, whereas doing the opposite would equal to declines of about 7%. Those gains were even bigger before the mid-February slide started, reaching more than 700%.

    Extreme volatility has gripped equity markets around the world since the coronavirus outbreak began, crippling the global economy and dropping projections of growth rates to levels that would be the slowest in modern history. Equity markets have been whipsawed, with the S&P 500 in March moving an average of 5% a day, the most for any month on record.

    “Investors benefit from receiving as much information as possible, and there’s a lack of information outside of market hours,” said Todd Rosenbluth, the New York-based head of ETF and mutual fund research at CFRA.

    It’s been different this year, with the virus outbreak starting in China before spreading around Asia and into Europe. That’s delivered a spate of negative headlines overnight, leading to wild trading. U.S. stock futures hit exchange-mandated barriers that prevent further gains or losses at least 10 times over the last month.

    Earlier: State Street Warns ETF Traders to Avoid Market’s Open and Close

    “If it’s China, if it’s Europe, sometimes bad news may come more off-trading hours than on trading hours,” said Mark Heppenstall, chief investment officer at Penn Mutual Asset Management, which oversees $28 billion. “Generally, the trend has been for less favorable news on the spread of the virus.”

    The S&P 500 was up 5.5% as of 12:43 p.m. in New York Monday after the reported death tolls in some of the world’s coronavirus hot spots showed signs of easing over the weekend.

     
    #31     Sep 29, 2020
  2. ValeryN

    ValeryN

    Those kind of graphs, over multiple decades, can skew a perception a lot.

    I ran this test myself and it is apparent that most of the performance came from 90-ies.

    Here are all possible combinations for SPY by last 3 decades without compounding.
    1. oc - open to close
    2. co - close to open
    3. cc - close to close
    4. oo - open to open
    2020-10-03_07-05-52.png
    2020-10-03_07-06-12.png
    2020-10-03_06-54-56.png
    Val
     
    #32     Oct 3, 2020
    markd01, userque, yc47ib and 3 others like this.
  3. Trade the idea, ignore the fees and slippage.
     
    #33     Oct 3, 2020
    Laissez Faire likes this.
  4. Nice work. What software do you use for your backtest?

    I suspect you'd get better results on this 'strategy' if some parameter to determine long term trend were used.

    For example, apply the parameter only when the market is closing above a moving average. Maybe a 20 MA.
     
    #34     Oct 3, 2020
  5. I have certainly made some nice wins overnight. My first $2k+ win was an overnight hold of BYND. (sold when I woke up, didn't wait for the bell, but I still would have been many hundreds in the green.)

    Nevertheless, I will wait for a multi-day up trend before I start holding futures overnight. I don't think I would sleep very well.
     
    #35     Oct 3, 2020
  6. Buying the close on big down days proved to be a good strategy. However, its works until it doesnt.
     
    #36     Oct 3, 2020
  7. ValeryN

    ValeryN

    Backtesting software - RealTest.
    Since it's so easy to do I ran a comparison with 20 MA.

    Note - I personally wouldn't consider any of this a tradable edge. Also, I'd fire a guy who would bring me this kind of "research" if he would be working for me. Even for an intern this is bad (referring to nytimes & bloomberg articles where this is discussed)

    Attached how code for this test looks like.

    2020-10-03_12-16-55 (3).png
    2020-10-03_12-18-12.png

    Val
     
    Last edited: Oct 3, 2020
    #37     Oct 3, 2020
    userque, yc47ib and Laissez Faire like this.
  8. Overnight

    Overnight

    I'd fire a guy who presents me with a graph of 5 runs on the x-axis in a legend, yet plots only 3 of them, two of which are so close in color we cannot make out which is which.
     
    #38     Oct 3, 2020
  9. narafa

    narafa

    Those results are with or without transaction fees? Where is that compared to buy & hold?
     
    #39     Oct 3, 2020
  10. Overnight

    Overnight

    You can assume the transaction fee in minimal. The assumption is a single buy at close, sell at open (Or O-to-O, etc.)

    Each is a single trade, not a whole bunch of trades. Otherwise it would just fall under "day-trading" in those timeframes, with all of it's messiness.
     
    #40     Oct 3, 2020