Actually, volatility is lower in the close-to-open vs open-to-close periods, while the drift is higher (for SPY since inception, for example): Code: 'c2o' : { 'sd': 0.00620, 'mean': 0.00032 } 'o2c' : { 'sd' : 0.00972, 'mean': -0.00003 } So it is kinda a real alpha, but it's so thin that it's not usable (some variations are, though). I have seen multiple explanations for why most of the drift happens overnight. The one that resonates the most with me is the liquidity/risk premium based - since you can't trade overnight, you get some compensation for being a captive holder. Then again, being a volatility person I am probably biased. To be frank, some people are. You do come across as arrogant and disinterested in learning things. It's very hard to thrive in this business if you think you are the smartest (guarantee that you are not).
So these trades are assumed right before close/right after open, or right after close/right before open?
My quick and dirty study assumes perfect MOC and MOO execution, with zero transaction costs. I can even attach a spreadsheet if someone would teach me how to do it here.
I don't "know" and never claimed to "know" why this effect happened during the time period chosen. I simply pointed out that it made sense that holding during the time period when news gets dropped simply amplifies the direction the market went during the time period you chose. The same as buying beta. And I pointed out that the Sharpe ratios (that's data, just so you know) tend to support that. That's what we call a hypothesis. Intelligent well meaning adults would agree or disagree with that hypothesis with reasoning. Blindly citing data sets that do nothing to prove or disprove that hypothesis isn't reasoning. Not a huge fan of Barbados, as it happens, but thanks for asking. Good place to store your boat during hurricane season but quite a few places in the Bahamas and Caribbean I enjoy spending time in more. If you want to talk Caribbean though happy to change the subject, but I'm more of a sailboat guy than a mansion guy, just to forewarn you.
I wasn't disagreeing that people are getting frustrated, I was disagreeing that "everyone" was. It's always dangerous to throw out absolutes like "everyone", "no one", etc. It's just more loosey-goosey statements that have nothing to do with the thread topic. I am interested in data and facts. I am disinterested in people who like to use others threads as their own personal soap boxes. Look at the 10 pages of this thread and my response to each person. People who came with something to add, such as more articles, data sets or analysis I welcomed and thanked. People who came with a soap box, people who wanted to argue with the data (apparently it's a figment of our imaginations?), and people who wanted to argue about random things, they get no welcome. You can call it whatever you want, I am trying to keep my thread focused on topic. I was very specific in what I was looking for. As usual, those who like to soap box came and soap boxed, and those who tend to add value, came and added value. After my short while here on ET I have pretty much figured out who is who.
My guess is they are taking the most simple approach of open/close prices, but I am not sure. I have seen a number of studies on it now, they all show the phenomena but each has slightly different results. I think the discrepancies are explain by varying start/end dates, the exact entry/exit times, and I think some may be using relative returns and others absolute returns. I'm currently downloading a few years of tick data for SPY so I can start testing this myself and tweak the entry/exit times. My data is going to go back at least 5-6 years. If I want to go back the full 25 years that is in the original post I will have to go for OHLC daily data or perhaps hourly if I can get my hands on it. But I would value the test more with minute or tick data. I will post some of that once it is ready.