You the one who asked us to comment on this and when you get a comment you whine that we have not done our 5 man years of effort to respond to you correctly? Seriously?
Good for you if you can live outside of time. A nice parlor trick. But from my perspective, WHEN the price of XYZ is at a certain level is important in making a trading or investment decision. Is the price of XYZ down 10% over the past 2 days...or is it over the past 2 years? TIME is a big factor in making an entry or exit decision, eh?
cave man mentality to worship the sun. how about 10% down from the last high or high[1] or high[2] ? maybe you can learn to use array's and vector's but that would take some skills beyond looking at a calendar.
Guys - you are missing the point. This simple system is a hold-overnight approach that doesn't take bull or bear markets into consideration. That may be it's one weakness. Otherwise, the equity curve is not bad at all....even with this limitation. You sell at the RTH close and buy at the next RTH opening. It's so simple, even a cave man can do it.
You say its so easy that even a cave man can do it, but didn't you get it wrong? LOL... You're supposed to buy at the RTH close and sell at the RTH open, no?
Yes, my bad. I was thinking of the bear market scenario....LOL. This system obviously needs some tweaks and enhancements. Always buying the close is stupid IMHO. So how to filter these entries is the big question....50 Day MA/200 Day MA crossover?
For the list of 51 ETFs in the attached lower_corr_liquid_etfs.xls and for the period from October 5, 2006 through July 15, 2019, I ran a simple test using dividend- and split-adjusted open and close prices from yahoo finance. The test simulated buying at the adjusted close price and selling at the next adjusted open price and found overall statistics for (today_open_price - prev_close_price) / min(prev_close_price,today_open_price) * 100 of numValues 161085 sum 4893.70209205738 min -28.4618144460554 max 35.2940922272187 mean 0.0303796262349528 sampleStdDev 1.01546064296849 median 0.0250821730300508 medianAbsDev 0.34083110081868 skewness 0.0966767197618975 excessKurtosis 52.6993377683063 >Thresh_0_Pct 53.34 Note dividing by the minimum of the two prices instead of the previous close represents losses by the equivalent gain to recoup the loss. Also, for SPMD (SPDR S&P 1000 ETF), the test ignored bad-looking data before July 9, 2013. So this simulation showed a low average 0.03 percent per-trade profit similar to my longer SPY-only test in https://www.elitetrader.com/et/thre...it-loves-the-night.329764/page-3#post-4890098 without accounting for transaction costs in either test. And an oversimplified, per-trade Sharpe ratio == mean / sampleStdDev == 0.0303796262349528 / 1.01546064296849 == 0.0299170888062626.
Thanks for that Ph1l - I am not sure about that formula. Shouldn't it be : today_open_price/ prev_close_price - 1 ?? Horrible value for Sharpe, but then again this is a "stupid" system that always goes long....so no surprise there. The system badly needs a proxy for detecting a bull / bear market so it can sell the latter....and buy the former.