This is funny, do you realize that you are denying a hard fact based on an (incorrect) interpretation of what glancing at a chart suggests to you? If you take a minute to think about it, a chart that is full of filled gaps (not full of unfilled gaps) is precisely what you would get if 100% or more of real gains happened overnight. EDIT: replied w/o seeing all the other posts that followed this. Looks like the denial is strong with this one.
Pekelo, you seem to be very focused on this particular scenario of a gap fill followed by a rally. Just because this scenario can happen doesn't really mean anything by itself. If you want to prove that using your own definition, all real gains actually happen during market hours, you're still going to need to do the actual work to prove it. Right now you are just making claims based on looking at a chart. By the way, have you considered that gaps often fill more than 100% by the close? That means the day loses all the night gain and then some.
1. I didn't say ALL gains happen during RTH. I actually demonstrated that about 85% did this year, and 15% during after hours. It probably varies a bit in different market conditions and time frames. One of you guys can run a sum on it. 2. You want actual work? How about today? We gapped up 6+ points in cash. Went higher and reached 11pts gains. Then it all fell apart and by just after 2 pm we closed the gap thus if you held long from yesterday's close, you had exactly zero gains. Now if we gain or lose today, the last 2 hours will be actually responsible for it, because we started from scratch at 2:05 pm. Beat this argument. 3. Gap fills can take multiple days, hell, weeks. Since we are talking here longer term gains, if a huge gap gets filled after a week, that week gains has just evaporated. So once we are back into where we started it didn't matter if that huge gain was achieved in after hours, it is like there was no gain at all. All I am doing is explaining data to you, not denying it. This thread reminds me Tesla analysts. One group says: they are going bankrupt tomorrow. The other group says: It is going to $4000. Yet both groups are looking at the SAME data.
Here is another study to add to the archive. This one shows the phenomenon is alive and well for gold also. http://jayonthemarkets.com/2015/11/...Wm2Smq9mmYIwlJTIjVsSs8fHPczkW7HNC0ucLJkgKZJtI
What a strange day! After all that upgap, 5 minutes until the close and we are back to breakeven!! My, my, what a strange world we live in... [shaking head, after the market makes his point for him] Edit: Closing with a huge 1.5 points gain...
Talk is cheap and everyone here who has posted there own studies and data sets are significantly more valuable than all of your jaw boning. If you want to prove something, then by all means post a study and add something to this thread.
Ok, I can't believe I actually put in the work to do this, but here is a quantitative study. You will probably notice that I'm not nearly as adept as most people here given my rudimentary methodology, but I believe its accurate, and would replicate exactly what a trader would do in order to actually trade this method. I took the charts that I posted earlier, exported the data, and added the totals in Excel. Once again, this is how the bars are drawn (for the one guy who didn't get it). The open of the bar happens at 16:14:59. Lets assume that you enter the market on the ES at this time and get a fill just before the market closes. This would be the open price of the bar. The close price is at 09:29:59, just before the RTH market opens. Then its a simple calculation of open minus close to arrive at the total made or lost for that day. I am presenting 2 charts. First is from the end of September just before the drop. There are 170 days, and the total comes to 153 points. So now the caveats. The price that we get for open and close might not be the price we get because of the bid/ask spread. We should therefore assume a tick extra for both the open and close. Also, commissions will be about 1/3 of a tick (ie. $4 cost to trade, which is 1/3 of the $12.50 you get from 1 tick profit). So putting it all together, we have 153 points over 170 days, hence 0.9 points profit per day. Now we subtract 0.5 points for realistic fills taking into account the bid/ask spread, and we get a profit of 0.4 points average per day. Then we lose 1/3 tick for commissions, so our net is 0.32. This means over 170 days, we make 54 points, which comes out to $2,700. This sounds like a good number for minimal work, but 0.32 points per day is pretty shitty. Lets look at the data going back to 2013. Here we see a total of 759 points over 1692 days, which works out to 0.45 points per day. Once we subtract losing a tick on entry and exit, we are already negative. I have no idea how any of those other studies calculate the data that makes this strategy sound awesome, but based on what I would do as a trader to replicate this, its a pretty shitty strategy based on the hard numbers from trading the ES. Maybe one could introduce hard stops on certain outliers like being down 20 points, or one could enter just before 16:00 as opposed to just before 16:15, but I doubt any of this would matter much.
I think that is where some of the people here got off track. Theses aren't studies to examine real-life trading strategies, they are studies on a market phenomenon. Can someone figure out a winning strategy to exploit the phenomenon? Probably. The goal of this thread was to study the phenomenon in it's entirety and from all angles, and then from there people can study to see if any strategies can be gleaned from it. A few people here have already been discussing if there is any edge in trading it daily, and it doesn't sound like it. I personally am still studying the phenomenon itself and not getting wrapped up in developing strategies. The phenomenon itself, ex-overhead trading costs, is already much more compelling than buy and hold. If certain correlations can be found, the volume of trading reduced, and risk adjusted returns maintained or increased, then there might be a trading strategy to be had here. The market's aren't going anywhere, so I am going to keep studying and posting any new bits of info I find.
I would conclude the opposite. Since we can't really exclude the costs, buy and hold is actually better. Plus, it takes just one day of missing an entry to lose perhaps a 20 point winner which will have a large impact on the bottom line. Also, given how this is a trading forum, buying the close and selling the open isn't exactly trading in my opinion. You're limited to just one trade a day, with no filters. Once you add filters, now you have even less trades. Sure, you could find a trigger for when that one overnight session per week will be the winner, but by the time you forward test this enough, you're probably looking at too much time. I know that day trading sucks money out of most people, but you get to see the results of your work faster, and hopefully make adjustments sooner. It sounds like you're eager to crunch data and make it work, and I would personally focus on a smaller time frame.
The gold study seems to support basic correlation. Swapping stocks in daytime for cash is bad for gold if fear/vol is not so bad. So , overnight is stocks, gold and possibly treasuries down as well, USD up , to be reversed in the US day session? Some serious extrapolation/guessing here on my part.