I’m just suggesting you look at the chart below and set it up in QT and watch it for a while on a daily basis. You may even want to play with the parameters. This is one of QT’s choices of Regression Channels (RC). Yes, RCs are not all set up the same. The parameters this one uses is in the parenthesis. It uses 0% Change, 1 tick, and a Width of 2. The Width does not impart any additional information, it just makes the thing more visible -- try a width of 0, and you will only get the center line. Going above a width of 4 does nothing. You could see what happens when you use more ticks, or you could set ticks to zero and use percentage width, or you can use both. I have not looked at this from day to day so I don’t know how many candles it takes to change direction -- I selected 1 tick because I think that makes it adjust the fastest. I suggest you try this setup then change the ticks to 2 and compare those two settings over time. I think you will be surprised. With these settings, I think the latest RC top was put in on 7/26 and the down slope started on 7/27 -- but that process needs to be verified. FWIW, my Swap system went to SDS on the small red candle of 7/26.
My point is that you could use these lines (channels) to either confirm your system, or trade these things alone (while maybe using something else for confirmation). For instance, one scenario could be that you would go long SSO only on a positive slope, you could stay long on a flat slope (or maybe not), but you would definitely sell SSO on a down slope and go long SDS at that time, and repeat the rules for holding and selling SDS. In other words, it’s a sort of swap system. I’m pretty sure that process will work very well; and since this process uses straight lines (which are determined by a least-squared best fit of closing prices) there is no argument about the direction of the slope (as there can be with curved lines). The trade process I explained would have definitely put you in SDS near the end of July and you would still be there having made a very good gain.
But the process only works if you take small losses when they occur so that you have the cash to move on and catch a better trade. Sort of like surfing: if you don’t catch a good wave, dump it and wait for a better one. For instance, this system would have you go long SSO in the middle of July and get back out on 07/27 for what would likely be a small loss. But just imagine the gain you would have had in SDS since then -- and that would pay for probably a few years of small losses.
A reminder, change the ticks to 2 and see what you would get and compare that with what your system obtained since your first entry. Use parameters (0,2,2). You will get more flexibility if you set ticks to zero and set % Change to a non-zero value. That will give you the ability to use fractional values. If I were to develop this system further, I would look at using ATR to determine the value of % Change. That would allow adapting this system to account for volatility. Another thing to look at would be to go to cash between positions instead of directly going to the new position directly from the last one. This would allow using a higher % Change value to get into a position and a smaller % Change to get out. That combination would tend to keep you out of horizontal markets where you have to get out with small gains or losses; and when you are in a trade, it gets you out a little quicker.