I have switched my focus back to the rate of change system from the linear regression slope. I would bet a little money that linear regression is more profitable but that is a discussion postponed for another day.
I typically develop strategies on Tradestation. That is relatively easy. At the moment, I want to continue to develop the ROC strategy on Excel. I have programmed market analysis in Excel VBA for over 25 years. Looking back on that period, I never did that to make money. Mostly I like to manipulate data, so market analysis was mostly a way to practice.
For example, downloading market data is an internet subculture.
The new concept (new for me) is to analyze multiple securities simultaneously through the lens of a strategy.
The idea is to decide if the strategy is sound and gain a deeper understanding of how it works in a variety of markets and market conditions.
This is a picture of an Excel worksheet that summarizes the results of the ROC strategy on each of the ETFs. The test goes back slightly less than 500 trading days, starting May 5, 2017.
The strategy seems to be profitable. It is outperformed by buy and hold, the few cases where it did better were not spectacular.. That is OK, if buy and hold was so bad, we wouldn’t compare strategies to it. It stays invested around half the time. There seems to be a positive correlation between Days In and Profit.
SPY Winner Detail
In order to improve the strategy or learn to play it properly, the details behind the summary have to be understood. The summary is built from detail and each trade can be analyzed.
This shows a position in SPY that was entered on April 11, 2018 at 258.74 and sold on the close of April 20 for 261.54 for a decent profit of 2.80. One obvious issue with the strategy is that it will sell after a minor decline, buy and hold doesn’t have friction like that. It would have been much better to sell a few days earlier. A major development goal is to enhance the exit strategy.
SPY Loser Detail
This is a losing trade that didn’t have to happen. It went long on November 30, 2018 and made as much a 5.27 but lost that and more by holding a day too long. This is an extreme example of selling too late, but the trade was also dangerous because of the relatively large percentage of days flat. Glancing at the data, positive trading results inversely correlate with distance between trades.
Dow Indu Summary
This is how the Dow Industrials (with the exception of DWDP) performed over the same period. There are many interesting possibilities with this data in conjunction with the ETF analysis.
The combination of trading algorithms with portfolios is quite exciting, especially for someone who has worked on an Excel VBA program for 20 years without really knowing why.
I plan to look at the major index behavior more carefully first.