Are the correlation end points significant

The first question I wanted to address is the role of the arbitrary end points in the correlation results. To do this, I changed the end points from 9 days before and after to 19 days. The following are the highest resulting correlations.

Day1	Day2	Day3	Num Corr	Ave Corr	Min Corr	Max Corr	nVar	Slope	Trend	Last Date
-19	0	19	139	0.476	-0.863	0.952	19	157.61	0.48	2019-05-19
-18	0	19	139	0.472	-0.878	0.951	18	161.98	-0.73	2019-05-19
-17	0	19	139	0.470	-0.875	0.952	17	165.37	-2.11	2019-05-19

Unfortunately, the new end points were the new best results. The average correlations were higher. What does that say? I don’t know. I do know that 2019 has been an overall up year for BTC prices. I also know 2018 was generally a down year for BTC. I ran the 2018 data through the program with the following results.

Day1	Day2	Day3	Num Corr	Ave Corr	Min Corr	Max Corr	nVar	Slope	Trend	Last Date
-2	0	1	365	-0.051	-0.999	0.998	2	-25.83	0.00	2018-12-31
-1	0	2	365	-0.052	-0.999	0.998	1	-25.83	0.00	2018-12-31
-2	0	2	365	-0.059	-0.985	0.989	2	-25.83	0.00	2018-12-31

Here the best correlations were negative with very small end point ranges. Do these confirm that in a rising market you should buy and hold, while in a declining market, don’t be buying. Probably.

In the next post, I’ll bring in the regression co-efficients to see if that adds clarity. I’ll tweet on Twitter, @billlanke, when I’m ready.