Hospital and skilled nursing facility

Sorry for the delay in posting. I did have the follow-up surgery a week ago. While it was successful, it set me back physically. I elected go from the hospital to a skilled nursing facility. I spent 100 days here in the spring and was happy with the results. They have good treatment nurses and decent physical therapists. I made many friends when I was here before. I expect to be here a couple of weeks. During the stay, I hope to continue working on the crypto currency pricing project.

The other cause for my slowness is a slight direction change. I had intended to start analyzing specific trading models. But it occurred to me that there was a fundamental question that should be addressed first. Do recent prices predict future ones? If the answer to that is no, then the entire project has no value. So, I’ve decided to look for this answer first. To tackle this, I wanted to look at the correlation between past and future prices. It’s turning out to be more complex than I originally thought. Rather than simply picking a few specific time frames, I am looping through multiple combinations. This has introduced complexity to the code and testing. But it’s the same complexity I expect when I start modeling specific approaches, so the effort won’t be wasted.

Follow me on Twitter, @billlanke, to know when I post some results.

More variations on scenario 2.

Since going to 11 coins in scenario 2 looked good, I’ve decided to explore other variations. I’ve also decided to move away from the 6 periods, and look at 2, 2018 and2019 through August 6th. This will cover a clearly down year and a clearly up one. As you may recall, scenario 2 looks at the percentage change in the last day for each of the  11 coins. I’ve also arbitrarily decided to not make trades on days when none of the 11 coins have a gain. The work spreadsheet contains the details on each coin, the trades, and the daily balances in the 1^st quarter of 2019. It can be viewed or downloaded at http://bit.ly/2YBeJsr. The following table has data from various scenario 2 runs that I made.

Net $	TradePct	Period	Description
-$4,324	1.0	6 separate quqrters	Trades every day
-$7,583	1.0	6 separate quqrters	Trades every day
$11,023	1.0	6 separate quqrters	Trades every day
-$7,585	1.0	2018 full year	Trades every day
-$7,583	1.0	2018 full year	Only trade if top gainer >0
$8,363	0.0	2019 thru 08/06	Trades every day
$11,474	0.0	2019 thru 08/06	Only trade if top gainer >0
$8,248	1.0	2019 thru 06/30	Only trade if top gainer >0
454	Trades	2019 Q1	Trades every day
404	Trades	2019 Q1	Only trade if top gainer >0

The first 3 entries have total results when the 6 quarters were treated separately (starting over with $10,000 at the beginning of each quarter). The “TradePct” is the optimum amount of the biggest gainer’s gain to be used. A value of 1.0 means use all of it, 0.0 means none, effectively buy and hold. There weren’t any intermediate values that were optimum.

The next 2 lines are the result of using 2018 as a single period. It was a bad year and the losses were similar. The next 3 lines are for 2019 data. The “TradePct” in the 1^st 2 lines indicate buy and hold. Basically, forget scenario 2. The 3^rd line ending just over a month earlier had a different optimum “TradePct”. Why? Apparently rapid price increases in those days. An indicator that scenario 2 is unstable.

I won’t be trading with scenario 2. Too many coins spread to thin. So, it’s off to scenario 3 and beyond. However, there might be a pause in the blog. I’m having another surgery this week and will be hospitalized for at least 5 days. Follow me on Twitter @billlanke to know when I return.