I had a lot of basic questions on algo functions, so I want to do some concise review of use:
This Algo only yields signals at 5PM Pacific closing.* There are no signals outside that time. There is no other reason for this other than it was what I tested to find its edge and I did so because 5PM Pacific has been a time of handoff between US traders and Asian traders so price anomolies in crypto often showed during that time.
This algo signals whether there is an edge to be all in (100%), 50% or 0%. The only signal that has a price definition is the move from 100% to 50%. The other signals measure momentum, whether a reversal of momentum to go from 0% to 50%, an acceleration for 50% to 100%, or a decay in momo to signal 0%. There has been an edge on being short with 0%, but that is not what I designed it for. It just shows in past data. So I cannot define prices for the momentum 'legs'. Also, the price for 100% to 50% can migrate slightly until close. So, when I post the stop so to speak in the swing trade portfolio it should not be frontrunned until the 5PM Pacific close.
This signal was hand tested, but since getting my hands on better tools, I was surprised to find it shows some long term robustness (walk-forward/ Monte Carlo simulation). However, I want to move us to more concrete price levels for the sake of everyone's sanity so am working with a candidate for replacement. I am likely to be less fancy in the candidate and suggest a binary position (100/0). If I find value in keeping both algos I will. But there is always reason to keep things simple.
*I noticed that my tradingview chart where I track this has not accommodated dailylight savings. I did not account for this and TBH I don't know whether it matters or not. I may force the time change into the the chart but I need to look at it and will know by tomorrow.
GBTCMomo (aka Momentum).
This is the first algo. It was created by hand then a friend ran it through walk forward optimization which led to modification. My buddy and I have since run walk forward optimization, and I have run MonteCarlo simulation. Those tests suggest it is not robust which is to say it may start to fail some time down the road. I can say based on what is under the hood, that it is unlikely to fail except if GBTC goes into a long term downtrend. Funny enough, it doesn't have much room for that with a price of $20. So, there are logical concerns with saying it is not robust. It's just what simulation suggests there is concern.
At any rate, this signal is Pure momentum based on 2H bar closes. I have no upfront understanding of those values, though I can 'guess' or estimate knowing how it works. Those guesses are only partially useful. The requirement for momentum to reverse in both directions makes it laggy. Though back tests look great, this is a weakness. This also means 'support' and 'target' in the tables are useless. If I post them they are arbitrary. This means to trade this algo you have to wait for my notification. To be more explicit, MOMO does not have a price stop but has a momentum condition and I must notify when that momentum decay has completed.
The attempt here was to get a functional algo that showed future robustness and could offer a true price level for the tables for your sanity as a subscriber. That is to say I wanted it more functional for EWT use.
The entries are buy stops based on a complex volatility equation and does fluctuate. I'll post the next entry as a buy stop but it is still based on a 2H bar so it requires a 2H close over the level. Dont' frontrun it.
The stop after entry starts at $1.40 for the first 2H. After that 2H it moves to $1.22 trailing based on 2H closes. That means only move your trail based on the 2H close minus 90 cents. However, once you do, it does not require a close. A breach is a breach so leaving the order in your account works. I do suspect that this 90 cent trailing will start to become obsolete over time as GBTC gains in price. But I am pretty well setup to retest that in intervals and modify that trailing. However, I am going to keep the algo stable as long as quarterly return do not wander outside expectation I do plan to objectify 'expectation' before the next quarter.
I hope this helps. Let me know if questions.
I have lots of things in the works. But work largely has to wait for weekends, and I make baby steps each weekend so I can't give any real meaningful indication of my next milestone, but I'm excited for where this goes.