I implemented the PT_LSTM for Argonaut about 3 weeks ago but have been pulled into $dayjob again for another surge of attention demand to get through audit season, so, I haven’t even had time to journal the update.
It went well, the numbers in the new system are manually created in an iterative fashion, so, I’m getting much better results instead of relying on internal library generation for metrics. This improved prophet accuracy as well for whole market analysis. The only problem is that prophet, and most of these models, are not much better than random when using only price data with a volume regressor, still. I’ll need to spend some time very carefully analyzing what metrics I can incorporate into the historical data to improve accuracy.
In terms of next steps, because I’m one person with alot of stuff to do, I’ve got to hit the pause button and revisit it, which I don’t like doing because I have to “get my head back in it” whenever I come back from a pivot, but, that’s just how it is, sometimes.
There are upcoming items.
First, the lease on my house rental is ending soon.
I’ve been considering moving to Pensacola over the next few months. Ohio has been great, but I think Pensacola will be more fun.
One of the big concerns there is if I ever need to change jobs, being that far away from Columbus and locked in to an area not near a major tech hub is a disadvantageous position to be in with all these companies’ executives pushing return to office mandates to justify 10 layers of middle management. This means if I end up out of work for whatever reason later, and land my next gig, I could potentially be in a position where I have to relocate to execute the role, and would potentially not have the funds to relocate in a timely manner. That would be bad.
The benefits are a warm climate, clean beaches, and some boating options for fun, which I very much want to be able to do alot of. It might reduce my need for vacations if I’m in a nice area like that. I’m getting old, and I don’t want to do as much as I did when I was younger.
Truthfully, I’m crazy for considering it because there are key pivots happening in slow motion that, depending on which way they go or do not go, may necessitate a job search, which would make the move a certain mistake. It’s one of those things where, “the timing is right” in certain conditions, but, it’s very much not in others. It’s easy to say “I’ll just move there for a year or two”, but, it never works out like that when rubber hits the road. Moves are also expensive and I just paid off the car. I might put it off for a year and relocate in Ohio, and just vacation in Pensacola this year. I actually hate moving at all, and this would double the number of times I’d move, but I really don’t want to take that big of a risk right now without more solid financial security in place in this economy. While I live well, these are hard times, and an extended period of no income would hit me just like it would almost anyone else, only in my case my reserves are still recharging from some debt restructuring I started last year that has played out well.
Then again, just playing out the scenario, Pensacola is an 8 hour drive from Galveston, which, I have family in, and Galveston is a 3 hour stretch from Austin, which is another thriving tech ecosystem. If I had to be in the office, and near a tech hub, that’s not ideal to have a 3 hour commute for a month in the worst case scenario but it’s able to be done. Galveston is also an hour and a half from Houston which is a slightly smaller but still a prime tech hub in the US. Galveston is 4 hours from Dallas, which is another thriving tech hub. Pensacola is also a 6 hour drive from Tampa, which is one of the top emerging tech hubs in the country, but doesn’t have a springboard point near it that I could bounce off of to transition there if I ended up needing to be in the office, so, it’s looking like my parachute would be westward. All of those scenarios are complicated by the dog.
Second, I need to create some safety padding with the Dayjob. Dayjob is a little weird right now. It’s probably secure, but it’s approaching my risk tolerance (and my patience threshold for certain things), so it might be safe to have a few options on the backburner just in case. I’m not expecting to take any offers currently, but, it’s more of a covering my arse type thing and I wouldn’t consider it an “active job search” so much as “maintaining options”, which I think really everyone should actively do in this day and age. This is not our grandfathers’ job market.
Stuff like that can go either way, so I’m not actually expecting to make a hop, it could also go very well if I stay put if things go the direction I want them to, and then I won’t have any issues I can’t just deal with. Sometimes the right thing to do is to tolerate the bullshit and get through it, but it never hurts to be safe — just need to plan for the long haul and manage risk exposure.
Regarding the move though, that may need to happen. This current rental keeps getting rent hikes, and it’s reached a point where I could just lease a 3 story house for what they’re charging, and I want to get out of here before the appliances start going out.
It would probably be safer to stay in Ohio for another year and save up for a Pensacola move planned a year from now when things are more certain. I could then use reserves as safeguards against needing to actually execute any of those scenarios.
Third, I want to revisit two projects and get an iteration in. One is SME-D, which I have an idea for “proving it out” that also meets some personal milestones. The other is a package manager for Dark Horse Linux. I’ve been analyzing it passively for a year or two now, and I’ve decided that RPM is full of legacy cruft, unused features, overengineered workarounds due to requirements identified after it was settled in design as a package manager, and so I think not designing a Dark Horse package manager and just using RPM will create unnecessary overhead in package maintenance by maintainers later.
I’ll need to spec out a new one that very closely resembles (I believe right now) what Slackware uses, only has extra features like network-based repositories and dependency management, file digests, digital signing etc. so that it can do all the same things and meets the security demands that RPM and DNF meet. Truthfully, it’ll have alot of the things RPM and DNF have, only a little more modern and a little more simple in how the pieces work. I’ll want to read the criteria for FIPS 140-2/3 and LSB about making sure that it can meet those standards, and it should also meet a plethora of enterprise use cases.
This will of course involve both a refresh of Dyad to the latest LFS and a refresh of Pyrois so that people can actually use it. Once a package manager is in place and relatively mature as its own project, it will make building the installer ISO much easier for a package-based descendant of the source-based product that exists today.