GildedPleb
commented
2 months ago Thanks for thinking of this and writing the request!
- For the Power Law, I'm presently using:
const powerLaw = (unixTimeMs: number, offsetYears = 0): number => {
const scaleFactor = 10 ** -17.3;
const unixStartTime = 1_230_940_800_000 - offsetYears * MS_PER_YEAR;
const daysSinceStart = (unixTimeMs - unixStartTime) / MS_PER_DAY;
return scaleFactor * daysSinceStart ** 5.83;I could certainly add your version as well if you can get someone to second that it's sufficiently warranted (a comment here would suffice). I would second, but personally I think all these models are baloney, so I find it a mute point: an 8% return performs better than the current power law on time-frames greater than 200 years, and by extension, an X% annual growth rate outperforms all power law growth on long enough timelines (bitcoin is going to be around in 200 years, and purchasing power will remain infinity / 21M--which is to say, on long enough timelines, Power Laws are not aggressive enough 🤷 ). I included all the models based on fame and diversity, not accuracy and wisdom--which you might have and I most certainly lack. If you don't care about these meaningless machinations and want me to do it anyway, so be it! My rate is 100,000 sats/hour and it would take me an hour.
-
That said, adding a walk which tapers vol over time (or a parameter that tapers walks in general) effectively removing bubble impact, is a fairly interesting feature request, premised on a sound assumption that I have not considered--vol should drop out of bitcoin. If I find the time, I'll look into it! Presently, if you set volatility to 0 with random walk, you can simulate this, but you are stuck at the current price as a % of model range.
-
Drawdown: Technically, you can simulate something like this via the "Inflation" param. Say, if you only think inflation is 2%, you can set the param to 3% to simulate a 1% increase in "Quality Of Life" over inflation. Adding waypoint changes is also interesting, but perhaps OOS, again, all this peters out at certain future distances--if we are lucky, we wont have many years left pricing in dollars anyway!
testing803
Hey there, great job on the tool you built! I've been building something like this on excel but this is way better.
Is it possible to use the Power Law Support Line as a model? I believe the regression will fall towards this over time without bubbles and it is more realistic than the median regression. I've calculated the 1% Quantile Regression in excel as =10^(-17.9289 + 5.9775 * LOG10(DATEDIF(DATE(2009,1,3), TODAY(), "d"))) . You'll notice the community is starting to theorize the support line is the real power law. Otherwise I think it's too aggressive in the simulation.
A second, less important suggestion, is on my spreadsheet I added the ability to add a cost of living increase/decrease in my spending. Say, if I start with 10k on the drawdown. I want to be able to increase my drawdown .5-1% a year or something in real terms so quality of life improves (or I need less). Then I can start very conservative and be more aggressive in time.
Just suggestions and I'd be happy to send you sufficient sats on Alby for your time if you can add these. Thanks and good luck!