Glitchet logo

Glitchet

Subscribe
Archives
May 14, 2020

Temporal Blindness and the Market

Have a song to read this to: Start Again - Kidnap Piano Mix. Melodic with a lovely, easy vocal and pulling beat and piano key elements.

Strap in. This is a long one. (With perfectly skippable sections.)

In 2016, Sam Winston, a 37-year old British artist, wielding over 200 meters of duct tape and blackout cloth, plunged himself into total darkness for a week. Then, two years later, he did it for an entire month--672 hours.

Sightlessness yields to the bizarreness of the mind filling in the blanks. Sam would feel time slipping away into something-like-morning, kinda afternoon, sort of evening, sure-I-guess-night. His brain would paint in where it thought the guesthouse should be, and oceans, landscapes, a cloud-filled sky, fields of stars.

It became difficult to tell the difference between dreams and hallucination. Old memories bubbled up with impressive obscurity (why was he such a dick to his first girlfriend 20 years ago?), and once or twice wondered if he had died, lost to fantasy.

Mitoraj, Riccardo Di Trani

While not in the dark, it does feel occasionally as if my ability to sense time has eroded, only tenuously held together by the mild awareness of clocks and the realization that the sun is coming up. I find myself gravitating to an essential very few routines that tie me to any sort of semblance of routine: hygiene, cooking, I started a side hustle doing tarot divination for people (see last section if you want one), both to make a little side cash but also to ensure that I have some sort of schedule that tracks daylight and the stock market.

How to (attempt to) predict the stock market with astrology

As I promised this last week, I'll discuss the deeply strange research I've been doing on and off for the last 2 years. I'll try to keep it brief.

A while ago I forced myself to learn some astrology in order to understand tarot correspondences better, and instead of engaging with it in a sterile fashion like I expected, it rocked my world. I also learned that while it's incredibly easy to lie to yourself (and other people) about who you are, astrology proves its efficacy extremely quickly when you begin to do synastry, or compatibility astrology between partners. Major relationship themes, chemistry elements, and problem points are usually immediately apparent in a chart and I found myself able to point at the primary aspects of relationships with ease.

So I thought, why not automate it? Of course, like any responsible programmatically-minded person I realized that I need to collect lots of data to determine what the real correlations are between relationship dynamics and synastry, which implied hundreds (thousands (millions)) of interviews. Which, of course, is a daunting amount of human interaction and manual data compilation that was quickly abandoned.

Reconstruction, Tim Curtis

Enter an acquaintance who introduces me to the wonderful world of financial astrology, the art of finding correlations between geocentric astrological configurations and action in commodities, the economy, and individual securities. Financial astrologers have actually been at working for decades, some modern and reputable, and others of controversial stature.

The idea is fairly simple: people's behavior is correlated with (not affected by, in my view) astrological configurations, and when astral bodies (planets, not projections) do things, people tend to do things emotionally as well in the markets. This accounts for the rough behavior of broader markets like the meltdown in the major market indices that began to the day that Mars entered Capricorn and was signified by the clustering of 5 planets in Capricorn, the sign that rules businesses, structures, and money, among other things.

However, my methodology goes a level deeper. A person's birth chart claims to capture the essential "energy" of the stars at that moment, and that potential is what a person lives through. So why can't stocks have birth charts? Therefore, a stock's birth chart is set for the day that it first traded on an exchange, called a "first trade date" (FTD).

My strategy is conceptually simple: if I can compile astrological data for every notable day for every stock I can get a FTD for, I can examine the statistical correlation for certain astrological interactions.

An example is required. Warning: this is where it gets dense and fairly technical. Scroll to the next header if you prefer to skip it.

The fiddly, math-y, statistical-y bits

This heatmap displays the presence (# of times an aspect was present divided by # of days data was captured for--58,727 dates) of a given aspect from transiting Jupiter to the position of any given astral point--a catch-all term for planets, the Sun/Moon, asteroids, and even "hypothetical" points invented by 19th century German astrologers--for the birth charts of roughly 200 biotech stocks on dates at which either a significant top or bottom (peak or valley) was formed.

Darker squares mean that the aspect pairing was more frequent--for instance, transiting Jupiter making a sextile aspect to a stock's natal chart's Zeus point (shorthand: Jupiter sextile Zeus) occurred 62 times, versus a Jupiter conjunction Neptune aspect that only occurred 5 times on these dates, leading us to believe that Jupiter sextile Zeus is deeply significant.

However, there are a number of data flaws that are present with this highly selective viewing of the dataset, the most obvious being caused by Jupiter's slow orbit:

  • Since Jupiter moves very slowly (12 years to make a full orbit), any biotech stock younger than a maximum of 12 years hasn't had Jupiter make a conjunction with its Neptune placement yet.
  • Also because of Jupiter's slow movement, we have a relatively small sample count to draw any correlations from, versus other planets that move extremely quickly.

So we already have a relatively sparse dataset! Unfortunately, there are also a few conceptual bases of financial astrology and astrology in general that make data gathering extremely difficult in general:

  • The birth chart of a stock determines its entire potential--whether it lives and dies as a penny stock or stays on the stock market for 40 years.
  • Significant action in a stock's behavior is indicated by aspects from large/significant planets cooccurring with other "trigger" planets that kick off events which is difficult to gather statistical data on
  • Planets behave differently depending on the sign they are in and the planets they are aspecting, as well as the aspect they're making
  • Events' impact can be muted by conflicting planets which means that even the most significant of aspects will rarely have a "clean" signal, which contributes to the difficulty gathering statistical data
  • Particularly to stocks with relatively binary correlation to astrological configurations (up or down by a certain amount), multiple types of aspects or astrological events can have duplicate effects

With these factors combined, you end up realizing that it's next to impossible to gather a dataset that paints a clean picture, and unfortunately a lot of the data you might choose to gather is based on certain assumptions about how you expect it to work, which is not a valid way to perform statistical studies. However, you don't need a high resolution camera to perceive that a figure exists.

This heatmap shows the percentage difference of presence for specific aspects between the data I gathered for peak/valley dates for biotech stocks pictured in the previous heatmap, versus the presence for the same aspects on all days for biotech stocks.

In the image above this means that, for example, the Jupiter sesquisquare Neptune aspect is 700% more likely to appear on a peak or valley day than it is on any random day.

This really doesn't look like a significant percentage if you consider the total presence of this aspect--0.1% on peak valley dates--but this is significant in comparison to the roughly 0.015% of times the aspect was present across all dates.

And this, of course, is just one single example. I have many other representations of this data for examining directionality (peak or valley) of the aspect, overall bullishness or bearishness, magnitude, etc. across various types of datasets. But let's zoom out.

Emerging from the data

It's honestly difficult for even me, who's been studying this in detail, to take the example above at face value for what it is: a 7x significance on meaningful dates versus every date for a data point. It seems so insane, so improbable (that's a pun--it's literally 7x more probable), and so tiny.

But that's what we're working with when we're talking about the deeply symbolic, holistic, whole-is-greater-than-its-parts world of astrological correlation. I'm also only looking at one small piece of the puzzle, ignoring retrogrades, the signs of planets, eclipses, declination, and ancient astrological methods that are apparently frighteningly accurate for these sorts of mundane matters.

The real way to figure this out is definitely machine learning. I'm no machine learning expert (or even a novice), but it's obvious that this dataset still presents massive challenges of dimensionality. Any astrological point can be in one of 12 signs, 4 modalities, 360 degrees, making up to 10 aspects (depending on if you count the minor ones or not), against a practically infinite number of potential natal configurations.

This problem is immense. And I'm fascinated by it.

However, there are small hints here. All you need to succeed in the market is the tiniest of edges. Even 1% will do with proper risk management. I've come to the conclusion that it's essentially impossible to create an algorithm that will predict the action of any given stock, but it's perfectly reasonable to try and build a signal generator that can point to stocks based on these small percentages.

What I outlined above looks mostly at significant turning points in a stock via its peak and valley dates, but I've also gathered datasets based on sudden increases in volatility and overall day-to-day bullishness and bearishness. My next step is to turn these datasets into technical indicators that can be looked at in combination, and identify strong trading opportunities when they converge strongly.

sulidos on twitter

And out

I don't really know what the point of sharing all this is, besides for your interest, as I realized something recently: what I know is not what you know, and what you know is not what I know. Most of my thoughts are extremely old hat to me and I don't think they'd be interesting to other people because of the cognitive fallacy that since I know it so well, other people must know it so well, too. Is there a name for this?

But my point is: share your thoughts and yourself, even if you think it's boring, uninteresting, or "nothing new". You might make unexpected friends or teach someone something interesting--or realize who your friends really aren't, which is sometimes just as valuable.

Anyway, I was going to write more, but I've already written plenty. Here's a closing piece from article from above.

"On day 28, just before dusk, his girlfriend helped him out of the guesthouse, leading him in a blindfold to the garden. Winston was gently hysterical, cackling and clapping, kneeling to touch the ground under his feet, stroking the bark of a tree and muttering, “Who made this?”"

Who made this, indeed?

Other stuff

A message from my fiance

My fiance, Ras Alhague, wants to tell you to STAY HYDRATED.

In case you didn't know, they also create erotic glitch art on instagram as ras.alhagve.

Also, they are extremely attractive.

Spotify playlist

If you're not aware, I actually keep the tracks I put into Glitchet occasionally in a Spotify playlist, and it's here. Link is now permanently in the footer, too.

(I should have probably chosen a longer track for this issue.)

Minecraft Patreon server poll

I have a hypothetical question for you! I started playing Minecraft again with my lovely fiance, and was pondering whether my Patreon subscribers would enjoy having access to our private server MC (which may end up being joined with her Patreon subscribers at some point in the future, which would be neat).

Don't get too excited yet--I'm just gauging interest, as we do like having it for just ourselves ATM, but who knows :) If you're already a Patreon subscriber, or would subscribe ($1 minimum), or even just like the idea hypothetically, hit this big 'ol link dressed like a button:

Sure

(It goes nowhere. This just lets me use clicktracking to check the number of clicks without messing with polling BS.)

Get/gift/share a tarot reading

If you'd like a tarot reading, or know someone who needs one, you can now book me through my website at divination.wayspurrchen.com! The link has all the details, really. But if you like my style of writing, you'll probably like my style of divination.

Some more links

Sure, the Velociraptors Are Still On the Loose, But That’s No Reason Not to Reopen Jurassic Park - McSweeney’s Internet Tendency

“Trump is shrugging off warnings by scientists that the easing restrictions taking place across the country could cause tens of thousands of death...

OK

Good bye. Please forgive any typos or bizarre inconsistencies (bizarre consistencies are expected). I love you. Thank you for reading. See you next week.

Don't miss what's next. Subscribe to Glitchet:
This email brought to you by Buttondown, the easiest way to start and grow your newsletter.