Read on Twitter
Here we are - the last day of sunshine before the Replicant Wave™ crests.
Alright, let me unpack this absurd little idea…
1/
Alright, let me unpack this absurd little idea…
1/
So as you may know I’ve been on the hunt of a curious little pattern that I’ve mentioned in passing a couple of times.
This thing:
2/ https://twitter.com/coloradotravis/status/1358151132841865217?s=20
This thing:
2/ https://twitter.com/coloradotravis/status/1358151132841865217?s=20
There are many ways to look at it, and I’ve tried looking through many lenses so far.
Some from finance and some from other disciplines.
3/ https://twitter.com/coloradotravis/status/1358903723938861057?s=20
Some from finance and some from other disciplines.
3/ https://twitter.com/coloradotravis/status/1358903723938861057?s=20
And the area where it’s most pronounced has been the major indices and their corresponding volatility instruments.
There are some interesting replicant structures in play.
4/ https://twitter.com/coloradotravis/status/1359738674972266502?s=20
There are some interesting replicant structures in play.
4/ https://twitter.com/coloradotravis/status/1359738674972266502?s=20
And I wasn’t sure how to line up the timelines precisely or why this would be happening, but in reading the work of others a few things started to come together
Then a few things other people were saying started to come together.
5/ https://twitter.com/Ksidiii/status/1361145401282355200?s=20
Then a few things other people were saying started to come together.
5/ https://twitter.com/Ksidiii/status/1361145401282355200?s=20
And I became convinced that this ties together a bunch of overlapping themes.
One is @profplum99 and his work on passive, which has a market thinning effect and also a “dehumanizing” effect on the market as algos take over.
6/ https://twitter.com/coloradotravis/status/1359360009465856003?s=20
One is @profplum99 and his work on passive, which has a market thinning effect and also a “dehumanizing” effect on the market as algos take over.
6/ https://twitter.com/coloradotravis/status/1359360009465856003?s=20
So the thesis here is: “algos trading volatility are the main participants as regards price movements.”
The market is more reflective of the aggregate ‘intent’ of models than it is human intent.
Which creates curious structure…
7/
The market is more reflective of the aggregate ‘intent’ of models than it is human intent.
Which creates curious structure…
7/
Algos use price history as a dominant input, and they also use forward-modeled volatility returns.
The former input will tend toward recurring patterns (same input, same output) and the latter will center these patterns around calendar-events (like week length and opex).
8/
The former input will tend toward recurring patterns (same input, same output) and the latter will center these patterns around calendar-events (like week length and opex).
8/
So what we want to look for is a “terrain-adjusted replicating wave” where “terrain” is the impact of calendar events that are highly relevant to volatility.
So the wave does a “module” of its overall form, then terrain triggers the next “module” of the broader wave.
9/
So the wave does a “module” of its overall form, then terrain triggers the next “module” of the broader wave.
9/
This results in a big, slightly distorted (by terrain) replicating waveform as these black box algos mechanically optimize for returns sans much human intervention
Given the same (or similar) inputs, they generate the same (or similar) set of outputs.
10/ https://twitter.com/farrisbaba/status/1360734331992432641?s=20
Given the same (or similar) inputs, they generate the same (or similar) set of outputs.
10/ https://twitter.com/farrisbaba/status/1360734331992432641?s=20
We’ve just hit some funny terrain - the same terrain as last year, which of course our meat minds reject as absurd.
That might matter if humans were running the market - expectation might preclude the event.
But perhaps humans aren’t, so their expectations don’t matter.
11/
That might matter if humans were running the market - expectation might preclude the event.
But perhaps humans aren’t, so their expectations don’t matter.
11/
So far, we're seeing these replica structures playing out pretty much as we'd expect, at least as regards the major indices and volatility.
12/
12/
The current terrain is:
- Structurally similar starting conditions in SPX, VIX, VVIX
- 4 day week before opex
- Echoes of similar credit market dynamics
- Completion of bigger waveform
13/
- Structurally similar starting conditions in SPX, VIX, VVIX
- 4 day week before opex
- Echoes of similar credit market dynamics
- Completion of bigger waveform
13/
And if that’s the case... why would the models spit out a different result?
Now - perhaps this is all just coincidence.
Or perhaps it really is the machines and we don't matter anymore.
14/
Now - perhaps this is all just coincidence.
Or perhaps it really is the machines and we don't matter anymore.
14/
Captain's log: Stardate 83474.43
The replica set is ramping into the close as we'd predicted.
Left: Last time, hourly.
Right: This time, hourly.
Red/Green bars for framing.
Let's see if futures fall off...
The replica set is ramping into the close as we'd predicted.
Left: Last time, hourly.
Right: This time, hourly.
Red/Green bars for framing.
Let's see if futures fall off...
