Trading Salience in a Hyperconnected Marketplace (Part 3)

And we’re back. This week has been legitimately crazy with the Gamestop saga. I’ve taken the self-imposed position of playing financial reporter through the whole charade, including being one of the first to break the news that Robinhood (and, as we found out later, other brokers) was de facto manipulating the price of Gamestop and other meme stocks by refusing to allow new opening of positions (and later, refusing even the purchase of shares over a cap).

That said, a brief recap of Parts 1 and 2 may be in order. In Part 1, I explained how largely due to the structure of its narrative, the rise of the Gamestonk was inevitable, owing largely to its salience and conformance with cultural hearthstones (the monomyth). I briefly touched on memetics, or why certain memes (in this case, Gamestop) prosper, and why others (This highlighted a few key aspects of “Why Gamestop?”):

1. A catalyst (the short squeeze thesis, console sales)
2. Salience (memorability and recall in the minds of low-information investors)
3. Humor

Similarly, in Part 2 we took a step back to explain the concepts of the theme (a well-defined topic, for instance pandemic/stay-at-home investing), the narrative (a story about the theme), and a trend (the propagation of the theme over time). Finally, I discussed my hypothesis on the impact of the hyper-connected state (information real-time super-hubs like WallStreetBets and its daily threads) on the reflexivity of the markets.

Let’s see now how these two hypotheses interact.

Surfing the Mind Map

As briefly touched on in Part 2, one of the most unique aspects of the Gamestop drama isn’t Gamestop itself, but the contagion it spread to other equities on its rapid ascent.

The most amusing by far have to be BLIAQ (Blockbuster Liquidating Inc.) and SHLDQ (Sears Holdings Corp).

For the less savvy here, Q stocks are traded over-the-counter only, and indicate the company is in bankruptcy proceedings. This almost always leads to a total wipeout of shareholder value, as evidenced by the extremely thin trading and near-zero value of both shares. Until well, this past week. Despite the fact that both companies are literally bankrupt and have effectively zero shareholder value, they rallied this week, handsomely rewarding risk-loving speculators.

So, the most obvious followup question here is why did they rally? We saw the rise of the meme stocks this week (AMC, BB, NAKD, GOGO, etc.) to the point where talking heads argued it presents a systemic risk to the equities markets (I disagree, but that’s another blog post in itself). However, even in the normal storybook stocks (NIO, PLTR, TSLA) there exists at least the illusion of massive cash flows in the indeterminate future for the patient and lucky investor. In the case of Q shares, with few exceptions, the end of the road leads asymptotically to zero. Q stocks, therefore, represent a pure play in speculation itself — much like a game of hot potato, Q equity is worth exactly the chance of selling it to someone for more money (the greater fool), and nothing more.

That said, how did these connections happen in the first place? We can easily construct a scenario where eager speculators, seeing the rise of Gamestop, used multi-level logic (the Keynesian beauty contest I discussed in previous posts) to reason what other market participants value. Seeing this as a way to cash in on the hype, they may have preemptively bought shares in literal nothing and triggered the price momentum and buying frenzy.

This is completely reasonable, and a large driver of reflexivity in the market places. People buy assets, especially in the current There-is-No-Alternative macro environment, in large part due to an expectation they will be worth more to others in the future.

However, this still doesn’t answer the fundamental question here — Why Blockbuster and Sears in the first place?

Let’s delve into a bit of faux psychology here (with my degree from Armchair University).

As humans, we intuitively develop semantic associations between concepts. This is a well understood concept (semantic memory), and tends to be best modeled as a semantic network, a form of graph where concepts are modeled as nodes and the relationships between them (semantic relations) form the edges.

Source: https://en.wikipedia.org/wiki/Semantic_network#/media/File:Semantic_Net.svg

This intuitively makes sense, I’m sure. The way we structure language is akin to a semantic network — when you speak about well, anything at all, you’re expressing an association between concepts (the subject and the object, to go back to grade school grammar).

It turns out this handy concept makes a lot of sense when you apply it to meme stocks. Much like the human mind (which makes sense, given the concept of meme stocks is a very human phenomenon), meme stocks tend to couple along semantic associations. This is the mind map erstwhile project managers have nightmares about.

Spooky.

I tend to envision the relationship as the following:

1. There is a god meme (borrowed terminology from https://twitter.com/goodalexander — a fantastic mind and follow)/L0

A god meme, in this paradigm, is the epicenter of the narrative, which usually follows (or sometimes sets) a theme and starts the trend. Two fantastic and simple examples of the god meme are Tesla and Gamestop. In Tesla’s case, while cultural shifts towards ESG have been an undercurrent for years, Tesla’s rise as the reigning meme stock of 2020 dramatically accelerated the zeitgeist. Tesla effectively was the nuclear bomb which in its blast radius (success) led to cash inflows (and therefore price increases) throughout its semantic network, to varying degrees:

a. The ARKosphere (L1)—Cathie Wood, in retrospect, may or may not be known as one of the most successful portfolio managers of all time (we’ll see in the coming years). Despite this, it is extremely clear she understands the zeitgeist. Owing strongly to the Tesla effect (and her prescient predictions of a $4,000/share stock price), ARK Invest received insane cash inflows throughout 2020. More interestingly though, because of the newest zeitgeist (the Gamestop effect), ARK Invest has, at least temporarily, been overshadowed. This points strongly to the semantic network effect — investors chased ARK throughout 2020 as part of the trend and its strong association to the theme (ESG). When a new god meme appears, the inflows move with it.

b. EV (L1) — This is a pretty obvious one. Hoping to replicate Tesla’s astronomical success, EV companies (Nikola, Fisker, Hyliion, Kandi, Nio, Xpeng) rallied, often completely detached from reality. Massive cash inflows came, largely tied to the Tesla effect. This can be neatly observed looking at one catalyst in Tesla’s momentum — November 16th and Tesla’s inclusion to the S&P 500.

Xpeng’s stock price as of Nov 16.

Nio’s stock price as of November 16th.

Tesla’s stock price as of November 16th.

What’s interesting to note here is the follow-the-leader effect: once the news of Tesla’s S&P inclusion hit, all EVs rallied (to varying degrees), but in general it took multiple days after the initial announcement (November 16th). This provides ample opportunity for speculation.

c. Electric battery technology and charging stations (L2) —As Tesla’s blast radius spread, the cash inflows hopping onto ESG spread with it. At a certain valuation for Tesla (maybe we hit the top, who knows), most market participants eventually will stop speculating on future price increase. This seems partly driven by fundamentals, but mostly driven by risk aversion. When that occurs, cash inflows (much like the L1 group) will radiate outwards, spreading at a rate over time and space to farther semantic associations. This was neatly demonstrated by the battery and charging station booms.

QuantumScape (NYSE: QS) was a dramatic example of this effect. A SPAC merger (hence the implication is the company itself valued its equity at ~$10/share), the deal was announced to the public long before the late November rally. All information on it was public, including the technology aspect. But then, this happened:

There’s our parabola.

What happened? Quite simply, this coincidences with a linear perspective of semantic association. On November 16th, the S&P 500 announced its inclusion of Tesla to the index, driving the price at peak up >100% from an already inflated price. Cash gets nervous at high valuations due to risk aversion, and starts radiating out to farther semantic associations: Tesla -> electric vehicles in general -> the batteries needed to power electric vehicles. And then QS happens.

d. Lithium mining (L3) — However, we can actually see that because of Tesla’s long reign as epicenter/the god meme, cash inflows radiated out even further through the semantic network. Much like QS’s late November/early December rally, the actual providers of raw material to the battery companies (lithium) also began to rally (partly also driven by hopes of a blue wave and renewed focus on electric vehicles under President Biden):

We can neatly observe LIT (best ticker name) rallied nearly continuously throughout 2020, but hiked up massively at the turn of the new year, following almost sequentially the battery boom.

—

This is the mind map, and makes sense in the context of how humans think. We can construct a similar semantic network to model the Gamestop effect (the vanguard of the trend and the current god meme). Most interestingly, this provides us with testable predictions, which are nice versus philosophical ramblings. To start, we can mine the narrative (the Gamestop “story” arc) and the theme (short squeeze, vintage) to construct associations:

1. The narrative — The major narrative characters of interest here are Melvin Capital (the “villain”), Point72 and Citadel (other “villains” per the evolving story), WallStreetBets, Ryan Cohen (Chewy genius billionaire), DeepFuckingValue, and Michael Burry. To keep it simple here, let’s focus on what mining the narrative can tell us for stock trend prediction:

• Michael Burry — One of the original heroes of the story (and full time Trump apologist on Twitter), Michael Burry bought into Gamestop early on through his fund, Scion Asset Management. Thankfully, we can easily view the 13F of Scion here. While there are a lot of safe, unexcitable “boomer” names (Altria, AllState) in the list, I immediately fixated when reading it on a few names: DBI (Designer Brands) and FL (Foot Locker Inc). This is because those names are salient given the god meme (similarly out-of-fashion retailers on a long term stock price downtrend, especially DBI). Based on this, I made the call to go long on DBI (given its lower price point and weaker recovery than FL). It already paid off, going from $11.xx to a peak of $13.33 on Friday.
• Ryan Cohen — The mentor of the monomyth himself. His most famous association pre-Gamestop is with the company he founded, Chewy. Let’s take a look what Chewy did during the Gamestop period:

Good boy!

It’s hilarious that a 50% price increase in one day looks so small here.

2. The theme — More interestingly, the god meme set the new trend in January 2021 — using the gamma hammer to print money from highly shorted, crowded trades. In this case, Melvin Capital is basically being whack-a-moled for money by people who don’t understand what delta is.

In this case, we can decompose three separate themes here — vintage, consumer-facing stocks (given Gamestop was last cool in about 2005), heavily shorted stocks (the short squeeze thesis), and existing meme stocks (PLTR, JMIA, OSTK). In general, we can observe the more salient names (AMC, BB) tended to outperform in the theme other representative names (LGND, RKT for example). The closer to the consumer mind, whether through clever ticker naming (FIZZ), nostalgia (BB, TR, BLIAQ), or episodic memory (AMC — going to movie theaters, remember those?) the more squeeze effect from the Gamestop blast radius we can observe.

Rambling and Speculations

The good news about most of what I’ve covered thus far is, even with knowledge gaps, most of this is indisputable given what we’ve all seen. I’m going to present a list of hypotheses I hope to test going forward about what I jokingly call “quantitative memetics”:

1. Cash inflows radiate from the god meme/L0 to semantically associated stocks in a predictable proportion and sequence — As we could see neatly through the Tesla effect, the strongest associations (ARK, other EVs) tended to rally to perhaps similar degrees (in fact arguably NIO returned more than Tesla did to investors in 2020) and at a similar time as Tesla. What’s more interesting is farther out connections (L2, L3, etc). These tend to enjoy (at least observationally) a lag period to the god meme, allowing speculators a chance to profit (or profit again, if they missed the god meme itself). Similarly, they tend to rally but to a smaller degree, the farther the association.
2. Associations can be made through time and space — There is no trivial way to map for instance, BLIAQ and GME. At least to a computer that is. In fact, it’s quite likely no post mentioned them together (pre-rally) in the same sentence online. This is a uniquely human phenomenon — we are designed to see patterns, even when the pattern may not exist. A savvy speculator can connect the dots here in ways that perhaps a computer can’t (at least not yet, to my knowledge).
3. Hyper-connectivity begets reflexivity which defines the speed and impact of the trend — Or, in English: the more meaningful at a personal level the cause is, the more rally (and farther spreading) the rally should be. We can neatly observe this with Gamestop, which morphed from the speculative hobby of a few tech nerds and value investors, to a casus belli against the financial elite. This gives it staying power. Similarly, Tesla and the promise of a better future allows investors to blind themselves to fairly massive red flags (especially on valuation and Twitter decorum of CEOs). The trend stays. Lesser salience implies a shorter, weaker trend.
4. Multiple themes can coexist in a trend, but one usually shows dominance — People, especially nowadays, tend to have short attention spans, and this is reflected in the market. While multiple trends can exist simultaneously, provoking rallies in disparate sectors (e.g. the stay-at-home/lockdown reopening twin narratives discussed in Part 2), in general the more impactful theme will ‘win’ (meaning a bigger, stronger rally).
5. Trends can ebb and flow in response to catalysts in the god meme, but the farther the association the less impacted — A clever investor can see that even in the broader trend (Tesla as the vanguard of ESG in 2020), there were substantial ebbs and flows. In general, these tended to coincide with catalysts in the god meme (S&P Inclusion, the stock split, Tesla Battery Day). Interestingly, these ebbs and flows are less reflected in L1 associations, and even less as you radiate outwards.
6. In general the god meme will be the most salient, simple narrative-We can see, in large part due to the interactions of low-information investors in hyper-connectivity (and the pile-on effect of momentum quantitative strategies), the god meme will reflect the most basal representative of the zeitgeist. Tesla is a car (and possibly energy) company, but due to the mad scientist founder Elon Musk also associates strongly with a bright future for humanity’s future. Gamestop is a video game store, but it is also a fight against the ruling class. These narratives win precisely because of how raw they are to us. We are an optimistic species, and we will turn on our blinders to reality for a rosy, simple picture.

Conclusion

I want to heavily thank contributors to these theories, including as I mentioned, Alexander Good (Twitter: https://twitter.com/goodalexander). Throughout my trading career I’ve also had the pleasure of finding people who just get it — for a lot of us, what I’m stating here is just a formalism of a more intuitive understanding of the markets (this is how I suspect my friend Patrick, also known as @final_october on reddit and Twitter, makes the calls he does).

I’d love to hear thoughts on this, and I’m looking forward to seeing this play out in real time (and iterating on these hypotheses).

Thank you for reading as always.

• Lily