Signal vs. Noise, Market Efficiency, & Evolving Alpha
Thoughts on parsing signal and noise in investing and the evolution of Alpha
As markets evolve, so do the areas in which investors seek alpha. And as the world evolves, it gets harder and harder to understand where and how one should focus in order to generate excess returns.
If one were to look at a variety of volatility-inducing events over the past few years you would think that markets have been caught off guard with minimal awareness. This has led to people overusing the term Black Swan Event as a substitution for “didn’t appreciate impact”.
Japan is a recent example of this where at some point over the weekend of August 2nd the market began to collapse with a -20% ETH candle and we woke up on a Monday morning to a 2x+ spike in the VIX + VX (volatility index + futures) and a broad sell off and panic.
The thing is, if you were talking to anyone who had their eyes on the markets the prior few weeks to this they could have told you that the Japanese economy was having problems and specifically that this interest rate move had very negative implications across T-bill demand all the way to a variety of carry trades.1
People just did not digest this information ahead of the move.
Reasoning Through Next-Order Effects
Humans are bad at having high conviction in next-order effects and reasoning through uncertainty. It’s easy(ish) to have a hypothesis, it’s hard to build a hypothesis on a hypothesis, etc.
The market may have known that the Yen would move strongly but didn’t think about just how risk-on people were on the “risk-free” trade of levering against it nor on the assets most impacted by a carry trade of that size.2
I’ve written about the idea of next-order effects often including in a recent post on technology forecasting, but it can be done in market forecasting as well.
The difficulty of reasoning through multiple events that stack levels of uncertainty is not intelligence necessarily but instead is a person’s ability to parse signal from noise in a growing environment of noise.
Noise creates errors and error rates compound at each step of abstraction.
There are many factors that can impact markets and these must be approached with:
IQ - what are the various permutations of what could happen
EQ - how will humans/machines made by humans interpret these permutations
Creativity - how well can one explore tail-risk permutations, and unrelated, how well can one express the resulting view of finding signal in noise
Without this, parsing signal from noise is very difficult.
Some may argue that the Japan event came at a time in which the conviction across most market participants has been quite low and noise to signal ratio has been quite high.
In 2024 we are inundated with election volatility, geopolitical instability, macroeconomic uncertainty, and technological change with people wondering if AI will create a utopia or destroy the world.
This dynamic feels like a moment in time but as time goes on, I increasingly believe it to be a more permanent societal shift where volatility and uncertainty are constants.
The world is not getting simpler, the patterns are not nearly as easy to match to prior times in history, and likely across all investing the best skillset to hone is how to minimize noise so that you have some chance of parsing signal.
Cliff Asness wrote a great paper on this point recently that supports how markets are less efficient than they’ve ever been as information continues to proliferate and scale.
I highly recommend reading the whole paper but this excerpt gets at each of the IQ/EQ/Creativity points and how technology disrupts this instead of enables it:
One could simplistically frame this shift as: Humans are more informed than ever before which actually means they are more confused and less efficient than ever before. While quant strategies perhaps have attempted to solve this by at least processing the IQ part of the equation well, we as humans have not evolved to process all three factors.
The Left Curve Gets Loud
People underestimate the impact of market structure shifts from a behavioral/demographic perspective. The largest and best understood market structure shift over the past few years has been that of retail resilience and risk-taking.
The Gamestop generation, memecoins, get rich or die tryin’, etc. are the best ways to think about this and it has fundamentally changed how market participants underwrite risk and model markets.
The thing is, historically these shifts have been temporary with the market “normalizing” and reaching “efficiency” far faster due to a more narrow set of outlets for large groups of people and dollars to congregate and thus find heuristics to create schelling points around.
The internet is a routing machine, allowing “niche” crowds to find their echo chambers and form resilient mindsets (a euphemism at times for absurd/stubborn/abnormal) in ways that were previously not possible. This may result in “resilient” mindsets that fly in the face of “traditional” viewpoints being formed by large groups more and more over time as these “outsiders” were previously less able to find like-minded individuals. This shift is a potential reason why this time, these market structures will not revert and is of course, often discussed through the lens of populism and politics throughout the world.
Framed another way, maybe the B(d)oomers/mid-curve investors could tamper/revert volatile viewpoints and conviction of the Zoomers/left-curve prior to the internet but now people can find their tribes and create large scale echo chambers that last nearly forever.
A great explanation for Keynes’ “Markets can remain irrational longer than you can remain solvent”.
BTFD
During early covid I was trapped inside and bored and thus enjoyed overtrading public markets as the world collapsed into a pandemic. I wrote about the concept of the Buy The Fucking Dip recession and how the vast majority of conversations within tech as well as amongst a moderately younger generation was built around continuing to hit the proverbial bid, buying up assets as prices fell, and hoping for larger dips to buy.
These people were handsomely rewarded if they moved moderately early enough, with huge government stimulus, a V-shaped recovery from March 2020 lows, and a massive rally across all risk assets for 12+ months. As markets felt toppy, they continued to assume past results meant future results and did not sell many assets, diamond handing across a broader collapse in risk assets as well as specific types of assets (tech + crypto) in the name of multi-decade trends.
Core assets like BTC/ETH/SOL and big tech have somewhat rebounded over the longer timeframe, while the majority of long-tail assets like NFTs, altcoins, and a variety of high-growth/emerging tech companies have nuked 80%+ with little chance of ever reaching close to all-time highs.
These market behaviors that were formed over the last bull cycle and in the first generation of many’s careers will likely persist as they continue to believe in heuristics and approaches that solidified the fortune of people around them and perhaps solidified their career positioning from this era.
Humans may continue to buy the dip across all sorts of seemingly absurd heuristics for long periods of time.
Alpha evolving from buying to selling
When any heuristic becomes consensus you see heavily correlated bubbles pop up that move even the top decile of participants to Market Beta, pushing perhaps only the top ~1% to Alpha Generators. This Market Beta has proven to be so strong and sizable that there is an open question about the amount of Alpha really left in risk assets like equities and later-stage venture investing.
The overly discussed view of venture firms becoming effective index funds and squeezing the middle of the asset class is a version of this in private markets and is often compared to the bifurcation of firms in investment banking.
The well-covered dominance of the Magnificent 7 stocks3 is exemplary of this in public markets, effectively steamrolling and outperforming the vast majority of those trying to generate Alpha on the long side over the past few years.4
Alpha is largely framed as an exercise of price entry (when do you buy an asset) because investing is a no called strikes business (among other reasons) so the buy is the key decision. However as assets go through tighter and more aggressive cycles and bubbles, and as markets have infinite bids and a certain gravity to them due to passive flows, one has to ask if the majority of Alpha is generated on the ability to exit.
I know this is almost an absurd thing to say but put another way: Is it possible that no longer is the vast majority of Alpha is accumulated from buying and owning outperforming and non-consensus assets, but instead from understanding when to sell mid-tail assets in order to harvest the cyclicality and gravity of markets and to utilize the high-consensus, low independent conviction market structure to the best of your advantage?
An interesting hat tip to this is that the Mag-7 trade was the most crowded trade in hedge funds for ~15 months straight and only recently has started to see some divergence as funds rotated out amongst tech weakness in late summer. In addition, the increasing prevalence of the 150/50 (a shift from 130/30) long/short fund structure is evidence of this as funds use leverage and market beta and pair it with alpha to outperform indices with theoretically less volatility.
George Soros is famous for saying “When I see a bubble forming I rush in to buy, adding fuel to the fire.”5
Perhaps the counter of this is to know when to be one of the first out the door as cycles and sentiment turn and the BTFD investors start diamond-handing before capitulating.
Panic → Euphoria & Persistent Noise
The complexity of market dynamics continues to outpace our ability to model and predict them. As information proliferates and participant structures evolve, the challenge isn't just about having more data, it's about developing new processes to parse signal from noise in an increasingly chaotic system.
Within this though, the role of an investor is still perhaps to see what others don't, understand what others can't, and act when others won't.
Hilariously, I wrote the majority of this post during the pre-market hours ahead of Monday August 5th as it looked like markets were set to open sharply down.
At open, markets basically V-shape recovered due to US economic data and upon reflection we saw a heavy bid by institutions and a large amount of public retail selling.
Since then, markets traded ~sideways as they basically stared at NVDA, up until they sold off due to DoJ actions in early September, leading to a mini cascade of market sell-offs.
Of course, the twitter timeline was filled with panic for a few hours and apathetic for weeks until Powell cut rates and China signaled turning on the money printer.
I’m sure some people bought the dip, and alas, euphoria has returned.
Thanks to Smac for thoughts on this post.
They maybe could tell you some vague musings around governance changes for Japanese companies as well but i don’t think anyone knew how this would actually impact prices
While this impacted both tradfi and crypto participants, it is especially ironic for crypto considering that market was upended by a different carry trade courtesy of the moon boys at 3 Arrows Capital
I really don’t like that name.
Indexes are also steamrolling probably 90%+ of professional public market investors.
For what it’s worth, I don’t think Mag-7 is a big bubble and am very long META and quite long AAPL and GOOGL to a lesser extent.