As the summer comes to a close and we round the corner towards Q4, I wanted to share a few thoughts about what we’ve seen so far in 2023 and why it feels as if sentiment in the digital assets market is nearing a nadir.
At Starkiller we track a set of benchmark indices, some of which are proprietary. The purpose is both to simply understand what’s happening in the market, how different baskets of assets are behaving, and to benchmark our performance against very simplistic incarnations of our core strategy. In the charts below that come from our internal dashboard you’ll see five different lines on the graphs representing relative performance (in % change) over the course of several time horizons for Bitcoin, our Top 10 index, Top 50 index, Top Quintile index and Trend Overlay index.
The top 10 and 50 indices are a basket of equally-weighted assets (excluding stable coins and with the application of several liquidity filters).
The Top Quintile index is a basket of assets rebalanced weekly based on the top quintile of cross-sectional momentum in the liquid universe (if you’d like to read more about how that is constructed please refer to our paper here).
The Trend Overlay index is simply the Top Quintile index that goes to cash when the 5/50 exponential moving average crossover trend-following model for Bitcoin produces a sell signal and reinvests on a buy signal.
Again, these are very simplistic incarnations of momentum and trend following strategies, but each are, historically, robust alpha generators over long horizons.
The main observation I want to share today revolves around the incredible outperformance of Bitcoin relative to everything else. In a year where Bitcoin is up 58% through the end of August, the Top Quintile index is up 2%, the Top 10 is up 32%, and Top 50 is up 25%. It is incredibly rare in the history of our asset class that we’ve seen such a massive divergence between Bitcoin and all other assets in an up year. Our thesis is that this divergence largely explains the incredibly poor sentiment throughout the digital assets ecosystem . It has also made it nearly impossible for fund managers who take beta exposure to beat their Bitcoin benchmark.
Since mid-February, it has been an historically abysmal period for cross-sectional momentum, as evidenced in the chart above. Our research in this space leads us to expect our momentum basket to significantly underperform the majors (i.e. Bitcoin or Ethereum) in down markets (which is why we use trend following models to reduce or eliminate beta during those periods), but it is rare throughout the historical time series of data that it would underperform in an up market for Bitcoin and Ethereum, especially over such a large timeframe. In a few minutes we’ll get to why we believe that’s been the case and when it may turn.
Despite a strong headline return for Bitcoin, 2023 has largely been an “anti-trend” regime. Outside of January, which produced buy signals coming off a long period of (accurately) sitting in cash, Bitcoin has been largely mean reverting all year, unable to hold price levels. In fact, all of the returns since the end of January occurred in just a handful of days. This means that trend models have gotten significantly whipsawed as price has drifted down only to be followed by two or three days of event driven gains (e.g. the regional bank failures, Ripple v SEC lawsuit, or ETF filings). By design, intermediate term trend models don’t turn on a dime and thus have been largely unable to capture the very short periods of upside amongst the vast majority of down days.
In terms of sentiment, this is also likely a large driver of the despondency felt in the market. Seeing red on your screen the vast majority of days only to have 10 or so of them produce all of the gains this year is tough emotionally, especially when the vast majority of digital assets in the market are underperforming the majors.
To drive this point home, consider the returns of the naïve Trend Overlay index in 2022. The model vastly outperformed Bitcoin in a year where Bitcoin lost 64%. It achieved this performance even though it allocates to the the Top Quintile momentum model, which did much worse than Bitcoin in 2022.
While 2022 was terrible in terms of overall performance, trend models produced a ton of alpha. 2023 has been the opposite.
One quick aside before we get into the why this is all happening. The chart below shows the price of Bitcoin since the market peak in April of 2021.
While, technically, the market did make a new high that November, April was the real top for risk assets and many other indicators like exchange volume and asset flows.
The two pink lines represent the VWAP (volume weighted average price). The first is anchored to that top in April and the second to the beginning of 2023. I won’t get into a tutorial of why VWAPs are important here, but the basic read of the chart is that in April and July of this year Bitcoin could not push through the VWAP from the market top, but also buyers in 2023 have defended the yearly VWAP three different times now.
What this says is that there are still a significant number of medium to longer term holders of Bitcoin who may want to exit their positions at break-even (assuming they were dollar-cost averaging along the way). Conversely, those who stepped (back?) into the market in January have been willing to buy more at break-even prices in support of their position.
These will be extremely important levels to watch as we move forward into Q4. The year-to-date VWAP also lines up pretty strongly with being a line in the sand that likely corresponds to US treasury rates spiking above the range we’ve seen since October of last year. A breakout there likely produces a significant risk off environment and Bitcoin losing its buyers from the beginning of the year. Subsequently it may also produce a further banking crisis in the US which was positive for BTC in March.
So, what is producing this anti-trend, anti-momentum environment and when might it abate? We have some strong guesses.
First, liquidity has significantly dried up. Order books are much thinner than they were in the past few years as some market makers have exited the space. More so, many market makers are no longer willing to keep significant capital on offshore exchanges with dubious legal and operational issues. The SEC is also in the middle of suing the largest US exchange, basically saying that everything except Bitcoin is an unregulated security. They are also suing the largest offshore exchange (Binance) for a range of indiscretions. The Binance issue for liquidity providers and institutional traders is compounded by a litany of less than transparent (to put it lightly) behaviors that call into question how much risk investors are taking by leaving money on the exchange.
Liquidity is incredibly important to price action in digital assets, but that is no different than what we know about other asset classes. In small-cap equities, for example, there is a direct relationship between liquidity and multiples: liquidity and multiples rise and fall in unison.
With less liquidity in the long tail of assets, the desire for institutions to hold them decreases significantly as it becomes more difficult to manage risk.
Our next culprit is the outflow of money from the asset class. Below is a look at the total market cap of all USD pegged stablecoins. This metric has been in nearly continuous decline since early 2022, meaning assets are leaving the industry and being converted back into US dollars. The first chart shows the trajectory of that overall market cap, the second is a daily look at flows. Notice the significant regime shift in the spring of 2022.
More interesting is the divergence between two specific stablecoins, USDC and USDT. USDC is largely used on Coinbase and within the on-chain DeFi ecosystem, while USDT is primarily used on offshore exchanges. That green line is USDT, the blue line is USDC.
What you’re seeing here are US institutions exiting the trading of assets on Coinbase as regulatory issues compound. The SEC lawsuit against Coinbase can be clearly seen in the chart above.
We’ve also seen the total supply of algorithmic stablecoins contract significantly as the desire for leverage on-chain has diminished. One primary use of this leverage was to buy long tail tokens, which explains the significant decrease in risk taking and prices for our momentum basket.
The broader macro-economic environment has also been conducive to an anti-momentum regime in crypto assets. The US Dollar index is now up on the year after three different attempts to move lower. This is happening on the back of stubbornly high US treasury yields as economic growth has been far more robust than many expected. The closing of “recession trades” has produced a severe backlash in the dollar and put a lid on risk assets. If you can get 5%+ sitting in treasuries it's a bit harder to move out on the risk asset curve, especially as the Fed continues to very directly signal that they fear a return of too-high inflation. As with treasury yields, these levels in the dollar also correspond with that year-to-date VWAP in Bitcoin being defended. A significant move above here will likely produce a crack in that wall over the short term.
Another massive trend weighing on the long tail of digital assets is the significant amount of supply being brought to the market every day through the unlocking of tokens from early-stage investment rounds. Unlike traditional finance, where VCs typically have to wait for an exit event (e.g. acquisition or IPO) to unlock liquidity from their investments, the supply of digital tokens often has a more continuous unlock schedule. Projects launched in 2021 and 2022 are beginning to unlock for many VCs, who are selling for liquidity rather than remain invested. As an example, the figure below plots the token unlock schedule for $APE, the token launched by Yuga Labs.
While there are some VCs that have been able to raise additional funds over the last year or so, many haven’t and are now in the mode of returning capital to LPs as LPs search for liquidity. Without retail flow throwing market orders at coins all day, institutional participation waning, these unlocks are a steady wet blanket on any momentum. The data is very clear in that names which have outperformed over the last 30 days end up performing far more poorly than their peers over the following week as VCs take advantage of any positive moves to sell their unlocking tokens. We also see a significant divergence between two groups: those tokens without upcoming unlocks and those that do. Coins like MKR (Maker DAO) which have completely unlocked all their coins have performed significantly better than those in the middle of their schedules.
One trade that has been very profitable in 2023, for those who can access enough offshore liquidity, has been in shorting FDV-to-market-cap ratio coins with big unlocked schedules vs simply going long Bitcoin and Ethereum (FDV stands for fully-diluted value)
The last, and possibly most important, variable is simply the regulatory environment. The saying “don’t fight the Fed” is generally correct for the sole reason that markets are about vibes (sentiment) and markets have come to pay more and more attention to the Fed in terms of its signaling of when it wants market participants to be more/less risk taking.
Similarly, Gary Gensler (Chair of the SEC) has been crystal clear about what he wants in terms of a lower level of participation and risk taking in the digital assets space. There is a direct signal being given to the market, especially in the long tail of assets where the real legal issues are occurring. While the SEC is likely to lose (on the important points) their case against Coinbase for unregulated securities trading, and have already lost cases (on the important points) against Ripple and Greyscale, the fact that they choose to continue their strategy of throwing a wet blanket on the asset class impacts sentiment directly. Digital asset companies winning court cases likely won’t change that, but the SEC losing those cases may eventually cause a shift in their strategy. If they are admonished enough by the courts, it may potentially even lead to a shift in their tone. We believe the shift in tone will be more important than the cases themselves, as it will indicate a significant change in regulatory risk for investors going forward.
This asset class is about technologies that are attempting to rewrite large, very regulated swaths of the economy, in a more efficient, global nature over a long period of time. Vibes, for the lack of a better term, are incredibly important to the level of risk market participants are willing to take. It wouldn’t surprise me to see signaling from the SEC change at a similar time as the Fed.
So, what type of things do we want to see as evidence that we’re exiting this anti-momentum/trend regime? We’re looking at four buckets which are all somewhat causally related: macro, vibes, user adoption, and flows.
On the macro side, rates and the US dollar simply have to stop going up. Investors are scared, mostly because the ones that throw around the most money have massive PTSD from the late 1970s when we had three successively growing waves of inflation. Adding to that, we have the debt hawks who believe long term rates are going to surge because the US will have an issue financing its debt (ironically, this is also their bullish Bitcoin thesis, as the easiest course of action would be for the US to print more dollars).
As stated above, the vibes are going to be directly correlated to sentiment regarding Jay Powell and Gary Gensler, or a change of party in the executive branch. Institutions want to know that when they make investments – especially highly speculative, long-term investments – that the government isn’t going to pursue their investments, subjecting them to expensive legal battles. Congress passing legislation providing safe harbor to a range of crypto activities could also accomplish this, but in the short-term we’re more likely to see narrow precedence set in existing court cases.
In terms of user adoption, we are 100% confident that there will be another major cycle. While the asset class wasn’t dealing with the kind of regulatory assault that it did in previous crypto winters, we have seen the same pattern play out in the data three times now. Similarly, in those previous cycles investors have questioned whether the next wave of user adoption would ever take place and if crypto was dead. This time is no different.
We like to use the corollary of fiber optic cable laid during the tech bubble of the late 1990s. It was obvious that user adoption and quality businesses on the internet were going to happen eventually, but it required enough speed and capacity for entrepreneurs to build real things people wanted. In the last 3-4 years our industry has invested heavily in R&D around speed and capacity of blockspace, along with things like ZK proofs and other technologies that will enable the next round of builders to produce real applications and use cases that drive users on chain. Yes, like the tech bubble that laid the fiber we also had a bubble in L1s, copy pasta lending markets and basically anything associated with making crypto into a casino. But in the end that R&D, and the casino itself which is the boot program for liquidity in the system, will produce a scalable ecosystem that is able to handle the blockspace demand in the next cycle for real use cases. It may be RWAs (real world assets) which ignite the user adoption, it might be social media type stuff, or it could be some use case none of us is thinking of (who called Uber being a huge winner from mobile phones years in advance?). The ability to quickly onramp retail liquidity was also built in the 2020 cycle, and should provide a path for capital to flow even more quickly into the space in the next cycle, assuming regulations don’t staunch those channels.
And when those things take place, we will see flows return. It will coincide with the launch of new tokens, hopefully with less predatory FDV-to-market-cap ratios. Institutions will return to trading in the US and be willing to stomach the risk of holding liquid tokens representing the business itself instead of some nebulous “governance” asset.
Some of these are going to take some time, others we may be days or weeks away from seeing play out. As stated above, this is the longest period of underperformance from that top quintile of cross-sectional momentum in our dataset. We would be extremely surprised to see that behavior persist in an up market for much longer, at least in the selection of coins without constant supply being brought to market by unlocking VCs.