How do crypto quantitative fund managers generate alpha?
TechFlow Selected TechFlow Selected
How do crypto quantitative fund managers generate alpha?
The value of the crypto market is equivalent to attention.
Navigating Web3 tides with focused insightsTopic: How Do Crypto Quant Fund Managers Generate Alpha?
Host
Zheng Naiqian @ZnQ_626
-
Founder of LUCIDA
-
Champion, Mixed Strategy Category – Bgain Digital Asset Trading League Season 1 (2019);
-
Composite Strategy Category – April Runner-up, May Champion, Season Bronze Medalist at TokenInsight Global Asset Quantitative Competition (2020);
-
Composite Strategy Category – Season Bronze Medalist at TokenInsight x KuCoin Global Asset Quantitative Competition (2021);
Guests
Ruiqi @ShadowLabsorg
-
Founder of ShadowLabs & Investment Director at DC Capital
-
Quant products managing over $300 million in AUM
-
Market-making consultant for multiple exchanges and well-known projects
Wizwu @wuxiaodong10
-
Multi-factor & Discretionary Strategy Fund Manager at RIVENDELL CAPITAL
-
Background in computer science + finance
-
$20M in non-traditional crypto strategies
-
Specializes in on-chain/off-chain data mining and market-neutral multi-factor strategies
What Is the Framework Behind a Fund Manager’s Alpha Strategy?
Zheng Naiqian @LUCIDA:
LUCIDA is a multi-strategy hedge fund. We develop various low-correlation diversified strategies to ensure our performance can endure both bull and bear markets.
Taking proprietary capital as an example, our return target is to outperform Bitcoin spot returns during bull markets. So first, we perform macro timing—assessing whether the market is near a bear bottom or a bull peak. This assessment is very low-frequency, roughly on an annual basis.
If we believe the market is at a bear bottom, we convert all funds into full exposure to Bitcoin and hold it through the entire bull cycle. On top of this, we enhance returns using quant strategies such as CTA, multi-factor models, and statistical arbitrage. These strategies are the core source of alpha during bull runs. We dynamically adjust allocations among these strategies based on current market conditions to maximize capital efficiency.
When we believe the market has reached the top of a bull phase, we exit all Bitcoin positions and switch to USD to ride out the bear market. During the bear phase, we continue generating returns using CTA and options volatility arbitrage strategies, growing our USD holdings until the next market cycle begins.
Therefore, alpha contributions fall into two main categories: one, macro market timing, which is one of our core competitive advantages; and two, return enhancement from quant strategies. For instance, if Bitcoin rises from $10K to $50K, perfectly buying at $10K and selling at $50K isn’t realistic. That's where quant strategies come in—to boost returns so we consistently outperform Bitcoin’s spot appreciation.
Wizwu:
When discussing alpha strategies, it ties back to our fund’s capital nature. We manage native crypto capital—all in tokens—so we’re inherently forced to generate alpha passively. Essentially, it’s an index-enhanced strategy. Within this framework, we employ multi-factor models and some discretionary approaches.
As an institution engaging in discretionary trading, we must consider many factors—holding periods, liquidity of small-cap tokens, etc.—which limits our investable universe. Holding too broadly makes it hard to beat the market; holding too narrowly means competing directly with project teams and VCs for allocations. So our approach is to do a bit of everything.
For example, when we identify a new factor, different people interpret it differently—some neutrally, some subjectively, some quantitatively. These represent distinct trading mindsets. That’s why I group subjective and multi-factor strategies together. The crypto market lacks precedent, so we borrow concepts from equities (data-driven factors), value investing (though we haven't found strong equivalents yet), and even commodity futures (like inventory and supply-demand analysis). Ultimately, success depends heavily on our ability to interpret data and trading signals.
Unlike native Web3 venture funds, we don’t have their level of research resources or broad ecosystem visibility. Instead, we focus on agility and data-driven decision-making. In this market, different players earn different types of returns—similar to futures markets. Industrial players earn industrial profits, quants earn quant profits, discretionary traders earn discretionary profits. Different methodologies yield different outcomes.
Overall, we operate primarily on a token-denominated basis. Our goal is to achieve Sharpe ratios of 3–4, with annualized returns exceeding 10%. Macro timing is minimal or extremely low-frequency. Based on market insights, we derive factors applicable across various strategies—including discretionary and multi-factor systems.
In factor discovery, we often adapt factors from traditional futures or equity markets for testing, while also incorporating our own trading experience.
Ruiqi:
We are a fully quantitative and automated team. From the outset, our alpha framework was built on principles of high engineering rigor and full automation, relying heavily on data-driven processes and execution. Internally, we divide our alpha framework into execution alpha and predictive alpha.
Crypto exchanges are highly fragmented, and investment tools vary widely. For example, to gain risk exposure, I could trade futures or spot, across different exchanges. At the execution layer, we compare funding costs across markets—futures vs. spot prices, basis, fees, slippage, borrowing costs, etc. After comprehensive cost analysis, we select the cheapest available instrument. Through this optimization, we generate approximately 5%–20% annualized returns, which we classify as execution alpha.
The second component is predictive alpha—forecasts across different timeframes, frequencies, and assets, including time-series and cross-sectional predictions. We use these forecasts to adjust risk exposure across instruments.
However, there’s a unique overlap between predictive and execution alpha. For example, suppose I make a directional forecast that solves only 20% of the problem—the remaining 80% lies in whether I can actually execute it effectively. This includes order placement tactics, fill probability modeling, conditional probabilities of funding costs, etc. These elements blend execution and prediction. Broadly speaking, this integrated system enables our alpha breakthroughs.
When conducting performance attribution, the contribution of each alpha type varies. As mentioned, execution alpha typically aims to beat the benchmark by 5%–20%, making it relatively stable but capped in upside. Predictive alpha differs—for high-frequency predictions, individual profits are razor-thin and deeply intertwined with execution. However, for medium-to-low frequency forecasts, predictive alpha tends to dominate overall returns.
What Is Your View of the Crypto Market? What Kind of Market Is It?
Wizwu:
Earlier I mentioned that different markets allow you to earn different types of returns—logical analysis in futures, and similarly in crypto. The defining feature of the crypto market is its high volatility. For example, U-denominated yields from funding rates alone can reach ~20% annualized during bull markets. To profit, you need to build strategies around such characteristics. If we receive USDT, we might first allocate it to arbitrage—locking in risk-free returns.
Currently in the bull market, risk-free yields on platforms like Pendle can reach 30%-40%. Even when calculating a precise Sortino ratio—subtracting expected minimum returns—the residual risk-adjusted return of active strategies becomes negligible. This explains why we focus on token-based alpha.
My market philosophy is simple: follow hot money—go where opportunities exist, where profits are clear, and logic is sound.
This year’s market rotation in crypto resembles China’s A-shares. Over the past five to six years, A-shares had a dominant theme each year—first carbon neutrality, now AI. Historically, during previous crypto bull/bear cycles I’ve experienced or reviewed, thematic rallies were rare—only this year has seen clear themes: AI and Memes. Before this, crypto lacked major themes—it was just dull. That’s what sets this year apart. If you caught the AI wave or jumped on Meme coins, you made substantial gains.
When identifying crypto trends and sector rotations, momentum is critical. Besides data, we monitor Twitter sentiment. But with fewer assets, the data we rely on leans more toward intrinsic value indicators.
Internally, we have a tool similar to Wind. We’ve been building factors for nearly two years, storing market data and Twitter sentiment. But we don’t focus much on sectors—we don’t chase sector rotation directly. Instead, our factors help identify coins within sectors that exhibit higher elasticity, and we buy those.
Ruiqi:
We view crypto as a highly speculative market, driven largely by continuous trading and sporadic event-driven activity. This is precisely why we remain actively engaged.
Compared to other financial assets, crypto exhibits stronger emotional and event-driven trading patterns—making it particularly suitable for quantitative capture, aligning well with our strengths.
Today, competition has intensified—both in execution and prediction. It’s a landscape of diverse players and strategies. Yet structural opportunities still exist, often rooted in emotional and event-driven behaviors. The market is undergoing structural divergence.
On predictability: pricing efficiency in established assets continues to improve. Trend formation used to take hours or even days; now, a trend may end within 10 minutes. Pricing errors driven by various factors get corrected rapidly. However, we still find strong alpha in new assets.
When participating in altcoins, narratives constantly introduce new assets—whether from competitions, startups, or emerging trends. Interestingly, previously effective factors often still work on these new assets. However, accessing them poses challenges—technical implementation, data integration, and execution stability remain weak points.
How Do Different Factors Contribute in Crypto Markets? What Are Their Underlying Return Sources?
Wizwu:
A key feature of crypto markets is high funding rates—large basis spreads. For futures, we can treat basis as calendar spread. Assuming they're equivalent, crypto calendar spreads fluctuate significantly. Arbitrage strategies are built around this phenomenon, and alternative factors often stem from the same logic.
Additionally, due to high volatility, certain altcoins exhibit extreme elasticity. Thus, real profits depend heavily on timing. We tested neutral momentum strategies and found their best-case outcome matches Bitcoin’s bull market returns. Without proper timing, achieving meaningful excess returns is difficult—largely due to crypto’s unique trading mechanics.
Moreover, exchange-provided data and off-exchange datasets differ significantly from traditional markets. Much of our excess return comes from leveraging these unique features and adapting strategies that are already saturated in traditional markets.
Ruiqi:
One representative sentiment-based factor is momentum—essentially chasing price increases and panic-selling declines. Its profitability stems from market overreaction.
For example, when retail investors see a coin rising, they assume the trend will persist and rush to buy. We can amplify this movement and profit from it. Alternatively, we can engage in momentum reversal trades—anticipating excessive reactions and positioning ahead of reversals. The core idea is exploiting market overreactions.
Event-driven factors profit from asset repricing, which requires reaction time. By monitoring Twitter feeds or latent large-scale market signals, we can react swiftly after events occur. For instance, when CPI data drops, Bitcoin may spike sharply. Rapid response enables profitable trades.
From a high-frequency perspective, many traders are insensitive to transaction costs. When executing large orders, they often do so entirely on a single exchange, creating significant market impact—and thus arbitrage opportunities. Liquidity factors remain persistently effective in high-frequency environments, serving as a key tool for alpha generation.
Compared to Traditional Financial Markets, What Are the Key Differences in Methodology for Generating Alpha in Crypto? How Can One Capture More Alpha in Crypto?
Zheng Naiqian @LUCIDA:
In recent years, I've increasingly realized that people are the most critical source of alpha. Despite its growth, the average skill level of participants in crypto—especially in secondary markets—is noticeably behind that of A-share markets.
Second, data and market infrastructure are extremely poor. There’s virtually no comprehensive data provider comparable to Wind or Bloomberg in equities. Data quality is low and highly fragmented. Simply acquiring usable data is a major pain point for many teams. Without data, how can you even begin modeling?
I believe institutions with clear advantages in talent and data relative to peers will have a stable edge in generating excess returns.
Wizwu:
Crypto differs from traditional finance in several ways: high volatility, high elasticity of small-cap tokens, and strong speculative sentiment. To generate alpha in crypto, strategies must be tailored to these characteristics.
A core issue is that risk-free arbitrage yields in crypto are simply too high. This undermines value factors. Stable USDT dividend-paying projects are extremely rare. When attempting to calculate value metrics—PE ratios, earnings yields—they pale in comparison to arbitrage returns in USD terms. Therefore, applying traditional value factors to assess crypto alpha is ineffective.
In crypto, the definition of “value” differs fundamentally. While equities prioritize fundamentals and P/E ratios, in crypto we may care more about "price-to-dream" ratios—the optimistic expectations for future potential and everything derived from realizing those visions.
A concrete example: a value-type factor could be tracking the number of native token holders with balances between 10–100 USDT on Layer 2 solutions like MATIC. Changes in this metric often signal emerging trends. When a blockchain anticipates a breakout app or mass adoption, growth in small holders usually acts as an early positive signal—highly correlated with market sentiment and price movements. From a factor perspective, addresses holding 10–100 USD worth of tokens likely represent real users rather than whales.
Ruiqi:
I’d summarize several key differences: Information asymmetry due to market fragmentation—crypto’s decentralized structure creates information gaps. Non-professional investors struggle to grasp market dynamics, making arbitrage opportunities more apparent.
Herd behavior and volatility—unlike traditional markets, crypto assets trade across multiple regional venues. This fragmentation amplifies herd mentality and frantic trading. Frequent attention shifts and irrational behavior are more prevalent in crypto.
Market manipulation—market manipulation is far more common in crypto than in traditional markets.
For most retail investors, exploiting this is nearly impossible. But for certain HFT firms, they can scale manipulative practices to extract alpha—activities that would be illegal and land you in jail in traditional markets.
Differences in Asset Management Product Landscape Between Crypto and Traditional Finance
Zheng Naiqian @LUCIDA:
I notice that over 80% of quant teams focus on neutral arbitrage strategies, leading to severe strategy homogenization.
From an effort standpoint, the underlying principles aren’t complex. If operating at lower frequencies, execution doesn’t demand much attention. This causes over 80% of products to compete fiercely in the arbitrage space, making strategies like CTA, options, or multi-factor models seem disproportionately labor-intensive relative to returns. Even high-frequency strategies require hardware upgrades and meticulous optimization—but still lag in AUM compared to arbitrage. Do you think arbitrage products will dominate the market long-term?
Wizwu:
Not just in crypto—even in traditional finance, fixed-income trading dominates volume-wise. Bond trading volumes across tiers are substantial, so arbitrage will always exist. As long as operations remain semi-compliant, crypto arbitrage yields can be 2–6x higher than traditional markets, offering vast capacity and return potential. This situation will persist.
As for other strategies like CTA—they offer scalable capacity too. They may gain broader recognition only after arbitrage yields decline. Then, our strategy’s Sharpe ratio will shine. Currently, arbitrage is measured in U, thanks to unified accounts offered by exchanges. We can run similar strategies on a token-denominated basis. Our current optimal allocation: run arbitrage in USDT, take risk in tokens.
Ruiqi:
I largely agree with Wizwu.
First, market fragmentation and barriers to entry will likely persist for the next two to three years. Therefore, arbitrage opportunities will continue to exist. Even if spreads narrow, arbitrage volume and capital capacity will remain dominant.
But eventually, arbitrage may cease to exist as managed products. Instead, it will be internalized by high-frequency quant teams running proprietary books—keeping profits in-house rather than sharing surplus with LPs. That’s likely the direction. Meanwhile, asset management products may settle for adjusted risk-return profiles—offering decent value, such as statistical arbitrage or CTA strategies. Such an environment may emerge in two to three years.
Zheng Naiqian @LUCIDA:
Crypto asset management product structures differ significantly from A-shares. In A-shares, the most popular products are index enhancers—tracking broad indices like CSI 300, 500, or 1000. Index-enhanced products are the easiest to sell, and most rely on multi-factor models underneath.
Yet I find such products almost nonexistent in crypto. Teams developing multi-factor strategies? Likely less than 10%. Why is the proportion so low?
Wizwu:
Because USDT yields in the market are simply too high. For example, I personally allocate nearly all my USDT to platforms like Pendle. In such cases, I wouldn’t even choose my own strategy—because after subtracting 30% risk and dividing by volatility, its performance doesn’t even match Sharpe ratios in traditional futures markets.
So, when risk-free yields are this high, everyone naturally opts for risk-free returns. Using standard metrics, any strategy must deduct this risk-free rate. When we use the true risk-free rate (~30% annualized) as a baseline, everything seems futile—no matter how you calculate it, it’s meaningless.
Our multi-factor strategies have become more diversified. Initially, we designed them along the lines of A-share or traditional futures neutral multi-factor frameworks. But over time, they’ve evolved—becoming more varied and incorporating more discretionary elements. The core reason: drawdown cycles in this market are short and change rapidly. Implementing pure multi-factor strategies faces structural issues. You can’t validate a factor’s long-term efficacy based on just two years of market data.
In traditional markets, we test a factor not just in A-shares but also in U.S. equities. Only if it worked for 20 years in the U.S. and 5 years in China would we consider it valid for large-scale deployment. In crypto, such validation opportunities are scarce. You might only have one or two years of backtest data—structurally insufficient.
Ruiqi:
My observation differs slightly—this depends on how we define the framework.
I observe that more players engage in time-series trading on major coins—trend following on Bitcoin and Ethereum is common. But very few teams apply trend-following across 100+ assets. Time-series trading is popular; cross-sectional trading is rare—that’s my takeaway.
Possible reasons:
First, data length. Most assets have gone through only one cycle—insufficient history for robust testing.
Second, even assets that survived multiple cycles—like EOS—became inactive post-2017/18, making them unsuitable for inclusion in tradable universes. Many similar cases exist in crypto. Few assets complete multiple full cycles while maintaining liquidity and activity—mostly just Bitcoin and Ethereum. Solana, for example, remained dormant for years before recently reviving.
Third, time-series factors tend to be more effective in practice than cross-sectional ones. The underlying mechanism—persistent emotional momentum—can be captured well using traditional trend frameworks. Cross-sectional relative strength factors are unstable because many assets themselves lack stability. Unlike traditional commodities or stocks, which endure multiple cycles, crypto assets from one cycle often disappear in the next—making relative strength comparisons unverifiable.
What Metrics Should Be Used to Assess Crypto Asset Value? Where Does Crypto Value Lie?
Ruiqi:
Currently, crypto value equals attention. It’s an attention-driven market. Regardless of a project’s fundamentals, if it captures attention, it gains value. This overlaps somewhat with Wizwu’s momentum point, but I see it differently—it’s more like a clickbait economy. Long-term, we hope—and many industry insiders and VCs are pushing—for value to reflect real utility and ecosystem competitiveness. But currently, the market isn’t fully aligned with that vision.
Bonus: What’s Your Current Market Outlook? Where Do You Think Bitcoin Is Headed Next? (Pure Speculation, No Liability)
Wiz:
Just guessing: it’ll keep consolidating here. Upside isn’t huge. Even if it breaks new highs, maybe another 30% up, then likely a pullback. At current levels, I think global risk assets generally have limited upside. Pure speculation—very exposed.
Ruiqi:
I’m more optimistic—I think rate cuts haven’t even started yet. Though I never had Bitcoin faith before, now I’m basically half a believer. So I think within this bull cycle, reaching $150K within two years is possible.
Join TechFlow official community to stay tuned
Telegram:https://t.me/TechFlowDaily
X (Twitter):https://x.com/TechFlowPost
X (Twitter) EN:https://x.com/BlockFlow_News
@lucidafund
