December 6, 2024
Almanak gives everyday users the ability to profit from quant trading techniques in DeFi
Ah, the golden days of DeFi Summer. A time when every day felt like a treasure hunt on Uniswap—sniffing out freshly deployed tokens and farming absurdly high APY food coins. It was messy, ridiculous, and thrilling.
And let’s be honest, we’ll probably never see anything quite like it again.
Back then, the playing field was almost—almost—level.
In fact, retail and degens had an edge. We were the first to experiment with new protocols, riding waves of high-yield farming opportunities before the big players even noticed. Things moved fast, but if you were plugged in, you could move faster.
But paradise never lasts forever.
It didn’t take long for the heavyweights to show up. Sophisticated players with deep pockets and even deeper strategies—MEV arbitrageurs, algorithmic market makers and liquidity providers. They saw the same opportunities and brought their institutional-grade tools to the DeFi sandbox. Suddenly, a scrappy retail trader like me didn’t stand a chance.
A study from the Bank for International Settlements (BIS) tells a common story in DeFi today: Retail liquidity providers are consistently outplayed by a small group of sophisticated, institutional participants, who control 80% of the total value locked (TVL). On most days, retail participants providing liquidity on Uniswap v3 lose money.
If the odds are stacked against us, how can the small boys start winning again?
Almanak’s vision is bold and straightforward: to make it as easy as possible for anyone to create AI agents that make money.
They are democratising institutional-grade trading technology and believe that AI-driven financial strategies should be accessible to anyone.
To bring this vision to life, Almanak has built a platform that empowers users to create, test, optimise, deploy, and monitor financial agents in crypto. These agents will work tirelessly on your behalf to trade, generate yield and manage risk.
In practical terms, Almanak puts advanced quant trading tools into the hands of more people — whether you’re building your own strategy or making use of other people’s strategies to generate returns.
Quantitative (or algorithmic) trading is well-known in the Tradfi world.
In traditional finance, 60-75% of trades are algorithmic, driven by sophisticated algorithms that crunch massive data sets to spot patterns and predict trends. It’s trading stripped of emotion, powered by pure logic.
But pulling this off isn’t simple. First, you need access to huge amounts of data. Then, you need powerful hardware to train machine learning models on that data, to churn out predictions. Finally, you need smart mathematicians and engineers to design algorithms that make it all profitable.
Most people don’t have those resources. And let’s be real—if you did, you’d probably be running a hedge fund, raking in millions, and definitely not reading this. (We’re kidding… mostly.)
If there’s one name that embodies the success of this style of trading, it’s the Medallion Fund, founded by James Simons—aka the “Quant King.” With a team of 90+ PhDs in math, physics, and computer science, Medallion uses statistical models and vast amounts of market data to identify patterns hidden in noise.
Their results are legendary. 66% annual returns before fees (39% after fees) over three decades, with only one negative year in its history. Medallion has consistently outperformed the likes of Buffett, Soros, and Lynch.
Now imagine if you had access to the power of those PhDs, while sitting at your home in your boxer shorts.
For years, quant trading was the exclusive domain of elite hedge funds with deep pockets, and proprietary infrastructure. But that’s changing. It is now breaking into the mainstream.
In Web 2, platforms like QuantConnect make quant trading more accessible. Since its launch in 2012, QuantConnect has empowered traders and engineers with institutional-grade tools. Its stats are eye-opening: over 315,000 users worldwide, churning out 2,500 new algorithms and 1 million lines of code monthly.
QuantConnect’s cloud-first infrastructure puts terabytes of financial, fundamental, and alternative data within users’ reach. It integrates features like backtesting and live trading, while its pay-as-you-go pricing model eliminates hefty upfront costs.
95% of QuantConnect’s revenue stems from cloud fees and its growing data marketplace, leveling the playing field for individuals and small firms.
What about in Crypto? There’s a diverse cast of players, including specialized crypto quant firms like Gauntlet, Quant Matter, and GSR Markets; TradFi heavyweights like Jump Trading and Citadel Securities crossing the bridge into Web3; and independent quants and small teams—the scrappy innovators coding strategies from their basements.
Will we see a similar disruption and expansion of the use of algorithmic strategies in crypto? We believe it is inevitable.
Launched in 2022, Almanak has spent two years quietly building the blueprint for the next generation of DeFi infrastructure — a world where there are more agents than real humans on the blockchain.
The idea makes sense. After all:
Almanak started small in its early days, initially building a simulator to optimise DeFi yields. But over time, that effort expanded into something much bigger: a generalised, agent-centric infrastructure tailored for DeFi.
Fast forward two years and the platform is already running a handful of live strategies with real capital deployed.
Here’s the playbook for a typical quant trading hedge fund:
Hedge funds often juggle thousands of strategies in their library, constantly testing them against current market conditions to pinpoint the most profitable ones. The process is dynamic. Strategies are deployed, monitored, and adjusted in real-time.
Almanak aims to simplify and automate these processes using machine learning and AI.
This is a game-changer for smaller crypto quant teams and individual traders. It simplifies the typically cumbersome DevOps processes and eliminates traditional barriers like high setup costs and technical complexity.
The platform’s backbone rests on two vital components: Simulator and Execution Engine.
Almanak’s secret sauce is in the simulator, which blends agent-based modelling and blockchain-driven Monte Carlo simulations. Its end-to-end infrastructure lets you build, test, deploy, and monitor trading agents with minimal overhead. Together, these create a predictive sandbox for financial strategies with uncanny precision. Almost like gazing into a crystal ball.
So, quant traders can focus on strategy instead of wasting time wrestling with DevOps or managing integrations across chains and trading venues. This cuts setup time by up to 50%, giving smaller teams a leg up with tools that typically take months of work and are now compressed into a few clicks.
Let’s say you’ve got a knack for spotting opportunities in the market. Almanak makes it easy to transform your ideas into actionable strategies. This could include:
• Providing liquidity on Uniswap V3 based on specific conditions.
• Hedging volatility by shorting on a Perp DEX like Hyperliquid.
Designing a Uniswap V3 strategy might seem straightforward, but critical questions arise: When do you rebalance? How does volatility impact your returns? What happens if the protocol changes parameters? Intuition alone won’t suffice, especially when real money is on the line. Markets are unpredictable, and even promising strategies can fail without rigorous testing.
That’s where the Almanak Simulator comes in.
It’s a digital sandbox designed to replicate real-world blockchain conditions with precision. Unlike traditional simulators that stop at price feeds, the Almanak Simulator goes deeper, simulating the blockchain’s state machine, core app logic, and even user behaviour.
Powered by a custom Ethereum Virtual Machine (EVM), it mirrors blockchain environments while integrating live market data. This presents a fast, scalable, hyper-realistic environment for traders to fine-tune their strategies.
Users can define parameters like duration, agents involved, and market conditions. The Simulator then models the agent behaviours across various market dynamics, delivering a detailed view of strategy performance across multiple scenarios.
Did your strategy turn a profit? What’s the risk/reward profile? Should you rebalance or pivot? The Simulator provides clear, actionable insights to optimise strategies for profitability and minimise risk.
The simulator’s core users include experienced quant traders—a good analogy would be the top 20% of Dune power users—and a secondary audience of less technical individuals who can tune pre-built strategies to suit their needs.
At the heart of the Almanak Simulator is the agent brain, powered by Agent-Based Modeling (ABM).
ABM models individual actors—human traders, arbitrageurs, bots—and their interactions. Think of it as observing a realistic representation of DeFi markets come alive in real-time.
ABM is not new. It’s already used in urban planning and epidemiology to simulate intricate systems. The concept is simple: agents follow rules to adapt, interact, and respond to their environment. The magic lies in how individual actions scale into larger patterns.
Fun Fact: Singapore uses ABM to optimise urban planning. By simulating the behaviour of thousands of agents—cars, pedestrians, buses, and trains—planners can predict traffic patterns and identify congestion points, eliminating bottlenecks before they even occur. If you were here during Token 2049, you probably noticed how smooth everything was. That’s ABM in action.
In DeFi, ABM is just as powerful. Every action, whether it's a trade or a protocol update, can create ripple effects that can destabilise or reshape the market. Traditional backtesting misses these complexities. ABM, on the other hand, thrives in this chaos. It captures the impact of things like MEV bots and protocol parameters (like funding or utilisation rates) that historical data alone can’t predict.
ABM also shines in edge cases and unexpected chain reactions, such as liquidation cascades and depegs.
The catch? ABM is data-hungry. Lucky for us, blockchains offer a treasure trove of recorded transactions to model user behaviour and system dynamics in ways that were never possible in traditional finance.
To complement ABM, the Simulator leverages Monte Carlo simulations, running countless “what if” scenarios to predict how strategies perform across various market trajectories. This approach stress-tests strategies against volatility, ensuring they’re robust and optimised for real-world conditions.
Monte Carlo is the famous casino town in Monaco, synonymous with expensive cars, beautiful celebrities, and chance. The method traces its roots back to the Manhattan Project, where scientists, inspired by calculating the odds of winning at solitaire, adapted it to solve complex problems like nuclear reactions.
In Alamank’s simulator, Monte Carlo simulations create randomised price movements and market conditions within a realistic blockchain environment. Each scenario represents one possible future. By analysing these scenarios, the system identifies statistically optimal strategies while highlighting vulnerabilities.
Here’s how it works: You set parameters—strategy duration, involved agents, and simulation conditions. The Simulator models agent behaviours and market dynamics to provide a comprehensive view of strategy performance across multiple scenarios.
Like me, you might be asking: what if you’re not a math genius or strategy designer?
Almanak’s Execution Layer makes it easy to deploy agents without coding or advanced know-how. Choose from a library of shared strategies, and Almanak will handle the heavy lifting.
Each agent runs on predefined strategies powered by machine learning models or rules-based logic. For example, a trading agent might analyse volatility to determine when to buy or sell, while another focuses on liquidity provisioning or rebalancing using preset parameters.
Almanak’s agents operate off-chain, meaning that strategies and code stay confidential—safe from reverse engineering. When it’s time to execute, agents interact securely with on-chain environments, ensuring every transaction is smooth and secure.
The platform integrates with trusted multi-signature wallets like Eulith and Gnosis Safe. Agents operate with strict, pre-approved permissions using an Almanak smart contract wallet. For instance, an agent managing a Uniswap LP strategy won’t suddenly start trading on Hyperliquid. Users' assets remain locked down and fully under their control, in keeping with an open, non-custodial system.
The Execution Layer is designed with flexibility in mind. Users can access it via a user-friendly interface or a software development kit (SDK) for deeper customisation. Pre-built libraries of agents and strategies save time and make it easy to get started.
Traditional machine learning models excel at crunching numbers, but large language models (LLMs) like GPT-4, Claude, and Llama-3 bring something new to the table: the ability to process unstructured data. They can parse whitepapers, tweets, and project blogs and weave them into trading strategies. They can scan the broader landscape, surface actionable insights, and align opportunities with a user’s risk profile.
For now, though, Almanak’s AI agents steer clear of LLMs. During the beta platform launch, strategies are powered by non-generative AI, and for a good reason: hallucination. LLMs can make mistakes or behave unpredictably, and when you’re managing millions of dollars, even a small error can have catastrophic consequences. We wouldn’t want that, would ya?
Almanak’s approach is deliberate. They’re starting by building a strong foundation—using non-generative AI that simplifies and accelerates every step of the quant trading process. From there, LLMs will be integrated thoughtfully, enhancing each stage of the workflow while ensuring 99.x% performance accuracy in their models.
The end goal is a future where LLMs handle the heavy lifting—developing quant strategies and executing them. This is blending the precision of traditional machine learning with LLMs' contextual intelligence to make smarter trading decisions.
Big things are on the horizon for Almanak. The plan is to start with tools for high-value quants and gradually expand to less technical users and retail markets, unlocking DeFi's full potential for everyone.
First up, a pre-liquid token community boostrapping event has just launched via Legion. The allocation is at a $45M fully-diluted valuation, with 30% unlock at TGE, and the 24-month linear vesting/6-month cliff. Legion is a new ICO-style platform backed by Delphi Digital that is democratising access to early-stage investments using a reputation-based scoring system to allocate tokens. This marks a key step in building momentum for Almanak’s ecosystem.
Legion investors will also gain early access to the Almanak platform, slated to go live in Q1 2025. To kickstart adoption, the team will provide an initial supply of agents and strategies, focusing on retail-friendly financial products for saving and investing. To ensure stability during the rollout, total TVL on strategies will initially be capped at $25M while the platform develops and optimises.
The token generation event is expected in H1 2025. Once the token is live, it will serve as a critical incentive mechanism to drive the development of new strategies and liquidity into the Almanak ecosystem.
After that, things get even more exciting. Through 2025, the focus will shift toward preparing for AI-driven quants, capable of autonomously analysing markets and designing strategies.
This means building infrastructure that integrates with advanced AI models like LLMs, and enabling plug-and-play functionality for future users. A marketplace for strategies is also in the works, offering vault-based strategies—where users can invest without revealing strategy details—and parameterised skeleton strategies available for purchase or lease.
Almanak is led by a team with diverse expertise across technology, finance, and operations.
The CEO, Michael, holds an MSc in Mathematics and has a background in decentralised finance, having worked with projects like Delphi Digital Risk, Layer Zero, Stargate, and Bancor. His experience in risk management and blockchain technologies provides a solid foundation for Almanak’s focus on building robust trading tools.
Lars, the CTO, has a BA in Computer Science and a history of technical leadership. He was previously a Lead Engineer at Delphi Digital and a Software Engineer at Deeploy, bringing valuable experience in developing scalable and reliable software infrastructure.
On the operational side, Lukas, the COO, combines legal and growth expertise. With an MSc in Law and a background as a securities lawyer, Lukas has also worked in growth roles at UBER and Founders Institute, adding a strong operational and strategic component to the team.
The team has raised $6.7M in funding so far: $1.2M pre-seed in 2022 and $5.5M seed in 2023.
Let’s ask some hard questions:
If I’m a smart quant with profitable strategies, why would I share them? Why not just milk the alpha myself?
The Almanak token is designed to tackle this existential question head-on. Its design is inspired by Bittensor and Curve. Almanak’s ecosystem aligns incentives across three distinct roles to ensure high-quality contributions and sustainable growth.
Firstly, let's examine the actors in the ecosystem.
Strategy contributors are the quants and developers who use Almanak’s platform and simulator to create profitable strategies. The code for these strategies may be private, private but shared (others can use) or public (code is open).
Agent managers are focused on optimising agent deployment and strategy selection. They allocate agents to the most profitable strategies, balancing risk profiles and timeframes.
Liquidity providers contribute their capital to the ecosystem, aiming to earn competitive returns by funding agents and strategies.
Alamanak uses token emissions to incentivise ecosystem participants, focusing heavily on strategy contributors who are the lifeblood of Almanak.
Strategy contributors will receive 75%, agent managers 5%, and liquidity providers 20%.
The proportion of emissions a strategy contributor receives depends on the returns (profits) made by their strategies. High-performance strategies that deliver strong alpha will see greater adoption by agents, earning their creators a larger share of tokens. Think of it as a decentralised, performance-based fee model.
This setup benefits quants who lack the capital to fully exploit their strategies. Instead of sitting on untapped alpha, they can monetise their expertise by sharing strategies and earning token rewards, similar to hedge funds charging performance fees.
Liquidity providers earn emissions on top of their strategy-driven returns, providing an added incentive to participate and enhance the ecosystem’s overall liquidity.
Token holders have governance rights, which include managing the project treasury and protocol parameters (emissions and fees).
Almanak generates revenue through three key streams, all flowing into the project treasury:
Let’s examine each of these and estimate the size of the demand using a back-of-napkin method.
To estimate transaction fee revenue, we can start with the total on-chain transaction volume. Using data from Coingecko on November 29, the daily trading volume across decentralised exchanges (spot and perpetuals) was approximately $16B. While this figure fluctuates with market cycles, it provides a solid baseline.
If Almanak’s agents capture 5% of this on-chain trading volume and charge a 0.05% fee, the revenue potential looks like this:
• Daily Volume Captured: $800M (5% of $16B)
• Daily Revenue: $400K ($800M x 0.05%)
• Annualized Revenue: $146M
This could be optimistic, but in essence, Almanak’s ability to grow transaction fee revenue will hinge on three key factors:
Broader crypto adoption and market sentiment will be critical. During bull cycles, total volumes can surge significantly above current levels.
The shift from centralized to decentralised exchanges (DEXs) will play a major role. Platforms like Hyperliquid with excellent UX ****will accelerate this transition, boosting DEX volumes relative to centralised exchanges.
Expanding the TVL managed by Almanak’s agents and increasing strategy adoption will drive the platform’s ability to capture a larger share of the on-chain trading pie.
Performance margins tie directly to the TVL managed within Almanak’s ecosystem and the profitability of the strategies.
Let’s assume $1B TVL and an average 10% APY generated across strategies. This aligns with examples like sUSDe on Ethena, which achieved a delta-neutral carry trade average APY of ~10% in October 2024. If Almanak charges a 5% fee on the profits, that would be $5M in annualised revenue.
This figure can scale significantly with higher TVL or increased profitability margins from Almanak’s strategies.
Unlike the other revenue streams, cloud computing margins are tied to user activity rather than TVL or profitability. Running Monte Carlo simulations and agent-based models requires significant computational power, which users pay for to train agents, optimise strategies, and run simulations. Almanak will likely source compute from AWS or decentralised GPU marketplaces, adding a 1–10% margin on costs.
As an example calculation: Lets say each simulation run costs $50, and each user runs an average of 10 simulations a month. At 5,000 users, this results in $2.5M in total compute costs monthly. If Almanak takes a 5% cut, it would generate $125K in monthly revenue.
While cloud margins are meaningful, they are unlikely to surpass performance-based fees or transaction fees in importance. Global compute costs are trending lower, reducing the overall revenue potential. However, as a revenue stream tied to user activity, cloud margins provide a steady and reliable contribution.
Another key demand driver comes from DeFi protocols themselves. These protocols can purchase and stake Almanak tokens to amplify emissions for strategies that utilize their platforms. This approach mirrors the mechanics of the Curve Wars, where protocols compete for liquidity by influencing token emissions.
By doing so, they create an incentive structure that attracts agent traffic and drives Total Value Locked (TVL) growth. This mechanism introduces a competitive dynamic among protocols vying for agentic traffic. In my opinion, this is a fascinating solution tailored to the emerging AI agent trend.
It’s an especially compelling strategy for newer protocols looking to bootstrap liquidity and gain traction in a crowded market.
Almanak’s agents have a singular mission: to make money. I respect that.
Unlike the flood of AI agents debuting daily on X, chasing memes and clout, these agents won’t be moonlighting as entertainers or influencers. The platform is purpose-built to create profitable traders, streamlining blockchain simulations and slashing the time taken to identify and capture alpha for winning strategies. The simulator is Almanak’s secret weapon.
An open marketplace for strategies has the potential to be far more powerful than a closed one. Imagine a smart developer in India creating a killer strategy but lacking the funds to capitalise on it. Almanak bridges that gap, connecting creators with capital.
But here’s the critical question: Will top-tier strategy creators share their best ideas on the platform if the incentives are right? With thoughtful design and alignment, the answer could be yes. But only time will tell. Without a steady stream of high-quality strategies, the ecosystem won’t thrive.
Designing a balanced incentive system is no small feat—players will inevitably seek to extract maximum value. Both Bittensor and Curve have grappled with challenges in their token models, and there are lessons to be learned. We’ve seen that effective incentive design on Bittensor subnets often requires major effort and iteration.
The team must anticipate and address potential gaming tactics when token incentives go live. For example, whales might park capital in their own strategies while hedging externally, exploiting token emissions without contributing real value. Such loopholes can distort key metrics, disproportionately rewarding strategies that are either unprofitable or excessively risky.
Almanak is flipping the script on DeFi.
It is quite possible that 80-90% of blockchain activity will be agent-driven within the next few years. In that future, Almanak is well-positioned. It’s taking the tools once hoarded by hedge fund giants and making them accessible to everyday traders and small teams.
The big question is whether it can nail the execution. If it does, Almanak will be rewriting the playbook for AI-driven finance. It's high stakes, but there's big upside.
Thanks for reading,
Teng Yan
This research deep dive was sponsored by Almanak, with Chain of Thought receiving funding for this initiative. All insights and analysis are our own. We uphold strict standards of objectivity in all our viewpoints.
To learn more about our approach to sponsored Deep Dives, please see our note here.
This report is intended solely for educational purposes and does not constitute financial advice. It is not an endorsement to buy or sell assets or make financial decisions. Always conduct your own research and exercise caution when making investment choices.
