Moritz Heiden & Moritz Seibert – Trend-Following Spreads (S7E25)

Episode Overview

• Explores the "anti-scale" hedge fund model, arguing that capping assets under management (e.g., at $500M) allows for superior performance by accessing niche, illiquid markets that large funds cannot trade.
• Details the mechanics and advantages of synthetic spread trading (calendar, location, substitution) to exponentially increase the number of independent, uncorrelated bets in a portfolio.
• Compares operational realities against academic theories, specifically why real-world friction (manual brokering, complex legal structures) creates a defensive "moat" for returns.
• Discusses the limitations of "prediction markets" for trend-following strategies, explaining why binary outcomes fail to capture the necessary "fat tail" returns found in traditional financial markets.

Key Concepts

• Capacity as a Performance Feature Most funds aim to grow indefinitely to maximize management fees, but this forces them into crowded, efficient markets. By strictly capping fund size, managers can trade "obscure" markets (e.g., sunflower seeds, canary seed, specific carbon credits). These markets are often less efficient and offer better trends, but they lack the liquidity required for multi-billion dollar funds.

• Operational Complexity as an Alpha Moat Alpha often hides behind friction. Many quants rely on "screen liquidity" (electronic execution). Takahe Capital seeks markets requiring "broker relationships," voice execution, and manual workflows (like negotiating ISDA agreements). Because these markets are operationally painful and legally complex, they are less crowded. This friction acts as a barrier to entry against large, automated competitors.

• Regulatory Arbitrage for Diversification True diversification requires access to distinct global drivers (e.g., Chinese commodities, European power). A US-domiciled fund structure is often restricted from these markets by CFTC regulations. Utilizing a Cayman Master structure provides a "wide open field," allowing the fund to access international instruments that US-centric funds cannot, thereby increasing the pool of uncorrelated opportunities.

• The Taxonomy of Synthetic Spreads To manufacture diversification without adding new asset classes, managers can trade the relationship between assets rather than the assets themselves. This "Combinatorial Alpha" includes:

• Calendar Spreads: Same asset, different dates (Long Jan Gas / Short Feb Gas).
• Substitution Spreads: Goods that replace each other (Arabica vs. Robusta coffee).
• Processing Spreads: Raw vs. Refined (Soybeans vs. Soybean Oil).

• Location Spreads: Same good, different place (Brent vs. Dubai Oil).

• Execution Reality vs. Academic Purity Academic models often suggest "constant maturity" spreads (e.g., always holding a 6-month spread), which implies daily rebalancing. In practice, this incurs prohibitive transaction costs via the bid-ask spread. The effective approach is trading specific contract pairs (e.g., Long June / Short Dec) and accepting the "messiness" of time decay and seasonality to preserve returns.

• Structural Failure of Prediction Markets for Trend Following Trend following relies on "unlimited upside" (letting winners run). Prediction markets (betting on elections or sports) are binary—they pay out 0 or 100. As a trend strengthens in a prediction market (probability rises from 50% to 90%), the potential future upside shrinks. This is the mathematical opposite of financial markets, where strong trends often beget further extensions.

Quotes

• At 0:06:17 - "We think quite frankly, all hedge funds should be doing that [charging 0% management fees]. If you're holding yourselves out as a firm that's capable of producing alpha... I think you should demand and be compensated with an incentive fee or a performance fee." - Discussing the philosophy of pure alignment with investors vs. asset gathering.
• At 0:11:13 - "We feel that they [niche markets] are really the most important ingredients for the way that we trade. Because these things go in at the front... and the more of these we can get, the more independent ones we can get, the greater the likelihood that we can crystallize the edge that we have." - Explaining the statistical "Blackjack" theory of increasing independent bets.
• At 0:19:58 - "When we put it into our quant machines, the statistical significance tests of that statement is fairly difficult to make... It's far easier for us... to say that all of these markets have the same expectation over time... So we're spraying these bets in an as diverse manner as we possibly can." - Arguing that niche markets provide diversification rather than inherently better trends.
• At 0:21:26 - "This is a negotiation... You like the pricing and you like the legal setup... It's just simply because of the fact that this is a bit more work and more complex from a technological point of view... that's probably one of the reasons why not everybody is doing it." - Highlighting operational friction as a barrier to entry for competitors.
• At 0:30:15 - "Diversification is really the thing that we're getting for free. I truly believe that... the code runs and whether you run it on 100 markets or 100 spreads or 200 things, it doesn't really matter. But if we can get an ever so small diversification benefit... then I think we should at least aim to harvest that." - On why the computational cost of spreads is zero but the portfolio benefit is high.
• At 0:31:37 - "Which gets away a lot of the permutations essentially because if you do not look only at the curve or contracts but also at liquidity... you can narrow down the number of tradable combinations to quite an extent." - Explaining how liquidity acts as a natural filter to avoid overfitting spread combinations.
• At 0:41:59 - "We don't say, 'Oh only because of that [volatility] we're not trading that'... but we give it a different risk budget because these things can exhibit more volatility, more punchiness... We would probably size that spread smaller." - On managing "widowmaker" markets through sizing rather than avoidance.
• At 0:51:30 - "We trade the same number of contracts on each leg. So this creates a notional mismatch... We could be accused of not being hedged. But... we don't want to be hedged. We want to be participating in the spread as it has been presented to us." - Challenging the academic idea of volatility-adjusting every leg of a spread.
• At 0:56:39 - "What we are doing in trend is always betting on strength... but in prediction market[s] it's completely the opposite... my potential win shrinks actually from initially 80 cents to win on the dollar to then in the end maybe only 40 cents... we need this unlimited upside." - Explaining why prediction markets are structurally incompatible with trend following.

Takeaways

• Target "Invisible" Liquidity: Do not ignore markets just because they lack screen volume; deep liquidity often exists in manual markets (voice brokering), offering alpha that fully electronic firms cannot access.
• Incentivize via Structure: Adopt or seek fee structures that charge for performance (alpha) rather than existence (AUM); this forces a focus on strategy integrity over marketing.
• Multiply Markets with Spreads: Use synthetic spreads (calendar, location, processing) to create new, uncorrelated return streams from existing data without needing to find new asset classes.
• Manage Volatility via Sizing: Do not blacklist volatile "widowmaker" markets (like seasonal Natural Gas spreads); instead, drastically reduce their position size to capture the trend while managing the ruin risk.
• Prioritize Operational "Messiness": Embrace specific contract dates and manual rolls rather than paying the continuous transaction costs required to maintain "constant maturity" theoretical positions.
• Evaluate Legal Constraints: Assess whether your fund structure (e.g., US vs. Cayman) is artificially limiting your diversification by restricting access to foreign regulatory regimes.