Video summary

The Secret to Uncorrelated Alpha in Crypto - Leigh Drogen on Starkiller Capital’s Sharpe Ratio of 4

Main summary

Key takeaways

Finance

Finance-focused summary (crypto investing, portfolio construction, risk/returns)

Core thesis / strategy overview

  • Starkiller Capital runs a crypto market-neutral strategy targeting “uncorrelated alpha.”
    • The alpha is framed as coming more from protocol/market-structure failures (e.g., hacks) than from general BTC/ETH price direction.
  • Reported performance:
    • Positive 27 of the last 28 months
  • Target profile:
    • Aim for a Sharpe ratio close to ~4 (“almost four”).

Yield generation in DeFi (market-neutral)

The approach combines multiple DeFi yield sources, allocated using a risk-adjusted expected yield framework:

  • Liquidity provision / yield before and after launch
    • Provide liquidity before a protocol/app is live (early liquidity provisioning)
    • Participate as it is live
  • Trading/arbitrage and carry-like trades
    • Basis / arbitrage strategies
    • Straight lending
    • Market maker strategies

Central process

  1. Build a risk score per opportunity using quantitative + qualitative factors.
  2. Compute an expected yield and derive a risk-adjusted yield score.
  3. Determine position size so that when an opportunity “eventually blows up” (speaker’s assertion), losses stay controlled via smaller sizing for higher blow-up probability.

Risk scoring framework (explicit)

Qualitative variables (diligence)

  • Reputability / track record of the founding team
  • Provenance of code (framed as hacking risk)
    • Many hacked protocols are forks of previously hacked designs
    • Lending often reuses older layer components, so code lineage matters
  • Who backs the team
    • Examples include large VCs/trading firms vs. angels
    • Backers may “fill the hole” after incidents, especially mentioned in bridges

Quantitative variables

  • Where the yield comes from and whether it’s sustainable
    • Avoid “magic money trees”
    • Example: 2020–2021 farms with ~1,000% APR, largely from governance token rewards priced at crazy valuations (unsustainable)
  • Persistence of incentives and lending/carry dynamics, including:
    • How long token/stablecoin incentive programs last
    • Whether incentives can be removed suddenly
    • Lending utilization / breakpoints where spreads close
    • Probability distributions for spreads and carry performance

Concrete trade example (carry / incentive-funded spread)

  • CME-basis “tokenized fund” example via Superstate
    • Long positions: Ethereum, Bitcoin, Solana (stated as “whatever”)
    • Stakes to earn yield
    • Uses collateral to short CME futures for the same assets
    • Reported fund yield: ~5% to 8% annually
    • Described as “very little counterparty risk”
  • Incentive/looping detail on Aave Arcane (private Aave variant)
    • Ripple incentivized borrowing rUSD against USCC
    • Borrow cost: ~2.5%
    • Implied strategy: earn the basis/fund yield while borrowing at ~2.5% to capture the spread
  • Key caution/risk point:
    • Incentives may be pulled
    • Spreads may compress if lending utilization rises
    • These dynamics are modeled prior to sizing/entering

“Three sources of alpha” (ranked)

  1. Diligence / risk management (largest alpha source)
    • Framed question: for every $10 deployed, how many blow up?
    • Position sizing aims to keep wipeout risk contained.
  2. Access (second)
    • Early liquidity provisioning programs (often VC-like)
      • Liquidity for 3/6/9/12 months
      • Receive native protocol yield plus token warrants at some valuation at TGE
    • Example: deposit $1m for 6 months
      • Native yield: 10%
      • Warrants struck at $125m FDV
      • Potential: +15% APR if valuation targets hit (e.g., $150m vs $300m)
  3. Nimble hopping between incentives (third)
    • Use AI tooling to scan and reallocate as incentives shift across DeFi
    • Human-only scaling across the whole landscape is implied to be impossible

Targets / capacity / sizing constraints (explicit numbers)

Target outcomes

  • Net return: 15%–20% annually
  • Sharpe: ~4
  • Very minimal drawdown

Capacity estimate

  • Belief that capacity at these targets is only ~$100 million
  • Peer comparison:
    • Others may run $400m–$600m but with lower net returns (~8%–9%) and lower Sharpe (~2)

Why capacity is limited

  • Incoming capital can compress yield opportunities
  • Growth in the number of competing incentive programs (“many experiments”)
  • Risk limits push them toward smaller books rather than accumulating assets

Position sizing rule (explicit “wipeout” risk constraint)

For each position/opportunity, estimate the probability of complete wipeout.

  • Portfolio rule:
    • They “never want” a position where the expected loss implied by wipeout probability is > ~1% of the book
  • Example sizing logic:
    • Experimental protocols with huge APR (e.g., >80%–100% APR long-tail): sized at ~1%
    • Basic lending like USDC in a Morpho vault: infinitesimally small wipeout probability; could be closer to full allocation (subject to their framework)

Market directionality vs uncorrelated alpha

  • Claims:
    • Returns are not correlated in direction with BTC/ETH beta
    • Loss probability is driven by tail events like hacks and market-maker strategy failures
  • However, magnitude is described as pro-cyclical:
    • When markets are “hot” and liquidity/leverage demand rises, available yield increases
    • Their risk posture extends outward → monthly return volatility increases
  • Monthly return range (market-neutral strategy):
    • Roughly 0 to ~250 bps per month depending on risk deployed

Trend-following / momentum strategy (directional “beta”)

Core belief / mechanics

  • They also discuss/run a trend-following / cross-sectional momentum directional strategy.
  • Belief: momentum is the only persistent alpha due to persistent human behavior.

Examples described:

  • Time-series: “50-day moving average on Bitcoin”
    • Long above the average; exit/short when below (speaker says “out when below”)
  • Cross-sectional:
    • Each week, rank assets by 30-day performance
    • Long top 20%
    • Short bottom 20%
  • Fundamental overlay:
    • Avoid pure “Ponzi/fraud/rugs”
    • Avoid long momentum exposure in structurally weak assets

Liquidities / concentration / liquidity limits in momentum

  • Liquidity floor exists for eligible tokens
  • Long side:
    • Typically ~20–25 tokens max
    • Long-tail candidates are smaller market caps; liquidity floors still apply
  • Short side:
    • Higher liquidity floor to avoid perps funding rate squeezes / liquidity constraints
  • Risk control on exits:
    • Positions shouldn’t be so large they can’t be unwound quickly
    • Speaker explicitly references “less than a week” to get out

Additional crypto risk points emphasized (beyond beta)

  • Vampire attacks on longs:
    • New protocols can steal users/mindshare when momentum reverses
  • Short-side market structure risk
    • Token manipulation/squeezes
    • Perpetual futures dynamics, funding rate behavior, “shorting heartbreaks”
  • Short concentration caution:
    • Less willing to concentrate in shorts due to squeeze unpredictability
    • Uses broader diversified short baskets

Crypto “fundamental” worldview (intrinsic value framework)

  • Block space is viewed as worthless/overabundant after Ethereum “blobs” made blockspace cheap/free.
  • Historical analogy:
    • Fiber optics (late 1990s): excess supply led to valuation collapse for fiber-related companies
  • Implications:
    • Many L1/L2 governance tokens likely trend toward zero
    • Long-term preference: assets where business cash flows have a direct relationship to token value
  • Example with “intrinsic value”:
    • Hyperliquid is cited positively for revenue/cash-flow-style reasoning (a P/E-like idea is mentioned, though no number is given)

Key numbers & performance metrics extracted

  • Sharpe ratio: close to ~4
  • Track record: positive 27 of the last 28 months
  • Portfolio targets:
    • 15%–20% net annual return
    • Sharpe ~4
    • Minimal drawdown
  • Capacity: ~$100 million
  • Peer comparison: $400m–$600m capacity, ~8%–9% net, ~2 Sharpe
  • Fund yield example (Superstate basis fund): ~5%–8% annually
  • Borrow rate example (Aave Arcane): ~2.5%
  • Yield/incentive examples:
    • 1,000% APR farms (governance-token reward-driven)
    • Aave lending example after “Kelp DAO hack”: ~17%–18% APR (stated for vanilla USD/EUSD/USDC lending)
    • Long-tail APR mentioned: north of 80%–100% APR
  • Market-neutral return range: ~0 to ~250 bps per month
  • Risk rule: expected wipeout-loss constraint implies ≤ ~1% of the book
  • Early liquidity program timing: 3/6/9/12 months
  • Warrant example: struck at $125m FDV, native yield 10%, potential +15% APR
  • Momentum illustration:
    • look back 30 days
    • weekly top/bottom 20%
    • “50-day moving average” for BTC example

Disclosures / disclaimers

  • No explicit “not financial advice” disclaimer appears in the provided subtitles/text.

Tickers / assets / instruments mentioned

Crypto assets (spot / long / baskets)

  • Bitcoin (BTC)
  • Ethereum (ETH)
  • Solana (SOL)
  • Dogecoin (DOGE)
  • Hyperliquid (token name mentioned; ticker not explicitly provided)

Stablecoins / DeFi currencies

  • USDC
  • rUSD
  • USDE (mentioned in “looping USDE” context)
  • USCC (collateral in the Aave Arcane example)

Protocols / venues / infrastructure

  • Aave Arcane
  • Aave
  • CME futures (no specific CME ticker)
  • Morpho (including “Morpho lending vault”)

Other instruments / concepts

  • Token warrants
  • Perpetual futures (no ticker)
  • Funding rates (perpetual futures mechanics)

Macro / benchmark mentioned

  • Nasdaq (risk-on / correlation context proxy)

Step-by-step / methodology frameworks explicitly described

Market-neutral DeFi portfolio construction

  1. Enumerate DeFi yield opportunities (pre-launch and live):
    • Liquidity provision
    • Arbitrage and basis trades
    • Straight lending
    • Market making
  2. For each opportunity:
    • Create a qualitative risk score (team, provenance, backers)
    • Create a quantitative risk score (yield source, sustainability, incentive persistence, utilization breakpoints, probability of wipeout)
  3. Estimate expected yield, compute risk-adjusted yield score.
  4. Size positions using the wipeout risk constraint:
    • complete wipeout implies ≤ ~1% of book loss
  5. Rebalance/allocate across strategies based on where available yield is highest relative to risk.

Trend/momentum directional strategy (conceptual framework)

  • Time-series example:
    • Use a 50-day moving average on BTC
    • Long above it; “out when below” (and the described directional intent includes shorting when below)
  • Cross-sectional example:
    • Each week: rank tokens by 30-day performance
    • Long top 20%, short bottom 20%
  • Fundamental overlay:
    • Avoid rugs/frauds/Ponzi-like assets
    • Filter longs where fundamentals look structurally weak

Presenters / sources mentioned

  • Lee Drogan (CIO), Starkiller Capital
  • Evan (podcast host; name partially shown as “Evan”)
  • Scott Phillips (source discussing fundamental momentum overlay)
  • Shane Coplan (Polymarket founder; historical story)
  • Paul Tudor Jones (mentioned via “Invest Like the Best”)
  • Estimize (speaker-linked company; mentioned as prior work)
  • Superstate (runs the CME basis fund; trade example)
  • Onyx (brand partner hiring Rust developers; not directly part of the finance strategy content)
  • JP Morgan (mentioned regarding launching a stablecoin)

Original video