The 2021–2025 generation of crypto funds is, on the whole, underwater. The cause wasn't bad luck. It was a mismatch — generalist capital, deployed passively, into a market that had quietly changed its rules. Our framework for timing deployment starts one level up from price: with the monetary regime that governs the whole risk complex.

Stability, not direction

Crypto is correctly understood as a risk asset that responds to monetary policy. But the common version of that idea is incomplete. Analysing every cycle since 2019, the more precise relationship is this: it is not the direction of policy changes that sets the duration and shape of a crypto cycle — it is the stability of the regime.

Hikes, cuts, and balance-sheet shifts introduce volatility; they can accelerate or end a trend. But the sustained uptrends that define bull markets emerge during periods of policy stability — intervals when the Fed holds rates steady and keeps a consistent balance-sheet posture. Direction sets amplitude. Stability sets sustainability.

Markets need roughly four months of stable rates before a durable uptrend resumes. Balance-sheet regime shifts work with a longer, eight-month-plus lag.

Two empirical anchors hold up the framework. First, markets typically need about four months of stable interest rates to resume sustained growth — a brief pause between hikes is not enough. Second, balance-sheet regime shifts operate with a longer lag: transitions between tightening and easing usually need eight or more months before they show up as sustained upside, the time it takes for reserve changes to propagate into risk-asset pricing.

Five regimes

Our quantitative work identifies five recurring regimes, each with a characteristic duration and return profile. The critical observation is uncomfortable: no regime delivers a positive median return at any horizon. Only 20–30% of token launches end up positive. That single fact reorganises how a portfolio must be built — entry-discount discipline, regime-aware timing, and systematic hedging, not stock-picking.

For short-lockup positions, Accumulation Above is the strongest regime — the least-negative median (−35%), the highest hit rate (29%), and by far the best tail. For longer lockups, Accumulation Below offers the softest decay and the best odds of a positive outcome at 545 days. Transitions are patterned too: Accumulation Above rolls into Bull Run ~67% of the time; Bull Run has gone to Bear 100% of the time in our sample; Bear resolves into growth or further accumulation.

Because venture entries precede a token launch by 6–9 months, and full liquidity arrives 12–18 months after it, the question that matters is not "what regime is the market in today?" It is "what regime will our projects launch and unlock into?"

The December 2025 shift

Two policy changes landed in December 2025. Quantitative tightening formally ended on 1 December — the Fed stopped letting bonds roll off unreinvested, closing the balance-sheet reduction programme that ran from mid-2022. Then the 10 December FOMC announced Treasury-bill purchases beginning 12 December, at an elevated initial USD 40bn per month, with maturing MBS (~USD 16bn/mo) reinvested into bills.

Together these mark a regime change: from active balance-sheet reduction to net accumulation. But the eight-month-plus lag means the effect on risk pricing is not immediate. The market is unlikely to enter a sustained growth phase before the second half of 2026.

The 2026–2028 window

We assess that the market entered a Bear Market regime in Q4 2025, likely overlapping with early accumulation as smart money positions for the next cycle. Based on the December shift and historical lags, the next stability window should open in the autumn of 2026 — defining the deployment window the fund is built around.

This is not a call on any single token. It is a regime read — the kind that decides whether a well-executed launch lands into a tailwind or a headwind. Within the window, project-level timing still matters, which is why we deploy into the specific regimes that have historically been kinder to new launches, and hedge every position at the portfolio level.

Structural tailwinds

Independent of precise timing, the ecosystem keeps scaling along three dimensions — adoption, capital formation, and financialisation. These are secular, not cyclical: the foundational demand against which the timing thesis operates.

What it means for the fund

The macro framework drives the structure. A 3-year fund life (versus the traditional 7–10) is calibrated to the cycle: a 2-year deployment period spanning accumulation and early expansion, and a 1-year harvest capturing the mature phase. 70% to Liquid Alpha keeps the majority of capital on a recycling cadence that can catch dislocations throughout the expansion; 30% to Venture Select targets the highest-conviction pre-launch positions when entry valuations are most compressed and our operational involvement matters most.

Capital is called in four tranches, each triggered by 50% deployment of the prior one — pacing flexibility for the GP, walk-away optionality for LPs. The cycle sets the tailwind. The work decides who catches it.

The DAT universe

Digital Asset Treasury Companies emerged after MicroStrategy's 2020 Bitcoin treasury strategy. These entities use public equity markets to accumulate digital assets, offering what proponents call "levered exposure" through regulated securities. The universe has grown from fewer than 10 companies in 2021 to 190+ by April 2026. We track detailed metrics for a cohort of 159 with sufficient data.

We classify DATs into three categories by underlying business model: Pure Treasury (41 — accumulation is the whole business), Web3-Native (45 — mining, exchanges, infrastructure, with treasury as a secondary activity), and Non-Web3 Native (66 — traditional operating businesses that adopted a treasury strategy).

Premium vs. discount

The market-to-NAV ratio (mNAV) measures whether equity trades above or below the value of underlying holdings. Above 1.0x is a premium; below is a discount.

Of companies with available data (n=149), 92 (61.7%) trade at a premium and 57 (38.3%) at a discount. But the aggregate hides everything — the split by company type is the story.

A notable subset of Pure Treasury DATs trade at severe discounts — equity worth a fraction of the assets it holds. These are not rounding errors; the market is pricing in illiquid holdings, dilution overhang, governance concerns, or anticipated forced selling.

Post-announcement performance

To assess investor outcomes, we tracked 41 companies with identifiable DAT announcement dates between December 2024 and November 2025. The pattern is consistent and unkind.

Pure Treasury companies generate the highest average pump (+134%) but land at returns indistinguishable from the rest. Treasury-only narratives create short-term excitement; they don't sustain value.

Extreme outcomes

The largest drawdowns follow a textbook pump-and-dump shape — triple-digit pumps that round-trip almost entirely.

Only three companies from the announcement cohort remain positive — and they share the same DNA.

The feedback loop

DAT valuations are reflexive. In the bull phase a positive loop compounds; in the bear phase the same mechanics run in reverse — and faster.

The method of capital formation is the single clearest signal of which loop a company is in.

The DAT wave

Announcements followed a distinct temporal arc, peaking in mid-2025 — and the timing of entry mattered, but not in the way most expected.

A diagnostic framework

The central question is not "does the company hold crypto?" It is "what is the structural path to sustained value creation?" The most reliable single indicator is the trajectory of treasury per share after a financing event — if it rises, the structure may be accretive; if it falls, shareholders are funding dilution.

Classification. ≥5 positive indicators → Strategic DAT, proceed with standard diligence. 3–4 → Transitional, monitor but don't chase momentum. ≤2 → Liquidity-driven, high probability of value extraction.

Conclusions

The DAT structure is a mechanism, not a strategy. Its efficacy depends entirely on the underlying business model, capital-formation discipline, and sustainable demand for equity.

Methodology. Data from CoinGecko (treasury valuations, mNAV), DeFiLlama (holdings verification), AlphaGrowth (universe identification), SEC EDGAR (filings), and company IR. Original data as of January 2026, updated with April 2026 prices. Announcement tracking covers Dec 2024 – Nov 2025. Limitations: treasury holdings may be self-reported; mNAV assumes full treasury accessibility; not all derivative instruments are captured.

Combined AI-crypto market capitalisation rose roughly 350% in 2024, and venture funding into AI-focused Web3 startups hit records through 2025 — surpassing all of 2024's total in the first eight months. Then came the excess. The AI-agent token sub-sector lost about half its value from peak, and an estimated 90% of AI-associated tokens have decayed to near-zero. The sector is now in its differentiation phase: genuine utility separating from narrative.

The correction did the useful thing. It cleared out the speculative layer and pulled valuations down into the range our strategy is built for.

Why they converge

The convergence is structural, not narrative-driven — each technology fixes a fundamental limit of the other.

Web3 fixes AI's trust and concentration problems. AI development is bottlenecked by compute and data, resources controlled by a handful of companies. Blockchains enable distributed compute and data marketplaces that reduce dependence on those providers — and provide mechanisms to verify and audit models and data provenance, addressing the opacity that erodes trust.

AI fixes Web3's usability and efficiency problems. On-chain interaction is hard; AI can power intelligent interfaces, automated strategy execution, personalised DeFi, and better security monitoring. The result is symbiotic: AI brings intelligence and automation; blockchain brings the verification, ownership, and incentive layers AI needs to operate trustlessly.

Where value is forming

The intersection spans several sub-sectors at very different maturities. We rate each on relevance to our two sleeves — Venture Select (pre-launch) and Liquid Alpha (listed, hedged).

Lessons from 2024–25

The hype cycle taught the sector — and us — four things that now shape how we deploy.

Two conclusions follow directly. Narrative alone does not create value — market cap accumulated rapidly around a story and collapsed just as fast without product. And hedging is non-negotiable: a sub-sector that lost ~50% from peak is subject to the same VWAP-anchored dynamic hedge as every other position in the fund.

What it means for the fund

AI × Web3 will be a meaningful share of deployment across both sleeves — but it is not a thematic concentration. The regime-aware, valuation-disciplined approach applies identically here; sector enthusiasm does not override entry discipline.

In Venture Select, we target pre-launch AI-infrastructure and AI-enhanced DeFi where we have strategic proximity and can add operational value post-investment — with entry FDV guardrails (below USD 30M) and structural requirements (minimum TGE unlock, maximum lockup, hedging precondition) applied without exception. In Liquid Alpha, post-correction valuations in listed tokens create OTC and secondary opportunities with established liquidity and perpetual-futures availability for hedging — well-suited to the 3–6 month recycling cadence.

Every AI × Web3 project gets extra scrutiny on three axes: genuine decentralisation, sustainable tokenomics, and real user adoption.

Outlook

The sector is moving from its speculative phase into its utility phase. The projects that matter in the next cycle will be the ones with working infrastructure, actual revenue, and actual users — not the ones built to capture a narrative. The structural drivers — decentralised compute, trustworthy data provenance, autonomous on-chain execution — are secular, not cyclical. Capital put into the strongest projects at today's compressed valuations is well placed as those trends mature through 2026–2028 and beyond.