What Kind of VCs Can Get Money from Fund of Funds? We Have the Answer After Reviewing 2000

marsbitDipublikasikan tanggal 2026-04-11Terakhir diperbarui pada 2026-04-11

Abstrak

Moses Capital, a fund of funds focused on early-stage VCs, reviewed over 2,000 funds over two years and invested in 46, a 2.3% selection rate. The market is larger than perceived, with many new funds remaining invisible to most LPs. They identified four GP archetypes: founder-turned-investor, spin-out from established VC firms, community-native managers, and quiet technical experts. The top reasons for rejecting 97% of funds included team inexperience (30%), poor portfolio construction (25%), weak track record (20%), strategy misalignment (15%), and fundraising challenges (10%). The highest-quality deal flow source emerged unexpectedly from blind founder reference calls during due diligence, where consistently praised investors became top targets. Building a reputation through prepared, respectful engagement led to strong GP referrals and trust within the VC ecosystem.

Author: Moses Capital & Lev Leviev

Compiled by: Deep Tide TechFlow

Introduction: Moses Capital is a Fund of Funds (FoF) focused on early-stage VCs. Over the past two years, they have reviewed more than 2,000 funds and ultimately invested in only 46, with an approval rate of 2.3%. This article reviews the four archetypes of GPs they discovered during the screening process, the specific reasons for the 97% rejection rate, and an unexpected due diligence method that became the highest-quality source of deal flow. For readers interested in the VC ecosystem and the LP perspective, the information density is very high.

When I founded Moses Capital, I thought I had a general understanding of the market for emerging fund managers. A few hundred funds, concentrated in a few common cities, and all you needed to know was where to look.

That assumption lasted about three months.

Over the past two years, we have reviewed more than 2,000 funds for Fund I. We conducted 553 preliminary calls, completed 276 full due diligence processes, and ultimately added 46 funds to our portfolio—an approval rate of 2.3%. When you sit through that many conversations, patterns naturally emerge.

Here’s what we learned.

This Market Is Larger Than Anyone Thought

Before we built a systematic sourcing process, our deal flow was like most FoFs: relying on networks and inbound referrals. VCs refer other VCs. This approach works, but it also means your perspective is limited by "who knows you."

When we started scraping SEC filing data in real-time, the picture changed completely. Dozens of new funds are launched every week, many of which don’t appear on anyone’s radar until months later—by which time they are already fundraising. By 2025, we covered about 95% of U.S. VC funds. The sheer number of new funds surprised even us.

The key point: most of these funds are invisible to the majority of LPs. Not because they are bad, but because they are too early-stage, too small, and haven’t built the network that gets you on shortlists. This is precisely the gap we aim to fill.

Four Archetypes of GPs

After 553 preliminary calls, patterns began to emerge. We broadly categorized the managers we met into four types:

  1. Entrepreneurs Turned Investors

Former founders or former operating executives, usually with one notable exit, who then decide to start a fund. They have credibility among founders and strong deal flow in their niche. The challenge is that managing a fund and managing a company are two completely different things—portfolio construction, follow-on investment strategies, post-investment management—many learn on the job. Some pick it up quickly, but more only truly get it by Fund II or Fund III.

  1. VC Spin-Outs

Former partners or principals from established funds (tier-one or tier-two) who go out on their own. They have brand recognition, track records to show, and usually strong networks. What we primarily look at: how much of that track record is theirs, and how much is the platform’s? After leaving a large fund, do they still remain competitive among founders?

  1. Community-Native Managers

A type that has clearly increased since 2020—managers who have built their reputation through community building, writing articles, hosting podcasts, and managing social media. They have inbound deal flow, visibility, and usually a real community moat.

Within this category, there are actually two subtypes: one is investors who built a community first, using it to drive deal flow and create network value for portfolio companies; the other is community operators who started investing because deal flow naturally came to them. The distinction between these two is important. For both, we look at two things—the quality of their investment discipline, and whether the community creates real value for the founders they want to back.

  1. Quiet Technical Experts

This is usually my personal favorite type. The GP has deep technical or industry expertise in a specific field, honed over many years. They are the people founders turn to for advice when facing problems, and over time, more and more founders want them on their cap table early—not for the brand, but to help build the business from day one.

These individuals deliberately stay low-profile, building their reputation on expertise and accumulated relationships. They almost never reach out to us proactively. We find them through systematic external searches or, more commonly, through founder references during due diligence on other funds. We ask every founder: among your investors, who provided the most help? The answer is often this type of person.

What the 97% Rejection Looks Like

We rejected over 97% of the funds we reviewed. Each pass decision was made as carefully as an investment decision, and this process was refined with every fund we examined.

  • About 30% of rejections were related to the GP or team. Insufficient fund management experience, lack of clear differentiation from existing players, or networks that don’t translate into unique deal sourcing capabilities.
  • About 25% failed on portfolio construction. Too much exposure to later stages, lack of discipline in follow-on strategies, insufficient target ownership, or over-diversification—mathematically killing the possibility of power law returns. If a fund isn’t designed to generate concentrated big winners, it probably won’t.
  • About 20% were due to track record issues. Investment history too weak or insufficient, or a track record that doesn’t match the current strategy (different geographies, sectors, stages, check sizes).
  • About 15% were due to strategy mismatch. The fund’s current strategy doesn’t align with our investment themes, unrelated to performance—fund size too large, investment scope too broad, or involvement in areas or regions we deliberately avoid.
  • The remaining 10% were due to factors like fundraising dynamics. If a manager can’t raise money, they can’t execute their strategy.

The Best Sourcing Channel We Never Planned

Our sourcing evolved in stages. Initially, it relied on networks and inbound referrals. Then we built a systematic outbound engine that scrapes every new U.S. fund in real-time, automatically filtering by size, strategy, and GP background. At its peak, this channel accounted for 70% of our meetings. We could reach managers before most LPs even knew the fund existed.

But the sourcing channel that ultimately proved most valuable wasn’t one we designed. It came from our due diligence process itself.

For every GP, we conduct blind founder reference calls—sometimes up to 10 if the track record allows. In these calls, we don’t just ask about the manager we’re evaluating. We go through the cap table, asking founders for honest feedback on their other early investors. The names that come up repeatedly become our next targets for outreach.

This proved to be our highest-quality source of deal flow.

Building a Reputation

Moses Capital’s reputation initially spread through our investments and the relationships built around them. Now we receive many proactive inquiries from GPs who heard about us through the VC ecosystem. We strive to be worthy of that trust.

We are not anchor LPs, we don’t sit on LPACs, and our checks aren’t large. But we do our homework. Before engaging with a GP, we usually have been tracking them for a while—monitoring their online presence, conducting references, and forming our own judgments. Our questions are prepared. We understand how fund economics work. We don’t disturb managers unnecessarily. If a fund isn’t right for us, we say so directly and explain why.

Managers appreciate this, and as a result, they refer other managers to us.

What We’ve Learned Over Two Years

Two years, 2,000 funds. We have a deeper understanding of this market and the people behind it. Every type of manager has the right to win—the key is knowing what to look for. This is an ongoing learning process, relying on our ability to see a broad enough funnel and our continuously improving dynamic sourcing mechanism.

Pertanyaan Terkait

QWhat are the four archetypes of GPs identified by Moses Capital after reviewing over 2000 funds?

AThe four archetypes are: 1. Founder/Operator Turned Investor, 2. Spin-Outs from Established VC Firms, 3. Community-Native Managers, and 4. Quiet Technologists.

QWhat was the single most effective sourcing channel for high-quality deal flow, which was not originally planned?

AThe most effective sourcing channel was conducting blind founder reference calls during their due diligence process. They would ask founders about other investors on their cap table who provided the most help, and those repeatedly mentioned became their next targets.

QWhat was the overall pass rate for funds reviewed by Moses Capital?

AThe overall pass rate was 2.3%. They reviewed over 2000 funds and ultimately invested in only 46 of them.

QWhat was the primary reason for approximately 30% of the fund rejections?

AApproximately 30% of rejections were related to the GP or team, such as insufficient fund operating experience, lack of a clear differentiation from existing players, or a network that could not be converted into unique deal access.

QHow did Moses Capital's perspective on the emerging manager market change after building a systematic sourcing process?

AThey discovered the market was much larger than anyone thought. By systematically scraping SEC filing data, they found dozens of new funds being formed weekly, most of which were invisible to other LPs because they were too early, too small, and lacked the network to get on shortlists.

Bacaan Terkait

Hari Pertama Peluncuran Langsung Terganggu 'Perselisihan Internal', Platform Peluncuran Robinhood Chain vlad.fun Dicurangi oleh Orang Dalam

**vlad.fun, Platform Peluncuran Token di Robinhood Chain, Alami 'Perselisihan Internal' di Hari Pertama** Pada 15 Juli, platform peluncuran token vlad.fun di Robinhood Chain mengalami insiden operasi jahat oleh pengembang eksternal tepat di hari peluncurannya. Menurut pernyataan resmi, dua pengembang yang terlibat dalam pembuatan platform secara diam-diam membuat token mereka sendiri dan menyandikannya ke dalam kode front-end. Hal ini menyebabkan pengguna hanya dapat melihat token tersebut saat platform diluncurkan, menyembunyikan semua token pembuat lainnya. Tim menemukan anomali dalam dua jam, kemudian memecat pengembang yang terlibat dan menghapus kode berbahaya. Kedua pengembang telah mengembalikan biaya pembuat sekitar $15.000 dan keuntungan dari token uji coba, dengan total sekitar 7,8 ETH yang dipulihkan. Biaya protokol sekitar 4,16 ETH dari peluncuran disimpan di alamat multi-tanda tangan tim. Platform vlad.fun, yang dibangun oleh tim kecil 5 orang dalam 48 jam, saat ini berstatus dihentikan sementara. Tim berjanji memprioritaskan penanganan masalah pengguna yang terdampak. Insiden ini menyoroti risiko kepercayaan dalam proyek crypto yang dibangun cepat dengan melibatkan pengembang eksternal. Salah satu anggota komunitas, Will Mexi, dan anggota tim inti @SOLsesame, mengklarifikasi bahwa mereka bukanlah pelaku dan telah lama berkolaborasi dalam ekosistem Solana. Meskipun vlad.fun terpaksa berhenti, ekosistem platform peluncuran di Robinhood Chain tetap aktif. Data DefiLlama menunjukkan TVL rantai sekitar $258 juta, dengan platform seperti PONS memimpin pasar dan token platformnya melonjak lebih dari 4200% dalam 7 hari terakhir.

Foresight News6m yang lalu

Hari Pertama Peluncuran Langsung Terganggu 'Perselisihan Internal', Platform Peluncuran Robinhood Chain vlad.fun Dicurangi oleh Orang Dalam

Foresight News6m yang lalu

Dari Emas ke Bitcoin: Pasokan Tetap + Demam Institusi, Akankah Ulangi 'Ledakan' Harga?

Dari tahun 2004 hingga 2026, dana ETF emas telah menunjukkan siklus kenaikan tajam, penurunan yang menyakitkan, dan pemulihan panjang yang pada akhirnya berujung pada rekor tertinggi baru. Analis ETF Bloomberg Intelligence, Eric Balchunas, melihat bahwa dana ETF Bitcoin yang diluncurkan pada 2024 mungkin mengikuti 'peta jalan' serupa. Baik emas maupun Bitcoin adalah alat penyimpan nilai tanpa hasil, yang harganya digerakkan sepenuhnya oleh sentimen investor dan permintaan institusional, bukan arus kas atau pendapatan. Kehadiran ETF Bitcoin arus utama telah membuka akses investasi yang lebih mudah, mendorong masuknya modal institusional. Namun, volatilitas tetap tinggi. Seperti yang ditunjukkan oleh penjualan besar Bitcoin oleh dana terkemuka BlackRock IBIT untuk memenuhi penebusan, aliran dana institusional dapat menyebabkan reaksi pasar yang signifikan. Akan tetapi, minat institusi yang berkelanjutan dan masuknya dana ke ETF dianggap sebagai penopang kokoh yang membantu menahan tekanan jual. Dengan pasokan yang hampir tetap (mirip dengan emas), ledakan permintaan yang terkonsentrasi berpotensi memicu kenaikan harga eksplosif pada Bitcoin. Meskipun perjalanannya pasti akan diwarnai fluktuasi yang hebat, banyak yang optimis tentang potensi jangka panjang Bitcoin sebagai 'emas digital'. Jika Bitcoin dapat mencapai bahkan sebagian kecil dari kapitalisasi pasar emas sebesar $28 triliun, ruang apresiasinya tetap sangat besar. Bagi investor, kuncinya adalah fokus pada tren jangka panjang, diversifikasi risiko, dan memahami sifat pasar yang berisiko tinggi dan berpotensi menghasilkan imbal hasil tinggi. Kisah Bitcoin masih jauh dari selesai.

marsbit13m yang lalu

Dari Emas ke Bitcoin: Pasokan Tetap + Demam Institusi, Akankah Ulangi 'Ledakan' Harga?

marsbit13m yang lalu

Model Besar Tak Lagi Hanya 'Berbicara Saja' Saat Memberi Skor pada Gambar, Gunakan Peta Struktur, Spektrum Frekuensi sebagai 'Bukti Fisik', dan 'Bukti Visual' untuk Memberi Skor pada Gambar

Model multi-modal besar (MLLM) sering gagal menilai kualitas gambar secara akurat karena mengandalkan bias semantik internal, bukan bukti degradasi visual yang terukur. Untuk mengatasi ini, tim peneliti dari **Northwestern Polytechnical University** dan **Hong Kong University of Science and Technology** memperkenalkan **IQA-T1**, sebuah framework baru yang memungkinkan MLLM menilai kualitas gambar dengan **bukti visual terstruktur**. IQA-T1 dilengkapi dengan **toolbox analisis** yang berisi alat-alat seperti *Noise Residual Map*, *Fourier Magnitude Spectrum*, dan *Gradient Orientation Coherence Map*. Alat-alat ini menghasilkan "bukti visual" yang memvisualisasikan noise, artefak, blur, dan degradasi lainnya. Model kemudian belajar **memanggil alat-alat ini secara adaptif** dan melakukan penalaran langkah demi langkah berdasarkan bukti-bukti tersebut. Framework ini dilatih dalam dua tahap: **Supervised Fine-Tuning (SFT)** untuk mempelajari cara menggunakan alat dan template penalaran, dan **Reinforcement Learning (RL)** dengan GRPO untuk mengoptimalkan strategi pemanggilan alat yang efisien. Tim juga membuat dataset **Q-Tool**, dataset IQA pertama yang dilengkapi rantai penalaran multimodal berbasis bukti. Dalam evaluasi pada 7 benchmark IQA (termasuk gambar dengan distorsi nyata, sintetis, dan hasil AI), IQA-T1 mencapai kinerja **terbaik secara keseluruhan (SOTA)** dengan skor PLCC/SRCC rata-rata **0.795/0.784**. Ia mengungguli metode MLLM lain sekaligus menyediakan proses penilaian yang **dapat dijelaskan dan dilacak**. Rata-rata, model hanya menggunakan **2.34 alat per gambar**, membuktikan efisiensi dan efektivitas pendekatan berbasis bukti ini.

marsbit17m yang lalu

Model Besar Tak Lagi Hanya 'Berbicara Saja' Saat Memberi Skor pada Gambar, Gunakan Peta Struktur, Spektrum Frekuensi sebagai 'Bukti Fisik', dan 'Bukti Visual' untuk Memberi Skor pada Gambar

marsbit17m yang lalu

Trading

Spot
活动图片