"Venture capital funds vintage 2012-2016 have a median net IRR of 16.8% according to Cambridge Associates but 18.2% according to Preqin."
The claim names two specific figures from two specific sources. Neither figure could be found in any publicly accessible document. The sources are both behind subscription paywalls, and no public reproduction of these numbers exists. The claim cannot be verified or falsified from public evidence.
What Was Claimed?
The claim is a concrete, source-attributed benchmark statement: VC funds that made their first investments between 2012 and 2016 have produced a median net internal rate of return of 16.8%, according to Cambridge Associates, and 18.2%, according to Preqin. The specific figures and the specific attributions matter — this is not a general claim about VC performance but a testable assertion about what two named benchmarking services have published.
What Did We Find?
Cambridge Associates is one of the most widely cited VC benchmarking services in the world. It publishes a US VC benchmark report that is widely referenced by LPs, fund managers, and academics. However, the full benchmark — including per-vintage-year median net IRR tables — is available only to subscribers. The publicly available Cambridge Associates materials (including a Q3 2025 summary PDF) contain only aggregate horizon returns over 5, 10, and 20-year periods. After searching more than 30 URLs across Cambridge Associates' website, academic preprint servers (SSRN, NBER), and financial news outlets, no public document reproducing the specific figure of 16.8% for VC vintage 2012-2016 was found.
Preqin, the other benchmarking firm named in the claim, also publishes fund performance data behind a subscription wall. Its public materials — including the annual Global Alternatives Report — cite aggregate venture capital return quartiles but do not break out per-vintage-year median net IRR tables. Searching for "18.2% Preqin venture IRR" across Google, academic databases, and financial press produced no results. Neither figure appears to have been publicly cited by Preqin or reproduced in secondary sources.
The 1.4 percentage-point gap between the two figures (16.8% vs 18.2%) is worth noting. It is plausible: Cambridge Associates and Preqin use different fund universes, different vintage year conventions, and different data collection methodologies. Cambridge Associates relies heavily on voluntary reporting by fund managers; Preqin supplements this with LP filings and freedom-of-information requests to public pension funds. These differences systematically produce divergent returns estimates for the same cohort of funds. So while we cannot verify the specific numbers, their difference is the kind of discrepancy one would expect.
What Should You Keep In Mind?
Proprietary benchmarking databases are the dominant infrastructure for VC performance measurement — and they are not publicly verifiable. Researchers, journalists, and practitioners routinely cite figures from Cambridge Associates and Preqin without being able to provide a publicly accessible URL. This is a structural feature of the asset class: most VC fund data flows through LP filings that are either confidential or only partially disclosed, and the benchmarking firms aggregate this data into products they sell.
This means that many claims about VC fund performance — even accurate ones — will receive an UNDETERMINED verdict under this proof's public-verifiability standard. That is not a criticism of the claim or the underlying data; it is a reflection of the fact that venture capital benchmarking operates in a regime of limited public transparency.
Finally, IRR figures for vintage years like 2012-2016 continue to shift as funds mature and exit positions. A figure published in 2022 will differ from one published in 2025. The claim does not specify the as-of date of the benchmarks, which adds additional verification difficulty even for subscribers who have access to these databases.
How Was This Verified?
This proof conducted targeted searches across Cambridge Associates' and Preqin's public-facing websites and documentation, searched academic preprint servers and financial news outlets for any secondary reproduction of the figures, and applied adversarial checks to all three potential routes to verification (primary sources, secondary sources, and combined citation). You can read the structured proof report for the full evidence summary, review the full verification audit for search query details, or re-run the proof yourself to reproduce the verification process.
What could challenge this verdict?
Check 1 — Are Cambridge Associates benchmark figures publicly available? Searched cambridgeassociates.com/research/ across 30+ URL attempts. The Q3 2025 public PDF contains only aggregate horizon returns. No specific 16.8% vintage 2012-2016 figure found.
Check 2 — Are Preqin benchmark figures publicly available? Searched preqin.com/benchmarks and Preqin annual reports. These cite aggregate quartiles, not the specific 18.2% figure. No match found.
Check 3 — Do both figures appear in any secondary source? Searched for "16.8% Cambridge Associates" and "18.2% Preqin" together for VC vintage 2012-2016. No secondary source reproduces both figures. The 1.4 percentage-point gap is plausible given known methodology differences (timing conventions, fund universe, valuation approaches).
detailed evidence
Evidence Summary
| ID | Fact | Verified |
|---|---|---|
| SC1 | Cambridge Associates: median net IRR for VC vintage 2012-2016 = 16.8% | No (paywall — no public source found) |
| SC2 | Preqin: median net IRR for VC vintage 2012-2016 = 18.2% | No (paywall — no public source found) |
Source: proof.py JSON summary
Proof Logic
SC1 requires Cambridge Associates to have published a figure of 16.8% for the median net IRR of VC funds with vintage years 2012-2016. Cambridge Associates is the most widely cited VC benchmarking service. Their public PDF reports contain only aggregate horizon returns (5/10/20-year), not per-vintage-year median net IRR tables. No public document containing this specific figure was found.
SC2 requires Preqin to have published 18.2% for the same cohort. Preqin's public materials (annual Global Alternatives Report, press releases) cite aggregate VC return quartiles but not this specific vintage-year median. No public source reproduces this figure.
Neither sub-claim can be confirmed or denied from public sources.
Conclusion
UNDETERMINED. Neither sub-claim can be verified from public sources. Cambridge Associates and Preqin are subscription-only benchmarking databases. The figures (16.8% and 18.2%) may be correct but are drawn from proprietary databases not accessible to public verification. To resolve this claim, a researcher would need a paid subscription to Cambridge Associates' US VC benchmark service and Preqin's fund performance database, then query median net IRR for vintage years 2012-2016.
Note: All citations come from unclassified or low-credibility sources. See Source Credibility Assessment in the audit trail.
audit trail
SC1 — Cambridge Associates: median net IRR, VC vintage 2012-2016 = 16.8%
- Data source: Cambridge Associates US VC benchmark database (subscription-only)
- Expected value: 16.8% median net IRR for VC vintage year cohort 2012-2016
- Verification status: Not verified — no public source found
- Search scope:
- cambridgeassociates.com/research/ — benchmark landing page confirms existence of VC benchmarks but requires subscription
- Cambridge Associates Q3 2025 public summary PDF — contains only aggregate 5/10/20-year horizon returns; no per-vintage-year median net IRR table
- SSRN and NBER searches for Cambridge Associates VC benchmark data — no match
- Financial news outlets (Bloomberg, Reuters, WSJ, FT) searches for "Cambridge Associates venture 16.8 IRR" — no match
- Over 30 URL attempts total across CA's domain and secondary sources
- Result: The specific figure 16.8% for VC vintage 2012-2016 cannot be confirmed from any publicly accessible source.
SC2 — Preqin: median net IRR, VC vintage 2012-2016 = 18.2%
- Data source: Preqin fund performance database (subscription-only)
- Expected value: 18.2% median net IRR for VC vintage year cohort 2012-2016
- Verification status: Not verified — no public source found
- Search scope:
- preqin.com/benchmarks — requires login; full benchmark tables behind paywall
- Preqin Global Alternatives Report (annual public PDFs) — cites aggregate venture return quartiles, not specific vintage-year median figures
- Google search: "Preqin venture 2012 2016 IRR 18.2" — no match
- Academic databases (SSRN, NBER, Google Scholar) for "Preqin venture median IRR vintage 2012" — no match
- Financial news outlets search for "Preqin venture 18.2 IRR" — no match
- Result: The specific figure 18.2% for VC vintage 2012-2016 cannot be confirmed from any publicly accessible source.
| Field | Value |
|---|---|
| Subject | Median net IRR of US venture capital funds with vintage years 2012-2016 |
| Property | Median net Internal Rate of Return as reported by two specific benchmarking firms |
| Operator | == (equality — two specific figures from two named sources) |
| Thresholds | SC1: 16.8% (Cambridge Associates); SC2: 18.2% (Preqin) |
| Compound operator | AND — both sub-claims must hold for the claim to be verified |
| Operator note | "Net IRR" = after management fees and carried interest. "Vintage year" = year of first investment. Both Cambridge Associates and Preqin are subscription-only; full benchmark tables are not publicly accessible. Verification requires either a paid subscription, a public press release reproducing the figures, or an academic paper attributing these figures to the named sources. |
The claim asserts two specific empirical figures from two named benchmarking sources: (SC1) Cambridge Associates reports a median net IRR of 16.8% for VC funds with vintage years 2012–2016, and (SC2) Preqin reports 18.2% for the same cohort.
Formal interpretation: Both are equality claims requiring exact figures to be machine-verifiable in publicly accessible sources. "Net IRR" means after management fees and carried interest. "Vintage year" refers to the year a fund made its first investment. Both Cambridge Associates and Preqin maintain subscription-only benchmark databases; the full tables are not publicly accessible.
Formalization scope: The claim is a faithful restatement of two source-attributed figures. The only narrowing is the requirement for public accessibility — if a subscriber to these databases can confirm the figures, the claim may be correct; however, public verification is not possible.
- Rule 1: No numeric values were extracted — the proof could not obtain publicly accessible quote text containing the figures to parse.
parse_number_from_quote()was not called because no verifiable quote was available. - Rule 2: Citation verification was attempted for both SC1 and SC2. Both returned no match across extensive URL search. The sources are behind subscription paywalls.
- Rule 3: Proof generated with date anchored to system time (2026-04-08).
- Rule 4: Claim interpretation explicit in
CLAIM_FORMALdict. Disambiguation of "net IRR," "vintage year," and the compound AND logic of the two sub-claims is documented. - Rule 5: Three adversarial checks performed — CA data availability, Preqin data availability, and secondary source triangulation. All three break the proof.
- Rule 6: No cross-check possible — neither primary source is publicly accessible. The 1.4 pp gap between CA (16.8%) and Preqin (18.2%) is noted as plausible given documented methodology differences.
- Rule 7: All comparisons via
compare()fromcomputations.py.
Cite this proof
Proof Engine. (2026). Claim Verification: “Venture capital funds vintage 2012-2016 have a median net IRR of 16.8% according to Cambridge Associates but 18.2% according to Preqin.” — Undetermined. https://proofengine.info/proofs/venture-capital-funds-vintage-2012-2016-have-a-median-net-irr-of-16-8-according/
Proof Engine. "Claim Verification: “Venture capital funds vintage 2012-2016 have a median net IRR of 16.8% according to Cambridge Associates but 18.2% according to Preqin.” — Undetermined." 2026. https://proofengine.info/proofs/venture-capital-funds-vintage-2012-2016-have-a-median-net-irr-of-16-8-according/.
@misc{proofengine_venture_capital_funds_vintage_2012_2016_have_a_median_net_irr_of_16_8_according,
title = {Claim Verification: “Venture capital funds vintage 2012-2016 have a median net IRR of 16.8\% according to Cambridge Associates but 18.2\% according to Preqin.” — Undetermined},
author = {{Proof Engine}},
year = {2026},
url = {https://proofengine.info/proofs/venture-capital-funds-vintage-2012-2016-have-a-median-net-irr-of-16-8-according/},
note = {Verdict: UNDETERMINED. Generated by proof-engine v1.11.0},
}
TY - DATA TI - Claim Verification: “Venture capital funds vintage 2012-2016 have a median net IRR of 16.8% according to Cambridge Associates but 18.2% according to Preqin.” — Undetermined AU - Proof Engine PY - 2026 UR - https://proofengine.info/proofs/venture-capital-funds-vintage-2012-2016-have-a-median-net-irr-of-16-8-according/ N1 - Verdict: UNDETERMINED. Generated by proof-engine v1.11.0 ER -
View proof source
This is the proof.py that produced the verdict above. Every fact traces to code below. (This proof has not yet been minted to Zenodo; the source here is the working copy from this repository.)
"""
Proof: VC fund vintage 2012-2016 median net IRR — Cambridge 16.8% vs Preqin 18.2%
Claim: Venture capital funds vintage 2012-2016 have a median net IRR of 16.8%
according to Cambridge Associates but 18.2% according to Preqin.
Generated: 2026-04-08
"""
import os
import sys
PROOF_ENGINE_ROOT = os.environ.get("PROOF_ENGINE_ROOT")
if not PROOF_ENGINE_ROOT:
_d = os.path.dirname(os.path.abspath(__file__))
while _d != os.path.dirname(_d):
if os.path.isdir(os.path.join(_d, "proof-engine", "skills", "proof-engine", "scripts")):
PROOF_ENGINE_ROOT = os.path.join(_d, "proof-engine", "skills", "proof-engine")
break
_d = os.path.dirname(_d)
if not PROOF_ENGINE_ROOT:
raise RuntimeError("PROOF_ENGINE_ROOT not set and skill dir not found via walk-up from proof.py")
sys.path.insert(0, PROOF_ENGINE_ROOT)
from scripts.computations import compare, emit_proof_summary
from scripts.verify_citations import verify_all_citations
# =============================================================================
# 1. CLAIM INTERPRETATION (Rule 4)
# =============================================================================
CLAIM_NATURAL = (
"Venture capital funds vintage 2012-2016 have a median net IRR of 16.8% "
"according to Cambridge Associates but 18.2% according to Preqin."
)
CLAIM_FORMAL = {
"subject": "Median net IRR of US venture capital funds with vintage years 2012-2016",
"property": "Median net Internal Rate of Return as reported by two specific benchmarking firms",
"operator": "==",
"operator_note": (
"The claim asserts two specific empirical figures from two specific named sources: "
"(SC1) Cambridge Associates reports a median net IRR of 16.8% for VC vintage 2012-2016, "
"and (SC2) Preqin reports 18.2% for the same cohort. "
"'Net IRR' means after management fees and carried interest. "
"'Vintage year' refers to the year a fund made its first investment. "
"Both Cambridge Associates and Preqin are subscription-only benchmarking databases; "
"their full benchmark tables are not publicly accessible. "
"Verification requires either: (a) a paid subscription to these databases, "
"(b) a publicly available press release or summary report citing these exact figures, "
"or (c) an academic paper reproducing these specific data points with attribution."
),
"threshold": 2, # sub-claims
"compound_operator": "AND",
}
# =============================================================================
# 2. EMPIRICAL FACTS — Sub-claim SC1: Cambridge Associates 16.8%
# =============================================================================
# No verified public source found for this specific figure.
# Searched: Cambridge Associates public benchmark page (cambridgeassociates.com/research/),
# CA public PDF reports (Q3 2025 benchmarks — image-based, no per-vintage tables),
# academic papers citing CA data, news coverage, institutional investor articles.
# All benchmark data is behind a subscription paywall.
empirical_facts_sc1 = {} # No verified sources
n_confirmed_sc1 = 0
sc1_threshold = 1
# =============================================================================
# 3. EMPIRICAL FACTS — Sub-claim SC2: Preqin 18.2%
# =============================================================================
# No verified public source found for this specific figure.
# Searched: Preqin benchmarks landing page (preqin.com/benchmarks),
# Preqin press releases, Alternatives Watch, Institutional Investor,
# Bain and McKinsey PE/VC annual reports citing Preqin figures,
# SSRN academic papers referencing Preqin VC IRR vintage data.
# All Preqin benchmark data is behind a subscription paywall.
empirical_facts_sc2 = {} # No verified sources
n_confirmed_sc2 = 0
sc2_threshold = 1
# =============================================================================
# 4. SUB-CLAIM VERDICTS
# =============================================================================
citation_results = verify_all_citations({}) # No public sources available
sc1_holds = compare(n_confirmed_sc1, ">=", sc1_threshold,
label="SC1: Cambridge 16.8% sources confirmed")
sc2_holds = compare(n_confirmed_sc2, ">=", sc2_threshold,
label="SC2: Preqin 18.2% sources confirmed")
all_sc_hold = sc1_holds and sc2_holds
# =============================================================================
# 5. ADVERSARIAL CHECKS (Rule 5)
# =============================================================================
adversarial_checks = [
{
"description": "Are Cambridge Associates benchmark figures publicly available at all?",
"verification_performed": (
"Searched cambridgeassociates.com/research/ — the benchmark landing page confirms "
"they publish US PE and VC benchmarks but requires a subscription or institutional "
"access. The Q3 2025 public PDF (ca.com) contains only aggregate 5/10/20-year "
"horizon returns, not per-vintage-year median net IRR tables. "
"No public summary report containing '16.8%' for VC vintage 2012-2016 was found "
"in over 30 URL attempts across CA's site, academic databases (SSRN, NBER), "
"and financial news outlets."
),
"breaks_proof": True,
},
{
"description": "Are Preqin benchmark figures publicly available at all?",
"verification_performed": (
"Searched preqin.com/benchmarks — requires login. Searched Preqin press releases "
"and the Preqin Global Alternatives Reports (annual public PDFs): these cite "
"aggregate venture return quartiles but not the specific 18.2% vintage 2012-2016 "
"median net IRR figure. Searched 'Preqin venture 2012 2016 IRR 18.2' in Google "
"and academic databases — no match found. No public source contains this figure."
),
"breaks_proof": True,
},
{
"description": "Do the two figures (16.8% vs 18.2%) appear in any secondary source?",
"verification_performed": (
"Searched for the specific pairing '16.8% Cambridge Associates' and '18.2% Preqin' "
"in financial literature. No secondary source reproduces both figures together for "
"VC vintage 2012-2016. The figures may be correct but originate from proprietary "
"database downloads that are not publicly citable. The 1.4 percentage-point gap "
"is plausible given known methodology differences (CA uses cash-on-cash timing; "
"Preqin uses its own fund database with different fund universe and valuation timing)."
),
"breaks_proof": True,
},
]
any_breaks = any(c["breaks_proof"] for c in adversarial_checks)
# =============================================================================
# 6. VERDICT
# =============================================================================
if any_breaks or not all_sc_hold:
VERDICT = "UNDETERMINED"
else:
VERDICT = "PROVED"
verdict_holds = compare(int(all_sc_hold and not any_breaks), ">=", 1,
label="Overall verdict holds")
# =============================================================================
# 7. FACT REGISTRY
# =============================================================================
FACT_REGISTRY = {
"SC1": {"label": "Cambridge Associates: median net IRR for VC vintage 2012-2016 = 16.8% — no public source found"},
"SC2": {"label": "Preqin: median net IRR for VC vintage 2012-2016 = 18.2% — no public source found"},
}
# =============================================================================
# 8. JSON SUMMARY
# =============================================================================
if __name__ == "__main__":
summary = {
"claim_natural": CLAIM_NATURAL,
"claim_formal": CLAIM_FORMAL,
"fact_registry": FACT_REGISTRY,
"sub_claim_results": {
"sc1": {
"description": "Cambridge Associates reports 16.8% median net IRR for VC vintage 2012-2016",
"n_confirmed": n_confirmed_sc1,
"threshold": sc1_threshold,
"holds": sc1_holds,
},
"sc2": {
"description": "Preqin reports 18.2% median net IRR for VC vintage 2012-2016",
"n_confirmed": n_confirmed_sc2,
"threshold": sc2_threshold,
"holds": sc2_holds,
},
},
"adversarial_checks": adversarial_checks,
"verdict": VERDICT,
"verdict_reason": (
"Both sub-claims rely on proprietary subscription-only databases "
"(Cambridge Associates and Preqin). No publicly accessible source was found "
"that reproduces the specific figures (16.8% and 18.2%) for VC vintage 2012-2016. "
"Verification would require paid access to these benchmarking services."
),
"key_results": {
"sc1_cambridge_confirmed": sc1_holds,
"sc2_preqin_confirmed": sc2_holds,
"claim_holds": all_sc_hold and not any_breaks,
},
"generator": {
"name": "proof-engine",
"version": "1.11.0",
"repo": "https://github.com/yaniv-golan/proof-engine",
"generated_at": "2026-04-08",
},
}
emit_proof_summary(summary)
Re-execute this proof
The verdict above is cached from when this proof was minted. To re-run the exact
proof.py shown in "View proof source" and see the verdict recomputed live,
launch it in your browser — no install required.
Re-execute from GitHub commit 1ba3732 — same bytes shown above.
First run takes longer while Binder builds the container image; subsequent runs are cached.
machine-readable formats
Downloads & raw data
found this useful? ★ star on github