Y Combinator on VCBench
Y Combinator score F0.5 8.6 on VCBench, the venture capital benchmark from the University of Oxford and Vela Research, with 14.0% precision and 6.9% recall on the 4,500-founder private test set. That is rank 32 of 32.
- Rank
- #32 of 32
- F0.5
- 8.6
- Precision
- 14.0%
- Recall
- 6.9%
- Cost / 1k founders
- n/a
Scored on the 4,500-founder private test set, mean over three folds. That is 0.8× the F0.5 of tier-1 VCs. See the full leaderboard.
Compared with the reference rows
| Entry | Precision | Recall | F0.5 | Cost / 1k |
|---|---|---|---|---|
| Y Combinator | 14.0% | 6.9% | 8.6 | n/a |
| Think-Reason-Learn Ensemble | 40.6% | 30.1% | 37.9 | $0.87 |
| Tier-1 VCs | 23.0% | 5.2% | 10.7 | n/a |
| Random Classifier | 9.0% | 9.0% | 9.0 | $0 |
Human rows are normalized to the dataset's 9% success rate, so every entry is compared on the same base rate. Costs use list prices on 2026-09-24.
How the baseline was measured
A human baseline, not a model. Precision and recall come from real investment decisions reported in the VCBench paper: at the inception stage the market index reaches 1.9% precision, Y Combinator 3.2% and tier-1 venture firms 5.6%, with recall near 6% for both. The paper normalizes these to the dataset's 9% success rate, which gives the precision and F0.5 shown here, so humans and models are compared on the same base rate.
Read more
- Introducing VCBench, the First Benchmark for Venture Capital: VCBench, the first benchmark for venture capital, tests LLMs and human investors on predicting founder success across 9,000 anonymized profiles.
Frequently asked questions
- How well do Y Combinator predict founder success on VCBench?
- Y Combinator scores F0.5 8.6 on VCBench, with 14.0% precision and 6.9% recall, rank 32 of 32 as of 2026-10-07.
- How was the Y Combinator baseline measured?
- A human baseline, not a model. Precision and recall come from real investment decisions reported in the VCBench paper: at the inception stage the market index reaches 1.9% precision, Y Combinator 3.2% and tier-1 venture firms 5.6%, with recall near 6% for both. The paper normalizes these to the dataset's 9% success rate, which gives the precision and F0.5 shown here, so humans and models are compared on the same base rate.
- Does AI beat Y Combinator on VCBench?
- Yes. Think-Reason-Learn Ensemble (Vela + Oxford) leads with F0.5 37.9, 4.4 times the Y Combinator score, and most LLM entries also score above it.