---
pretty_name: YCbench
dataset_name: ycbench
language:
  - en
license: mit
tags:
  - finance
  - startups
  - ycombinator
  - benchmarking
  - tabular
  - prediction
  - venture-capital
task_categories:
  - tabular-classification
  - text-classification
size_categories:
  - n<1K
configs:
  - config_name: startups
    data_files: yc_w26_startups.csv
  - config_name: traction
    data_files: yc_w26_traction.csv
  - config_name: scores
    data_files: yc_w26_pre_demo_scores.csv
  - config_name: mentions
    data_files: yc_mentions.csv
  - config_name: mentions_early
    data_files: yc_mentions_early.csv
---

# YCbench

**A live benchmark dataset for forecasting startup outperformance in Y Combinator batches.**

This dataset was introduced in the paper:  
**["YC Bench: a Live Benchmark for Forecasting Startup Outperformance in Y Combinator Batches"](https://huggingface.co/papers/2604.02378)** by Mostapha Benhenda.

## Dataset Overview

YCbench provides structured public signals for Y Combinator startups (focused on the **W26** batch of 196 companies). It enables rapid evaluation of models that predict which startups will **outperform their batch peers** in the short term (until Demo Day).

Performance is measured using a **Pre-Demo Day Score** that combines traction signals and web visibility.

### Available Configurations

| Config          | File                          | Description                              |
|-----------------|-------------------------------|------------------------------------------|
| `startups`      | `yc_w26_startups.csv`         | Basic startup information                |
| `traction`      | `yc_w26_traction.csv`         | Traction metrics                         |
| `scores`        | `yc_w26_pre_demo_scores.csv`  | Pre-demo day scores + velocity           |
| `mentions`      | `yc_mentions.csv`             | Google/web mention counts                |
| `mentions_early`| `yc_mentions_early.csv`       | Early-stage mention data                 |

## Quick Load

```python
from datasets import load_dataset

# Load specific parts
startups = load_dataset("benstaf/ycbench", "startups")
scores   = load_dataset("benstaf/ycbench", "scores")
mentions = load_dataset("benstaf/ycbench", "mentions")
```

Or with pandas (simple way):
```
import pandas as pd
df = pd.read_csv("hf://datasets/benstaf/ycbench/yc_w26_pre_demo_scores.csv")
```

## Links

- **Paper**: [arXiv 2604.02378](https://huggingface.co/papers/2604.02378)
- **Live Benchmark**: [ycbench.com](https://ycbench.com/)
- **GitHub**: [benstaf/ycbench](https://github.com/benstaf/ycbench)