---
license: cc-by-4.0
pretty_name: BookieBench
language:
- en
- de
- es
- pt
- zh
- fr
- ru
task_categories:
- question-answering
- multiple-choice
- text-classification
tags:
- calibration
- probabilistic-reasoning
- coherence
- dutch-book
- bayesian-inference
- benchmark
size_categories:
- 10K<n<100K
configs:
- config_name: in_family
  data_files:
  - split: dev
    path: data/in_family/dev.parquet
- config_name: in_family_v2
  data_files:
  - split: dev
    path: data/in_family_v2/dev.parquet
- config_name: prior
  data_files:
  - split: dev
    path: data/prior/dev.parquet
- config_name: surface_transfer
  data_files:
  - split: dev
    path: data/surface_transfer/dev.parquet
- config_name: new_mechanics
  data_files:
  - split: dev
    path: data/new_mechanics/dev.parquet
- config_name: realcoh
  data_files:
  - split: dev
    path: data/realcoh/dev.parquet
- config_name: stress_anchoring
  data_files:
  - split: dev
    path: data/stress_anchoring/dev.parquet
- config_name: stress_base_rate
  data_files:
  - split: dev
    path: data/stress_base_rate/dev.parquet
- config_name: stress_conjunction
  data_files:
  - split: dev
    path: data/stress_conjunction/dev.parquet
- config_name: stress_disjunction
  data_files:
  - split: dev
    path: data/stress_disjunction/dev.parquet
- config_name: stress_format_chat
  data_files:
  - split: dev
    path: data/stress_format_chat/dev.parquet
- config_name: stress_format_csv
  data_files:
  - split: dev
    path: data/stress_format_csv/dev.parquet
- config_name: stress_format_email
  data_files:
  - split: dev
    path: data/stress_format_email/dev.parquet
- config_name: stress_format_log
  data_files:
  - split: dev
    path: data/stress_format_log/dev.parquet
- config_name: stress_format_markdown
  data_files:
  - split: dev
    path: data/stress_format_markdown/dev.parquet
- config_name: stress_lang_de
  data_files:
  - split: dev
    path: data/stress_lang_de/dev.parquet
- config_name: stress_lang_es
  data_files:
  - split: dev
    path: data/stress_lang_es/dev.parquet
- config_name: stress_lang_pt
  data_files:
  - split: dev
    path: data/stress_lang_pt/dev.parquet
- config_name: stress_lang_zh
  data_files:
  - split: dev
    path: data/stress_lang_zh/dev.parquet
- config_name: stress_negation
  data_files:
  - split: dev
    path: data/stress_negation/dev.parquet
- config_name: stress_nested
  data_files:
  - split: dev
    path: data/stress_nested/dev.parquet
- config_name: stress_paraphrase
  data_files:
  - split: dev
    path: data/stress_paraphrase/dev.parquet
---

# BookieBench

**Can a model's probabilities be trusted?** BookieBench asks a model for many linked probabilities about one
situation. It then checks four things:
- **coherence**: whether a bookie can Dutch-book the answers, whether they are invariant to option and evidence
  order, and whether updates follow a martingale;
- **calibration** against sampled gold;
- **skill** against exact posteriors, relative to the evidence-blind `uniform_joint` and to train-fitted priors;
- **evidence sensitivity**.

Code, metrics, runners and full documentation: **https://github.com/Mapika/bookiebench**. Scoring requires the `bookiebench` package:

```bash
pip install -e .                # from a clone of the GitHub repo
bookiebench run oracle data/release/test --out results/oracle
bookiebench score results/oracle --data data/release
```

## Licence

- The **generated synthetic data** is released under **CC-BY-4.0** (`LICENSE`). It covers every simulator config
  and the stress text we generated. Copyright 2026 Mark Marosi.
- **realcoh** is real text, and each source keeps **its own licence**; see `metadata/realcoh_NOTICE.md` and
  `metadata/NOTICE.md`. The licences include cc-by-4.0, cc-by-sa-4.0, mit, apache-2.0, gpl-3.0 and cc-by-nc-4.0, and
  some are research-only or undeclared.
- Sources that may not be redistributed ship **ids only**. Each such row has `evidence: null`, a `source_ref`, a
  sha256 of the text, and the questions.

## Configs

All configs have a single split, `dev`. These are the public dev files. Leaderboard editions are regenerated from
secret seeds and are not published here.

| config | rows | what | train data for these families? |
|---|---|---|---|
| `in_family` | 2,100 | 7 v1 simulator families (urn, dice, medical, alarm, sensor, cards, witness) | yes (unseen worlds) |
| `in_family_v2` | 3,900 | 13 v2 pack families (mechanics, programs, tables) | yes (unseen worlds) |
| `prior` | 600 | random Bayes nets and random HMMs | yes (unseen worlds) |
| `surface_transfer` | 1,200 | factory, weather, spam, hiring: re-skinned v1 mechanics | no, but they are not new reasoning |
| `new_mechanics` | 3,300 | genetics, tracking + 9 pack families never in train | **no** (the transfer headline) |
| `realcoh` | 5,000 | 25 real-world sources x 200 states; no exact law (coherence + gold where labelled). `release_mode` = keep (full text) / ids_only (evidence null) | n/a |

Stress transforms of the 11 v1 test/heldout families:

| config | rows | |
|---|---|---|
| `stress_anchoring` | 3,300 | |
| `stress_base_rate` | 3,300 | |
| `stress_conjunction` | 3,300 | |
| `stress_disjunction` | 3,300 | |
| `stress_format_chat` | 3,300 | |
| `stress_format_csv` | 3,300 | |
| `stress_format_email` | 3,300 | |
| `stress_format_log` | 3,300 | |
| `stress_format_markdown` | 3,300 | |
| `stress_lang_de` | 2,156 | |
| `stress_lang_es` | 2,535 | |
| `stress_lang_pt` | 2,451 | |
| `stress_lang_zh` | 2,119 | **BETA** |
| `stress_negation` | 3,300 | |
| `stress_nested` | 3,300 | |
| `stress_paraphrase` | 2,250 | |

Total: 63,911 rows.

LLM stress items (paraphrase, lang_*) are kept only if **both** judges pass every segment. Judge 1 is
Qwen3.8-27B-FP8, the rewriting model, at temperature 0. Judge 2 is deepseek-v41-flash. **`stress_lang_zh` is BETA.**
In Chinese, the judges detect only 0.64 and 0.66 of planted query-negation flips, versus 0.96-1.00 for the other
languages and paraphrase. A rule filter removed 64 more items. Report lang_zh separately, and keep it out of pooled
stress numbers. `lang_hu` and `framing` are excluded from v1.

Compare stressed and clean scores on the same source ids only. The ids are in
`metadata/stress_manifest.json` → `transforms[t][family].kept_source_ids`.

## Format

Each row is one benchmark instance.

| column | contents |
|---|---|
| `instance` | the instance **verbatim as a JSON string** |
| `id`, `family`, `split`, `group`, `level`, `pack`, `transform`, `source_file`, `row` | index fields |
| `n_variables`, `n_steps`, `n_queries` | counts |
| `has_exact_joint`, `evidence_withheld` | flags |
| `source_id` | source instance id (stress) |
| `release_mode` | keep / ids_only (realcoh) |
| `beta` | true for stress_lang_zh |

Instances hold exact posterior joints whose nesting depth equals the number of variables. That varies within a
file, so the instance is stored as a string instead of a nested Arrow type. Parse it with `json.loads`.

```python
from datasets import load_dataset
import json
ds = load_dataset("Mapika/bookiebench", "new_mechanics", split="dev")
inst = json.loads(ds[0]["instance"])
```

Writing `instance` in `row` order for each `source_file`, one per line, reproduces the GitHub release JSONL files
byte for byte. Their sha256 values are in `metadata/CHECKSUMS.sha256`, and this was verified when the dataset was
built. The instance schema (variables, evidence steps with exact joints, queries, gold, perms, next_evidence) is
documented in the GitHub README.

## Not included (regenerable)

| what | how to get it |
|---|---|
| **train split** (22 files, 600k instances) | `bookiebench generate --out data/release --train-only`. It is deterministic, byte-identical to our build. |
| **ids_only realcoh text** | `bookiebench rebuild-realcoh --release data/release/realcoh --out <dir>`. Reads the Hugging Face datasets at the revisions pinned in `metadata/realcoh_manifest.json` and checks hashes. E-mail addresses are masked as `[EMAIL]` in every source. |
| **stress long_4k/8k/16k/32k** (third-party filler) and **scaling** (large, deterministic) | `bookiebench rebuild-stress --long` / `--transforms scaling`. Pinned filler datasets and tokenizer, per-instance sha256 in `metadata/stress_recipe.json`. |

## Known limitations (summary; full list in the GitHub README)

- Some questions can be answered from the question alone. The rate is 0.119 for dice and 0.115 for alarm.
- Some negated conjunctions are double negations.
- `surface_transfer` measures surface transfer, not new mechanics.
- Evidence-blind, train-fitted references reach skill of about 0.16 on `in_family_v2`, so use `skill_prior` there.
- Contamination flags:
  - 10 realcoh sources overlap decider's training mixture;
  - 23 of 25 realcoh sources are likely in pretraining data.
- The judge-based filtering of the LLM stress items makes the kept sets judge-dependent.

## Citation

See `CITATION.cff` in the GitHub repository. Author: Mark Marosi.