BondFoundry/bondfoundry-quant-sample

BondFoundry定量样本数据集 — 10条记录

这是一个为构建领域特定定量金融模型的企业AI团队提供的高级合成指令调优数据集。包含来自BondFoundry定量金融目录中深度最高的10条记录，每条记录都由资深量化投资经理级别的人物在真实机构约束下生成，如追加保证金压力、风险委员会反对意见、监管期限和投资组合缩水情景等。

平均每条记录568字

样本主题包括：
- 缩水压力下的因子模型构建
- 信号冲突时的风险委员会演示
- 市场压力下的监管资本要求
- 流动性约束下的投资组合再平衡
- 具有交易对手风险敞口的衍生品定价

质量认证：
每条记录都通过BondFoundry的三阶段QA流程：
1. 技术筛选 — 字数、标记污染、领域元数据
2. 深度评分 — 推理深度评分1-10，通过/标记/拒绝
3. 资深量化PM同行评审 — 领域专家评估

完整数据集：
quant-research-instruct-v1 — 978+条记录，每晚更新

BondFoundry Quant Sample — 10 Records

Premium synthetic instruction-tuning data for enterprise AI teams building domain-specific quantitative finance models.

Whats inside

10 of the highest-depth records from BondFoundrys quantitative finance catalogue. Each record is generated by a senior quant PM-level persona operating under real institutional constraints — margin call pressure, risk committee pushback, regulatory deadlines, portfolio drawdown scenarios.

Average word count: 568 words per record

Sample record topics include

Factor model construction under drawdown pressure
Risk committee presentation with conflicting signals
Regulatory capital requirements during market stress
Portfolio rebalancing under liquidity constraints
Derivatives pricing with counterparty risk exposure

Quality certification

Every record passes BondFoundrys three-stage QA pipeline:

Stage 1 — Technical filter — word count, markdown contamination, domain metadata
Stage 2 — Depth scorer — reasoning depth scored 1–10, PASS/FLAG/REJECT
Stage 3 — Senior quant PM peer review — domain expert persona evaluation

Full dataset

quant-research-instruct-v1 — 978+ records, growing nightly