Datasetpublic

huyxdang/adaption-market-analysis-3

This dataset is a remastered version prepared using Adaption's Adaptive Data platform. adaption-market-analysis-3 This dataset contains prompt/completion pairs for updating dated market-research notes using only supplied evidence. Each entry includes specific financial data points, such as historical actuals or analyst forecasts, and a corresponding response that strictly separates observed results from forward-looking claims. The completions emphasize causal limitations… See the full description on the dataset page: https://huggingface.co/datasets/huyxdang/adaption-market-analysis-3.

sourceHugging Faceupdated 2mo agoView on Hugging Face

0likes21downloads

Dataset Card

This dataset is a remastered version prepared using Adaption's Adaptive Data platform.

adaption-market-analysis-3

This dataset contains prompt/completion pairs for updating dated market-research notes using only supplied evidence. Each entry includes specific financial data points, such as historical actuals or analyst forecasts, and a corresponding response that strictly separates observed results from forward-looking claims. The completions emphasize causal limitations, preserve counterevidence, and avoid providing executable trade instructions or personalized advice.

Dataset size

There are 15,197 data points in this dataset. This is an instruction tuning dataset.

Adaptive Data recipe

Prompt and completion both rewritten, with response_constraints. A second generation under the same recipe as dataset 2, giving an independent variant of every source item.

Asset family

—companyandindustry (71.1%)
—fundsandetfs (14.2%)
—macro_synthesis (8.2%)
—fixedincomecash (3.8%)
—real_estate (2.7%)

Task family

—invalidatedorinsufficient_evidence (37.3%)
—marketstateinterpretation (30.6%)
—actualvsforward_looking (17.0%)
—marketdrivertransmission (13.0%)
—comparativemarketanalysis (1.4%)
—fundandbenchmark_analysis (0.4%)
—macromarkettransmission (0.1%)
—realestatemarket_analysis (0.1%)

Evidence relationship

—causal (31.4%)
—scenario (30.9%)
—valuation (23.3%)
—comparative (7.2%)
—countervailing (5.3%)
—temporal_update (1.0%)
—hierarchical (0.8%)
—expectation_gap (0.1%)

Language

—English (100%)

Coverage

—As-of dates: 2010-12-31 to 2026-12-31
—Leakage groups: 1,753
—Rows carrying an automated review flag: 2,794 (18.4%)

Schema

Field	Description
`prompt`	Source instruction with system framing, as-of date, and evidence block
`completion`	Source analyst-style response
`enhanced_prompt`	Adaptive Data rewrite of the prompt
`enhanced_completion`	Adaptive Data rewrite of the completion
`evidence`	Evidence spans supplied to the model
`original_context`	Source passage the row was derived from
`question`	The analytical ask, derived from the response's own content
`reasoning`	Reasoning trace for the expected answer
`as_of_date`	Report date the note is anchored to
`asset_family`	Asset class of the subject
`task_family`	Analytical task type
`relationship_type`	Evidence-to-conclusion relationship
`leakage_group`	Grouping key so related rows stay on one side of a split
`relaxed_quality_flags`	Automated review flags
`record_id`	Stable row identifier
`response_constraints`	Explicit constraints the response must satisfy

Related datasets

One of three Adaptive Data generations combined into huyxdang/adaption-market-analysis-final.