# What Predicts Correctness in Text-to-SQL? A Selective-Prediction Study

arXiv: https://arxiv.org/abs/2607.06799 · HTML: https://mlsanity.com/paper/2607.06799/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- Placed in Knowledge- and tool-augmented reasoning, one of 17 directions in LLM reasoning.
- Builds on 16 of its 17 references that are in the map.
- First posted on arXiv 7 Jul 2026.

## Placed in

- LLM reasoning / [Knowledge- and tool-augmented reasoning](https://mlsanity.com/lr/knowledge-and-tool-augmented-reasoning/)

## Builds on (most cited first, 16 of 16)

- [Self-Consistency Improves Chain of Thought Reasoning in Language Models](https://mlsanity.com/paper/2203.11171/) (2022)
- [On Calibration of Modern Neural Networks](https://mlsanity.com/paper/1706.04599/) (2017)
- [Reflexion: Language Agents with Verbal Reinforcement Learning](https://mlsanity.com/paper/2303.11366/) (2023)
- [Spider: A Large-Scale Human-Labeled Dataset for Complex and Cross-Domain Semantic Parsing and Text-to-SQL Task](https://mlsanity.com/paper/1809.08887/) (2018)
- [Teaching Large Language Models to Self-Debug](https://mlsanity.com/paper/2304.05128/) (2023)
- [Selective Classification for Deep Neural Networks](https://mlsanity.com/paper/1705.08500/) (2017)
- [Semantic Uncertainty: Linguistic Invariances for Uncertainty Estimation in Natural Language Generation](https://mlsanity.com/paper/2302.09664/) (2023)
- [Can LLM Already Serve as A Database Interface? A BIg Bench for Large-Scale Database Grounded Text-to-SQLs](https://mlsanity.com/paper/2305.03111/) (2023)
- [DIN-SQL: Decomposed In-Context Learning of Text-to-SQL with Self-Correction](https://mlsanity.com/paper/2304.11015/) (2023)
- [RAT-SQL: Relation-Aware Schema Encoding and Linking for Text-to-SQL Parsers](https://mlsanity.com/paper/1911.04942/) (2019)
- [Conformal Risk Control](https://mlsanity.com/paper/2208.02814/) (2022)
- [Learn then Test: Calibrating Predictive Algorithms to Achieve Risk Control](https://mlsanity.com/paper/2110.01052/) (2021)
- [Spider 2.0: Evaluating Language Models on Real-World Enterprise Text-to-SQL Workflows](https://mlsanity.com/paper/2411.07763/) (2024)
- [MAC-SQL: A Multi-Agent Collaborative Framework for Text-to-SQL](https://mlsanity.com/paper/2312.11242/) (2023)
- [LGESQL: Line Graph Enhanced Text-to-SQL Model with Mixed Local and Non-Local Relations](https://mlsanity.com/paper/2106.01093/) (2021)
- [Reliable Text-to-SQL with Adaptive Abstention](https://mlsanity.com/paper/2501.10858/) (2025)

## Built on by (most cited first, 2 of 2)

- [Certified Against Which Oracle? Execution Labels Set the Reported Risk of Conformal Abstention for Text-to-SQL](https://mlsanity.com/paper/2609.25938/) (2026)
- [Does a Governed Confidence Score Know When a Text-to-SQL Answer Is Right? An Exploratory Study on BIRD and a Pre-registered Confirmation on Spider](https://mlsanity.com/paper/2610.17256/) (2026)

## Abstract (arXiv)

Evaluating uncertainty in AI-generated SQL queries requires estimating whether a query is correct, where correct means it executes to the same result as a human-written reference. We study which signals predict correctness on hard multi-table text-to-SQL, using AUROC to measure how well each ranks correct queries above incorrect ones. On BIRD and Spider, black-box signals such as string, structural, and execution self-consistency, a schema-relevance score, and query executability all fall between about 0.61 and 0.68 AUROC, with string self-consistency strongest at 0.675; white-box log-probability is similar (0.67). The signals that move past this ceiling are verification-based: an LLM judge scores from 0.72 (GPT-4o-mini) to 0.78 (Claude). Judges from different providers make different errors, so a two-provider ensemble reaches 0.82 AUROC with a well-calibrated probability (expected calibration error 0.03) and supports useful abstention frontiers (for example, answering 27% of questions at 24% selective risk) where self-consistency offers no valid low-risk subset. The pattern holds across two benchmarks, two generators, and two judge providers. We also ask whether a verifier can be trained. Fine-tuned verifiers, both encoder and generative, reach about 0.77 to 0.79 AUROC in-distribution but fall to about 0.66 on unseen schemas; scaling to 7B, adding schema diversity, distilling a strong judge's rationales, and cross-benchmark training all fail to close that gap. Cross-schema transfer appears to track model scale and reasoning rather than fine-tuning. In practice, correctness uncertainty for text-to-SQL lives in reasoning-based signals: a fine-tuned verifier is a good in-domain tool, but a verifier that generalizes across schemas currently means a large frozen reasoning model.