Georgia Tech ML researcher building LLM forecasting agents, economic nowcasting, and reproducible agent benchmarks
2
papers here
1
reviewed by Evolve
1
first-author
0
corresponding
—
works · OpenAlex
2
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-10-17) and cover the author's entire public record, not just the papers listed here. Reproducibility signal comes from Evolve's own reviews.
Research profile
Output—
Output scoring is not yet available: the snapshot lacks complete venue-normalized publication counts, so this axis — which would measure cumulative and first-author productivity over time — is withheld rather than estimated from the partial yearly tallies.
Impact—
Impact is not yet scored: citation signals are too sparse at this stage (h-index 2, i10-index 1) to support a rating. This axis would measure citations, benchmark adoption, and downstream influence once the record accumulates.
Focus100/100
Focus scores 100/100: the OpenAlex topic profile is tightly clustered around text-centric machine learning — Topic Modeling (10 works), Computational and Text Analysis Methods (5), and Sentiment Analysis and Opinion Mining (4) — with Stock Market Forecasting Methods (3) and Multimodal Machine Learning Applications (3) as secondary threads, so the catalog field distribution collapses to a single field, LLM agents and event forecasting.
Collaboration55/100
Collaboration scores 55/100: the last three years show 13 works with 15 distinct co-authors, a moderate network for an early-career researcher. Recurring partners include Xinyue Zhao, Ruiyi Zhang, Rui Cao, Pengtao Xie, and Sudheer Chava, a mix that lines up with the text-analysis and market-forecasting topic threads.
Activity77/100
Activity scores 77/100: OpenAlex records 13 works in the last three years, with yearly output climbing from 2 works in 2022 to 7 in 2025 and 6 already in 2026, indicating a sustained and accelerating publishing rhythm alongside the first-authored Forecast-Dojo arXiv release.
Reproducibility100/100
Reproducibility scores 100/100, but from a single assessed paper (reproPapers = 1, reproMean = 100): Forecast-Dojo is rated high reproducibility readiness, and its replayable-environment design — putting all 12 evaluated models on the same replayed footing — is itself an artifact built for reproducible agent evaluation.
Research summary
Liqin Ye is a machine learning researcher at Georgia Institute of Technology whose work sits at the intersection of large language model agents and event forecasting. The OpenAlex snapshot shows an early-career profile (h-index 2, i10-index 1) with a sharply concentrated topic signature: Topic Modeling (10 works), Computational and Text Analysis Methods (5), Sentiment Analysis and Opinion Mining (4), Stock Market Forecasting Methods (3), and Multimodal Machine Learning Applications (3). Activity is on a clear upswing — 13 works in the last three years, rising from 2 indexed works in 2022 to 7 in 2025 and 6 in 2026 — supported by a moderate collaboration network of 15 distinct co-authors including Xinyue Zhao, Ruiyi Zhang, Rui Cao, Pengtao Xie, and Sudheer Chava. The flagship contribution in the Evolve catalog is Forecast-Dojo (2026, first author), a replayable-environment framework for benchmarking and training LLM forecasting agents on prediction markets, with explicit attention to agent memory and belief updating. Evolve's review calls it a well-engineered, replayable prediction benchmark and treats its headline finding — research tooling helped all 12 evaluated models, yet every model still trailed the market — as credible, partly owing to the architecture design; reproducibility readiness is rated high. A second 2026 arXiv paper extends the agenda to real-time LLM nowcasts of macroeconomic indicators. The overall trajectory points toward rigorous, infrastructure-first research on whether LLM agents can genuinely forecast real-world events.
Collaboration pattern
Collaboration is moderate but growing: 13 works over the last three years with 15 distinct co-authors. The recurring circle — Xinyue Zhao, Ruiyi Zhang, Rui Cao, Pengtao Xie, Sudheer Chava, Haorui Wang, Fardin Ahmed, Rongzhi Zhang, among others — pairs machine-learning collaborators with finance- and market-oriented partners, consistent with the topic mix of text analysis and stock market forecasting.
Research keywords
Highlights
First-authored Forecast-Dojo: replayable environments for LLM forecasting agents
Finding: tooling lifts all 12 models, yet none beats the market
Real-time evaluation of LLM nowcasts on macroeconomic indicators
A 2026 co-authored arXiv study, 'Can LLMs Take the Pulse of the Economy?', that evaluates LLM nowcasts against macroeconomic indicators in real time; it is listed in the catalog without an Evolve verdict or reproducibility assessment yet.
Timeline
- 2022Earliest year in the snapshot: 2 works indexed in OpenAlex.
- 2025Output expands to 7 indexed works, the busiest year on record.
- 20266 indexed works so far, including the first-authored Forecast-Dojo and a co-authored evaluation of LLM nowcasts on macroeconomic indicators, both on arXiv.
Papers in the Evolve catalog
Frequent co-authors
About this profile
This profile is AI-structured by Evolve.Pub from public bibliographic data (OpenAlex) and Evolve's own paper reviews. It is not written or endorsed by the researcher unless the profile is claimed. Radar dimensions are computed from real bibliometrics with fixed formulas — raw figures are shown alongside; missing dimensions mean insufficient data, never zero. Errors? Claim the profile to correct or hide it.
- Bibliometrics are drawn from OpenAlex only; output and impact scores are withheld because citation signals (h-index 2, i10-index 1) are too sparse to score reliably.
- The reproducibility score rests on a single reviewed paper (reproPapers = 1), so the 100/100 figure should be read as an early signal, not a track record.
- The profile is name-resolved from public bibliographic data (OpenAlex identity plus ORCID); 'Liqin Ye' may correspond to multiple researchers, and no biographical details beyond institution and country are verified.
- Only one catalog paper carries an Evolve review; the macroeconomic nowcasting paper is listed without verdict, keywords, or reproducibility assessment, and countsByYear shows no indexed works for 2023-2024.