finding

active

finding:the-three-code-agents-never-overbid

The three code agents never overbid

Deterministic heuristics avoid the overbidding failure mode entirely.

Source paper

extracted_from

Cattle Trade: A Multi-Agent Benchmark for LLM Bluffing, Bidding, and Bargaining

(2026) · Robert Müller · Clemens Müller

Neighborhood — ranked by edge-count

Claims (2)

claim

Behavioural traces surface recurring LLM failure modes including overbidding, self-bidding, bankrupt TC initiation, and weak opponent-state adaptation that never appear in code agents.
supports
LLMs exhibit systematic errors that deterministic logic avoids.
Conditional logic already suffices where LLMs still fail, as code agents avoid systematic failures
supports
contrast between rule-based and LLM reasoning

Related by similarity (8)

cosine ≥ 0.65 · no typed edge

Entities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.

Two heuristic code agents outperform most tested LLMs, and behavioural traces surface recurring LLM failure modes including overbidding, self-bidding, bankrupt TC initiation, and weak opponent-state adaptation.quote0.773
Abstract sentence summarising performance and failures.
G3-F wins 67.9% of 28 mixed games (vs three code agents)finding0.766
Robust performance against algorithmic baselines.
Two heuristic code agents outperform most tested LLMsclaim0.765
author assertion that deterministic heuristics surpass many LLMs
Two heuristic code agents (TrackerAgent and SetRaceAgent) outperform most tested LLMs.claim0.760
Calibration that conditional logic can beat cost-efficient LLMs in this setting.
Code agents operate on structured data with exact arithmetic, while LLMs must parse natural-language observations and track state across turns; some failures may partly reflect numerical parsing or working-memory limitationsclaim0.755
discussion of potential confounds
Gemini 3 Flash wins 67.9% of its 28 mixed-format games against code agentsfinding0.754
In the 172-game exp2 slice, G3-F has the highest LLM win rate against deterministic baselines.
G2.5-FL overbid rate=1.20%, highest among all agentsfinding0.740
highest overbid frequency observed
Smith & Boyd (1991) criteria are irrelevant across the vast majority of the space of possible agents.claim0.738
Dismissal of earlier criteria as too narrow.