AgentHorizon
Harbor: pendingA benchmark of 1,373 computer-use tasks with paired instructions to evaluate agentic judges on long-horizon trajectory success verification across three OSs.
- Samples
- 1,373
- Released
- 2026
- Scores
- 10
Synthetic Trading Benchmark
Synthetic Trading Benchmark for Execution Assumptions
Evaluates stability of LLM trading agent rankings under varying execution assumptions using 2,462 synthetic runs across six settings.
How to run
ltb · synthetic-trading-benchmark
$ npm i -g lettokenburn-cli$ ltb bench info synthetic-trading-benchmark
External harness. No Harbor adapter has been generated for this benchmark yet. `ltb bench info` prints the known links.
Runs happen on your machine with your own subscription; only the score summary is pushed. CLI docs
Scores
5 sourced entries| Harness | Source | Confidence | Link | |||
|---|---|---|---|---|---|---|
| GLM-5 | 9.2 | N/A | Independent | medium | Oct 4, 2026 | arxiv.org |
| Claude Opus 4.7 | 9.0 | N/A | Independent | medium | Oct 4, 2026 | arxiv.org |
| GPT-5.5 | 9.0 | N/A | Independent | medium | Oct 4, 2026 | arxiv.org |
| DeepSeek V4 Pro | 8.2 | N/A | Independent | medium | Oct 4, 2026 | arxiv.org |
| Gemini 3.1 Pro | 7.7 | N/A | Independent | medium | Oct 4, 2026 | arxiv.org |
Every number links to where it was published. Low-confidence rows are muted and need per-cell verification.
Reproduced by the community
All runsLoading community runs…
About
The benchmark uses the TradeArena testbed to simulate trading for 12 policies (5 LLM, 7 classical) on synthetic price paths. It varies execution conditions from ideal fills to high-stress scenarios involving latency, spread, and impact.
Scoring measures rank agreement (Kendall tau) between execution settings rather than absolute returns. Key findings show ranking stability drops significantly in high-volatility regimes, highlighting sensitivity to execution conventions.