# Financial Advisory Preference Ranking on S&P 500 2017 (held-out)

Source: [Financial Advisory Preference Ranking on S&P 500 2017 (held-out)](<https://www.sota2.com/research/sota/financial-advisory-preference-ranking-on-s-and-p-500-2017-held-out>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-04-28T09:31:05.758Z

Evaluation of model performance in ranking financial advisory preferences using head-to-head comparisons against other models, applied to the held-out 2017 data subset of the S&P 500.

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

| Scope | Recorded value |
| --- | --- |
| Task | Financial Advisory Preference Ranking |
| Dataset (including any stated split) | S&P 500 2017 (held-out) |
| Metric | Win Rate vs Qwen3-VL-235B |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 2 of 2 |
| Coverage | All results currently returned for this benchmark |

Evaluation settings may differ between results. The original papers and recorded configurations are needed to establish comparability. A shared metric name alone does not establish an identical protocol.

## Metrics

- [Win Rate vs Qwen3-VL-235B](<https://www.sota2.com/research/sota/financial-advisory-preference-ranking-on-s-and-p-500-2017-held-out.md?metric=Win%20Rate%20vs%20Qwen3-VL-235B>): Higher is better; unit: PERCENT.

- [Win Rate vs Claude Sonnet 3.7](<https://www.sota2.com/research/sota/financial-advisory-preference-ranking-on-s-and-p-500-2017-held-out.md?metric=Win%20Rate%20vs%20Claude%20Sonnet%203.7>): Higher is better; unit: PERCENT.

- [Win Rate vs Qwen3-VL-4B](<https://www.sota2.com/research/sota/financial-advisory-preference-ranking-on-s-and-p-500-2017-held-out.md?metric=Win%20Rate%20vs%20Qwen3-VL-4B>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Win Rate vs Qwen3-VL-235B | Result date | Report paper |
| --- | --- | --- | --- |
| Qwen3-VL-4B + Hindsight DPO | 56.8 | Not recorded | 69f03f2c55a3c32af89fa19f |
| Qwen3-VL-4B + SFT (Stage 1) | 48.7 | Not recorded | 69f03f2c55a3c32af89fa19f |

### Qwen3-VL-4B + Hindsight DPO

Win Rate vs Qwen3-VL-235B: 56.8. Result ID: 69f07dba47f34ce924d6bc5c.

[Reporting paper](<https://www.sota2.com/research/paper/69f03f2c55a3c32af89fa19f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69f03f2c55a3c32af89fa19f.md>)

Recorded evaluation configuration: {"Training Stage":"Hindsight Direct Preference Optimization","Base Model":"Qwen3-VL-4B","Evaluation protocol":"LLM judge","Number of evaluation runs":"5"}. This record does not assert independent reproduction.

### Qwen3-VL-4B + SFT (Stage 1)

Win Rate vs Qwen3-VL-235B: 48.7. Result ID: 69f07dba47f34ce924d6bc5b.

[Reporting paper](<https://www.sota2.com/research/paper/69f03f2c55a3c32af89fa19f.md>)

Recorded evaluation configuration: {"Training Stage":"Supervised Fine-Tuning (Stage 1)","Base Model":"Qwen3-VL-4B","Evaluation protocol":"LLM judge","Number of evaluation runs":"5"}. This record does not assert independent reproduction.

Page 1 of 1. This is an index page, not the full collection.