# Financial Forecasting on Exchange-rate 5 rolling-origin windows GPT-5.2 (test)

Source: [Financial Forecasting on Exchange-rate 5 rolling-origin windows GPT-5.2 (test)](<https://www.sota2.com/research/sota/financial-forecasting-on-exchange-rate-5-rolling-origin-windows-gpt-5-2-test>)

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

Data updated: 2026-04-13T07:39:38.450Z

Evaluation of models performing Financial Forecasting on the test split of the Exchange-rate 5 rolling-origin windows GPT-5.2 dataset, assessed using Sharpe ratio and Root Mean Squared Error (RMSE).

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 Forecasting |
| Dataset (including any stated split) | Exchange-rate 5 rolling-origin windows GPT-5.2 (test) |
| Metric | Sharpe Ratio |
| Unit | SCALAR |
| Direction | Higher is better |
| Results loaded | 10 of 10 |
| 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

- [Sharpe Ratio](<https://www.sota2.com/research/sota/financial-forecasting-on-exchange-rate-5-rolling-origin-windows-gpt-5-2-test.md?metric=Sharpe%20Ratio>): Higher is better; unit: SCALAR.

- [RMSE](<https://www.sota2.com/research/sota/financial-forecasting-on-exchange-rate-5-rolling-origin-windows-gpt-5-2-test.md?metric=RMSE>): Lower is better; unit: SCALAR.

## Recorded results

| Method | Sharpe Ratio | Result date | Report paper |
| --- | --- | --- | --- |
| TimesNet multivariate Student-T | 4.214 | Not recorded | 69dc788a7d0172140ec78aa3 |
| PatchTST factor-first PCA(3) + reconstruct | 2.78 | Not recorded | 69dc788a7d0172140ec78aa3 |
| Transformer policy net (Sharpe loss) | 0.748 | Not recorded | 69dc788a7d0172140ec78aa3 |
| Stacking ensemble (turnover-penalized) | 0.697 | Not recorded | 69dc788a7d0172140ec78aa3 |
| TCN multivariate risk-parity | 0.395 | Not recorded | 69dc788a7d0172140ec78aa3 |
| VAR(1) | 0.17 | Not recorded | 69dc788a7d0172140ec78aa3 |
| drift | 0.16 | Not recorded | 69dc788a7d0172140ec78aa3 |
| TSMixer global returns | 0.121 | Not recorded | 69dc788a7d0172140ec78aa3 |
| TFT cross-currency with lag covariates | 0.077 | Not recorded | 69dc788a7d0172140ec78aa3 |
| (22 experiments with negative Sharpe) | 0.05 | Not recorded | 69dc788a7d0172140ec78aa3 |

### TimesNet multivariate Student-T

Sharpe Ratio: 4.214. Result ID: 69dc9d8447f34ce924c34c87.

[Reporting paper](<https://www.sota2.com/research/paper/69dc788a7d0172140ec78aa3.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69dc788a7d0172140ec78aa3.md>)

Recorded evaluation configuration: {"LLM":"GPT-5.2","variant":"multivariate Student-T"}. This record does not assert independent reproduction.

### PatchTST factor-first PCA(3) + reconstruct

Sharpe Ratio: 2.78. Result ID: 69dc9d8447f34ce924c34c88.

[Reporting paper](<https://www.sota2.com/research/paper/69dc788a7d0172140ec78aa3.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69dc788a7d0172140ec78aa3.md>)

Recorded evaluation configuration: {"LLM":"GPT-5.2","components":"factor-first PCA(3) + reconstruct"}. This record does not assert independent reproduction.

### Transformer policy net (Sharpe loss)

Sharpe Ratio: 0.748. Result ID: 69dc9d8447f34ce924c34c89.

[Reporting paper](<https://www.sota2.com/research/paper/69dc788a7d0172140ec78aa3.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69dc788a7d0172140ec78aa3.md>)

Recorded evaluation configuration: {"LLM":"GPT-5.2","loss\_function":"Sharpe loss"}. This record does not assert independent reproduction.

### Stacking ensemble (turnover-penalized)

Sharpe Ratio: 0.697. Result ID: 69dc9d8447f34ce924c34c8a.

[Reporting paper](<https://www.sota2.com/research/paper/69dc788a7d0172140ec78aa3.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69dc788a7d0172140ec78aa3.md>)

Recorded evaluation configuration: {"LLM":"GPT-5.2","regularization":"turnover-penalized"}. This record does not assert independent reproduction.

### TCN multivariate risk-parity

Sharpe Ratio: 0.395. Result ID: 69dc9d8447f34ce924c34c8b.

[Reporting paper](<https://www.sota2.com/research/paper/69dc788a7d0172140ec78aa3.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69dc788a7d0172140ec78aa3.md>)

Recorded evaluation configuration: {"LLM":"GPT-5.2","strategy":"risk-parity"}. This record does not assert independent reproduction.

### VAR(1)

Sharpe Ratio: 0.17. Result ID: 69dc9d8447f34ce924c34c90.

[Reporting paper](<https://www.sota2.com/research/paper/69dc788a7d0172140ec78aa3.md>)

Recorded evaluation configuration: {"type":"baseline"}. This record does not assert independent reproduction.

### drift

Sharpe Ratio: 0.16. Result ID: 69dc9d8447f34ce924c34c8f.

[Reporting paper](<https://www.sota2.com/research/paper/69dc788a7d0172140ec78aa3.md>)

Recorded evaluation configuration: {"type":"baseline"}. This record does not assert independent reproduction.

### TSMixer global returns

Sharpe Ratio: 0.121. Result ID: 69dc9d8447f34ce924c34c8c.

[Reporting paper](<https://www.sota2.com/research/paper/69dc788a7d0172140ec78aa3.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69dc788a7d0172140ec78aa3.md>)

Recorded evaluation configuration: {"LLM":"GPT-5.2"}. This record does not assert independent reproduction.

### TFT cross-currency with lag covariates

Sharpe Ratio: 0.077. Result ID: 69dc9d8447f34ce924c34c8d.

[Reporting paper](<https://www.sota2.com/research/paper/69dc788a7d0172140ec78aa3.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69dc788a7d0172140ec78aa3.md>)

Recorded evaluation configuration: {"LLM":"GPT-5.2","features":"lag covariates"}. This record does not assert independent reproduction.

### (22 experiments with negative Sharpe)

Sharpe Ratio: 0.05. Result ID: 69dc9d8447f34ce924c34c8e.

[Reporting paper](<https://www.sota2.com/research/paper/69dc788a7d0172140ec78aa3.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69dc788a7d0172140ec78aa3.md>)

Recorded evaluation configuration: {"LLM":"GPT-5.2"}. This record does not assert independent reproduction.

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