Quant Models

Run Forecast Model Tournament

Backtest approved forecasting models on a holdout tail, rank by error, and refit the winner for the requested forward horizon.

Version: v2

Stability: stable

Sandbox: Available

Production: Entitlement required

Target latency: ≤ 1,800 ms

POST/api/v2/quant/model-tournaments

Candidate forecasts are persisted as auditable model runs. FinanceGPT does not represent tournament winners as guarantees of future performance.

Required scope: quant:models:run

Try it in the FinanceGPT sandbox

The playground uses deterministic isolated sandbox fixtures. It never queries production workspace data and does not require you to paste an API secret into the browser.

Example response

{
    "data": {
        "winner_model_key": "forecasting.holt_linear",
        "winner_forecast": [
            137.80000000000001136868377216160297393798828125,
            140.19999999999998863131622783839702606201171875,
            142.599999999999994315658113919198513031005859375
        ]
    }
}

Code examples

curl

curl -X POST \
  https://financegpt.dev/api/v2/quant/model-tournaments \
  -H "Authorization: Bearer $FINANCEGPT_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"series":[100,102,103,106,108,111,113,116,118,121,124,126,129,131,134,137],"holdout":4,"forecast_horizon":3,"metric":"rmse"}'

python

import os, requests

url = "https://financegpt.dev/api/v2/quant/model-tournaments"
headers = {"Authorization": f"Bearer {os.environ['FINANCEGPT_API_KEY']}"}
payload = {
    "series": [
        100,
        102,
        103,
        106,
        108,
        111,
        113,
        116,
        118,
        121,
        124,
        126,
        129,
        131,
        134,
        137
    ],
    "holdout": 4,
    "forecast_horizon": 3,
    "metric": "rmse"
}
response = requests.request("POST", url, headers=headers, json=payload)
response.raise_for_status()
print(response.json())

javascript

const response = await fetch('https://financegpt.dev/api/v2/quant/model-tournaments', {
  method: 'POST',
  headers: { Authorization: `Bearer ${process.env.FINANCEGPT_API_KEY}`, 'Content-Type': 'application/json' },
  body: JSON.stringify({"series":[100,102,103,106,108,111,113,116,118,121,124,126,129,131,134,137],"holdout":4,"forecast_horizon":3,"metric":"rmse"})
});
if (!response.ok) throw new Error(`FinanceGPT ${response.status}`);
console.log(await response.json());

php

<?php
$ch = curl_init('https://financegpt.dev/api/v2/quant/model-tournaments');
curl_setopt_array($ch, [CURLOPT_RETURNTRANSFER => true, CURLOPT_CUSTOMREQUEST => 'POST', CURLOPT_HTTPHEADER => ['Authorization: Bearer '.getenv('FINANCEGPT_API_KEY'), 'Content-Type: application/json'], CURLOPT_POSTFIELDS => '{\"series\":[100,102,103,106,108,111,113,116,118,121,124,126,129,131,134,137],\"holdout\":4,\"forecast_horizon\":3,\"metric\":\"rmse\"}']);
$response = curl_exec($ch);
if (curl_getinfo($ch, CURLINFO_HTTP_CODE) >= 400) throw new RuntimeException($response);
echo $response;

Versions & changelog

v2 published

M19 publication of this governed FinanceGPT API operation.

Ready to build?

Explorer gives immediate isolated sandbox access. Production plans require a commercial entitlement.

Create account