SkibidiBreaddd/BBCA-Chronos-14062026-v0

Fine Tuned to BBCA Stock Data of 2001-2023

This repository contains a fine-tuned version of amazon/chronos-bolt-small, optimized specifically for forecasting the stock price of Bank Central Asia Tbk (BBCA).

The model was trained on historical daily data spanning from April 16, 2001, to January 6, 2023.

Deployed Model

https://huggingface.co/spaces/SkibidiBreaddd/BBCA-Forecaster

Space above contains the deployed model forecasting real time stock prices of BBCA

Usage Guide

You can use this model for inference using the official chronos library.

from chronos import ChronosPipeline
import torch
import pandas as pd

pipeline = ChronosPipeline.from_pretrained(
    "SkibidiBreaddd/BBCA-Chronos-14062026-v0",
    device_map="auto",
    torch_dtype=torch.bfloat16,
)

# Example: Provide your historical context data (e.g., past 50 days of BBCA)
# context = torch.tensor([...])

# Generate forecasts
# forecast = pipeline.predict(context, prediction_length=12)

Improvements

Metrics

	MAE	RMSE	MAPE(%)	MASE
Baseline (Zero-Shot)	208.533783	253.615631	2.508856	8.371802
Fine-Tuned	191.213867	233.045746	2.300375	7.676476
Delta (%)	-8.305568	-8.110654	-8.309825	-8.305568

lower delta = Better Results

Metric Comparison

Residuals