Is there any certified case that timesfm could predict the trend of a stock?

[JackeyLee007]

I tried it with China's A-share 000001 and found that it was not accurate. Is there anything that I used improperly?

The blue-grey line is real data, and the red line is predicted value. I used the value from 20210101 to 20210430 to predict the value of the next 32 days.

import timesfm
import os
import numpy as np
import akshare as ak
import pandas as pd

stHist = ak.stock_zh_a_hist('000001', 'daily', '20210101', '20210430')
incVals=range(1, len(stHist)+1)
stHist['unique_id'] = incVals
stHist = stHist.rename(columns={'日期':'ds','收盘':'value'})
stHist['ds'] = pd.to_datetime(stHist['ds'])
print(stHist)
tfm = timesfm.TimesFm(
    context_len=128,
    horizon_len=32,
    input_patch_len=32,
    output_patch_len=128,
    num_layers=20,
    model_dims=1280,
    backend='cpu',
)
tfm.load_from_checkpoint(repo_id="google/timesfm-1.0-200m")

forecast_df = tfm.forecast_on_df(
        inputs=stHist,
        freq='D',
        value_name='value',
        num_jobs=1)