SFX warmup issue of ma_std_dev

[femtotrader]

Hello,

I think SFX have a warmup issue (on ma_std_dev)

See data

Date,Open,High,Low,Close,Volume,atr,std_dev,ma_std_dev
02/22/2013 12:00:00 AM,151.16,151.89,150.49,151.89,106292600,,,
02/25/2013 12:00:00 AM,152.63,152.86,149.0,149.0,245483200,,,
02/26/2013 12:00:00 AM,149.72,150.2,148.73,150.02,186331600,,,
02/27/2013 12:00:00 AM,149.89,152.33,149.76,151.91,143932600,,,
02/28/2013 12:00:00 AM,151.9,152.87,151.41,151.61,123724200,,,
03/01/2013 12:00:00 AM,151.09,152.34,150.41,152.11,170612700,,,
03/04/2013 12:00:00 AM,151.76,152.92,151.52,152.92,97328800,,,
03/05/2013 12:00:00 AM,153.66,154.7,153.64,154.29,121238200,,,
03/06/2013 12:00:00 AM,154.84,154.92,154.16,154.5,94284300,,,
03/07/2013 12:00:00 AM,154.71,154.98,154.52,154.78,86002000,,,
03/08/2013 12:00:00 AM,155.46,155.65,154.66,155.44,122805700,,,
03/11/2013 12:00:00 AM,155.33,156.04,155.13,156.03,83235500,1.5841666666666658,2.0828806174782697,
03/12/2013 12:00:00 AM,155.92,156.1,155.21,155.68,104108300,1.5263194444444423,2.1939593521505545,
03/13/2013 12:00:00 AM,155.76,156.12,155.23,155.9,91051500,1.4732928240740735,1.90548532021518,9992.0607750966146683
03/14/2013 12:00:00 AM,156.31,156.8,156.22,156.73,123614200,1.4255184220679011,1.6973876987889336,1.9322774570515562
03/15/2013 12:00:00 AM,155.85,156.04,155.31,155.83,133153200,1.4250585535622415,1.5738955563258366,1.7255895251099835
03/18/2013 12:00:00 AM,154.34,155.64,154.2,154.97,126518300,1.4421370074320568,1.27914576530155,1.5168096734721068
03/19/2013 12:00:00 AM,155.3,155.51,153.59,154.61,167049900,1.4819589234793842,0.9684781876738394,1.273839836433742
03/20/2013 12:00:00 AM,155.52,155.95,155.26,155.69,113607800,1.4701290131894336,0.7065226858038988,0.9847155462597628

9992.0607750966146683 looks odd here (compared to ma_std values which where 2.0828806174782697, 2.1939593521505545, 1.90548532021518

Data https://github.com/QuantConnect/Lean/blob/master/Tests/TestData/spy_ppo.txt

Code to reproduce

import pandas as pd
from pathlib import Path
import talipp
from talipp.ohlcv import OHLCVFactory
from talipp.indicators import SFX

def main():
    fname = "spy_ppo.txt"
    datetime_fmt = "%m/%d/%Y %I:%M:%S %p"

    df = pd.read_csv(fname)
    df["Date"] = pd.to_datetime(df["Date"], format=datetime_fmt)
    df = df.set_index("Date")

   def generate_reference_data_for_mimo_indicator(
        indicator_type, parameters, output_names, fname_stem
    ):
        df_in = df[["Open", "High", "Low", "Close", "Volume"]]
        input_values = OHLCVFactory.from_dict(
            {
                "open": df.Open,
                "high": df.High,
                "low": df.Low,
                "close": df.Close,
                "volume": df.Volume,
            }
        )
        indicator = indicator_type(**parameters, input_values=input_values)
        output_values = {output_name: [] for output_name in output_names}
        raw_output_values = list(indicator)
        for raw_output_value in raw_output_values:
            if raw_output_value is None:
                for output_name in output_names:
                    output_value_for_name = np.nan
                    output_values[output_name].append(output_value_for_name)
            else:
                for output_name in output_names:
                    output_value_for_name = getattr(raw_output_value, output_name)
                    output_values[output_name].append(output_value_for_name)
        df_out = pd.concat([df_in, pd.DataFrame(output_values, index=df.index)], axis=1)
        df_out.index = df_out.index.strftime(datetime_fmt)
        df_out.to_csv(Path("out") / f"{fname_stem}.csv")

    generate_reference_data_for_mimo_indicator(
        SFX,
        {"atr_period": 12, "std_dev_period": 12, "std_dev_smoothing_period": 3},
        ["atr", "std_dev", "ma_std_dev"],
        "spy_with_sfx",
    )

if __name__ == "__main__":
    main()

Kind regards