[HELP] Difference from RSI calculated by Python and RSI show on Chart MT5

New comment
 

Hi all,

As tittle, im trying calculate rsi using python follow this code:

# Calculate RSI
        window_rsi = 14
        delta = df['close'].diff()
        gain = (delta.where(delta > 0, 0)).rolling(window=window_rsi).mean()
        loss = (-delta.where(delta < 0, 0)).rolling(window=window_rsi).mean()
        rs = gain / loss
        df['rsi'] = 100 - (100 / (1 + rs))

But the result is difference to Chart. I tried the other library like ta, ta-lib, still difference to Chart

Can some one help me how can i calculate rsi exact chart.

Thank you so much.

 

This should be the right way to do it ..


delta = df['close'].diff()

gain = delta.where(delta > 0, 0)

loss = -delta.where(delta < 0, 0)


# Calculate the first AvgGain and AvgLoss

avg_gain = gain.rolling(window=window_rsi).mean().dropna()

avg_loss = loss.rolling(window=window_rsi).mean().dropna()


# Calculate the remaining AvgGain and AvgLoss

for i in range(len(avg_gain), len(gain)):

    avg_gain.loc[gain.index[i]] = (avg_gain.loc[gain.index[i-1]]*13 + gain.loc[gain.index[i]]) / 14

    avg_loss.loc[loss.index[i]] = (avg_loss.loc[loss.index[i-1]]*13 + loss.loc[loss.index[i]]) / 14


rs = avg_gain / avg_loss

df['rsi'] = 100 - (100 / (1 + rs))


 

Its still difference 

i tested on XAUUSD M1 exness, on chart at 21:43 is 62.40 but calculate python is 79.12 bro.

Joseph Obasi #:

This should be the right way to do it ..


delta = df['close'].diff()

gain = delta.where(delta > 0, 0)

loss = -delta.where(delta < 0, 0)


# Calculate the first AvgGain and AvgLoss

avg_gain = gain.rolling(window=window_rsi).mean().dropna()

avg_loss = loss.rolling(window=window_rsi).mean().dropna()


# Calculate the remaining AvgGain and AvgLoss

for i in range(len(avg_gain), len(gain)):

    avg_gain.loc[gain.index[i]] = (avg_gain.loc[gain.index[i-1]]*13 + gain.loc[gain.index[i]]) / 14

    avg_loss.loc[loss.index[i]] = (avg_loss.loc[loss.index[i-1]]*13 + loss.loc[loss.index[i]]) / 14


rs = avg_gain / avg_loss

df['rsi'] = 100 - (100 / (1 + rs))


 
Are you sure you are applying to exact same data? Data can be quite different depending on sources (especially in M1).
 
Yashar Seyyedin #:
Are you sure you are applying to exact same data? Data can be quite different depending on sources (especially in M1).


Yes, im sure that price open, close, high, low and period is correct.

 

To closely match the RSI values you see on most charts, you might want to use an exponential moving average (EMA) instead of SMA for the average gain and loss. Here's how you can adjust your code to use EMA, which might reduce the discrepancies:

import pandas as pd

# Assuming df is your DataFrame and it contains a 'close' column
window_rsi = 14
delta = df['close'].diff()
gain = delta.where(delta > 0, 0)
loss = -delta.where(delta < 0, 0)

# Use EMA instead of SMA
gain_ema = gain.ewm(alpha=1/window_rsi, adjust=False).mean()
loss_ema = loss.ewm(alpha=1/window_rsi, adjust=False).mean()

rs = gain_ema / loss_ema
df['rsi'] = 100 - (100 / (1 + rs))

Please note, the exact calculation might still differ slightly from what you see on some charts due to variations in how initial values are handled or how data boundaries are treated. Additionally, ensure your data (specifically the 'close' prices) is clean and matches the data used by the chart you're comparing against, as discrepancies in data can lead to different RSI values.

 
Nguyen An Nguyen #:

To closely match the RSI values you see on most charts, you might want to use an exponential moving average (EMA) instead of SMA for the average gain and loss. Here's how you can adjust your code to use EMA, which might reduce the discrepancies:

Please note, the exact calculation might still differ slightly from what you see on some charts due to variations in how initial values are handled or how data boundaries are treated. Additionally, ensure your data (specifically the 'close' prices) is clean and matches the data used by the chart you're comparing against, as discrepancies in data can lead to different RSI values.

Thank you so much, its exact what i need. 

Love u 3000 <3

New comment
Reason: