Golang : Calculate Relative Strength Index(RSI) example

Alright, below is a simple example on how to calculate the Relative Strength Index(RSI) using data from Binance.com and Golang. This example should be able to work with little modification if you're pulling your data from exchange other than Binance. All you need is a slice with Close price values.

UPDATE : Please use the RSI function from "github.com/markcheno/go-talib" It matches the TradingView.com result.

Here you go!

 package main

 import (
  "context"
  "fmt"
  "math"
  "strconv"
  "time"

  "github.com/adshao/go-binance"
 )

 var (

  //Binance
  binanceAPIKey = "<your binance api key>"
  binanceSecretKey = "<your binance secret key>"
 )

 func humanTimeToUnixNanoTime(input time.Time) int64 {
  return int64(time.Nanosecond) * input.UnixNano() / int64(time.Millisecond)
 }

 func main() {

  client := binance.NewClient(binanceAPIKey, binanceSecretKey)

  targetSymbol := "BNBUSDT"

  // reference https://golang.org/pkg/time/#Date
  // func Date(year int, month Month, day, hour, min, sec, nsec int, loc *Location) Time

  // NOTE : Binance only accepts unix nano timestamp, not unix time stamp
  // For example
  // 1527616800  <----- will return empty result or wrong date range
  // 1527616899999 <--- ok

  //startTimeHumanReadable := time.Date(2018, time.October, 10, 0, 0, 0, 0, time.UTC)
  //start := humanTimeToUnixNanoTime(startTimeHumanReadable)

  //endTimeHumanReadable := time.Date(2018, time.October, 24, 0, 0, 0, 0, time.UTC)
  //end := humanTimeToUnixNanoTime(endTimeHumanReadable)

  targetInterval := "1d" /// case sensitive, 1D != 1d or 30M != 30m

  data, err := client.NewKlinesService().Symbol(targetSymbol).Interval(targetInterval).Do(context.Background())

  if err != nil {
 fmt.Println(err)
 return
  }

  totalGain := 0.0
  totalLoss := 0.0

  for i := 1; i < len(data); i++ {

 previous := data[i].Close
 current := data[i-1].Close

 // convert string to float64
 previousClose, _ := strconv.ParseFloat(previous, 64)
 currentClose, _ := strconv.ParseFloat(current, 64)

 difference := currentClose - previousClose

 if difference >= 0 {
 totalGain += difference
 } else {
 totalLoss -= difference
 }
  }

  rs := totalGain / math.Abs(totalLoss)
  rsi := 100 - 100/(1+rs)
  fmt.Println("RSI for "+targetSymbol+" : ", rsi)
 }

Example output:

Thu Oct 25 11:35:33 +08 2018

RSI for BNBUSDT : 47.84161851469368

NOTE:

UPDATE : Please use the RSI function from "github.com/markcheno/go-talib" It matches the TradingView.com result.

I've tried to compare the result to see if it matches the RSI values calculated by TradingView.com (TV) and what I found out is that BTCUSDT and BNBUSDT RSI values are pretty close to the values calculated by this code example. I noticed that in TV's RSI row, it has complete data dating back to more than 1 year.

Most of the mismatched values found in TV have data less than 5 months. Some said that the data needs to be smoothed out first, etc. So far,I'm pretty ok with the values given by this script.

References:

https://stockcharts.com/school/doku.php?id=chartschool:technicalindicators:relativestrengthindex_rsi

https://stackoverflow.com/questions/22626238/calculate-rsirelative-strength-index-using-some-programming-language-js-c