anandanand84/technicalindicators

Looks like most of indicator calculation is wrong.

dipankar08 opened this issue · 8 comments

dipankar08 commented

I have tried to use this indicator and I found significant different the actual value and the values provided by this indicator module. Looks like RSI, PSAR, MACD proving the wrong results. ( while examine with talib, tradingview ). I see only SMA is working as expected.

I think the libs is not yet ready for prod usecase.

da7a90 commented

I have tried to use this indicator and I found significant different the actual value and the values provided by this indicator module. Looks like RSI, PSAR, MACD proving the wrong results. ( while examine with talib, tradingview ). I see only SMA is working as expected.

I think the libs is not yet ready for prod usecase.

I believe the same, I was using a library that depends on this one and the results are off and only sma and ema are working

moshtaghi commented

Since the calculation is based on historical data and serially, to have an accurate result, the number of input candles must be increased to 1000. :)

rlmomin commented

Since the calculation is based on historical data and serially, to have an accurate result, the number of input candles must be increased to 1000. :)

I agree with his statement. I use this library and use over 1000 candles. The calculations match exactly with my charts on the TD Ameritrade ThinkorSwim platform.

rishikesh2003 commented

Yes, I also agree. I've used most of the indicators in this library and I still don't find any errors in the value. You should give more input data in order for the calculations to be accurate

DroidBoyJr commented

Confirmed. it's the sample set size. For ema interval 50, I took history of 200 points and it showed correct value. If I take only 51 or so, it shows incorrect EMA calculation.

saru2020 commented

it's a good library, where most indicators seem to be working correctly.

it is true that giving all candles/data works for some indicators like RSI and MACD(for SMA only) but not for all, or at least I see an issue with MACD(EMA based) even after feeding with all the candle/data as input, looks like EMA isn't working properly, are there any inputs to fix this issue?

liesauer commented

i feed 500 continuous candles to calculate Stochastic Oscillator, sometimes it looks good(still a little deviation), but it looks ridiculous in next second. only the last candle is realtime updating, how does it make such huge difference?

{ k: 77.77777777777779, d: 59.391851851851904 }
{ k: 11.11111111111111, d: 37.16962962962969 }
{ k: 11.11111111111111, d: 37.16962962962969 }
{ k: 11.11111111111111, d: 37.16962962962969 }
{ k: 11.11111111111111, d: 37.16962962962969 }
{ k: 11.11111111111111, d: 37.16962962962969 }
{ k: 11.11111111111111, d: 37.16962962962969 }
{ k: 11.122222222222275, d: 37.17333333333341 }
{ k: 11.122222222222275, d: 37.17333333333341 }
{ k: 11.122222222222275, d: 37.17333333333341 }
{ k: 77.77777777777779, d: 59.391851851851904 }
{ k: 77.77777777777779, d: 59.391851851851904 }
{ k: 77.77777777777779, d: 59.391851851851904 }
{ k: 77.77777777777779, d: 59.391851851851904 }
{ k: 77.77777777777779, d: 59.391851851851904 }
{ k: 77.77777777777779, d: 59.391851851851904 }
{ k: 77.77777777777779, d: 59.391851851851904 }
liesauer commented

i don't think it's the data samples issue.

image