Wilder's Smoothing is applied in both the numerator and denominator of the calculation. In addition, you need to use a lot of bars of data to get the calculations to converge on the correct value.

For example, in the Personal Criteria Formula Language, a decent approximation of a 14-Period Wilder's Smoothed RSI would be:

50 * (C - XAVGC27.1) / 14 / (.071435345156143 * (ABS(C - C1) + 13 / 14 * (ABS(C1 - C2) + 13 / 14 * (ABS(C2 - C3) + 13 / 14 * (ABS(C3 - C4) + 13 / 14 * (ABS(C4 - C5) + 13 / 14 * (ABS(C5 - C6) + 13 / 14 * (ABS(C6 - C7) + 13 / 14 * (ABS(C7 - C8) + 13 / 14 * (ABS(C8 - C9) + 13 / 14 * (ABS(C9 - C10) + 13 / 14 * (ABS(C10 - C11) + 13 / 14 * (ABS(C11 - C12) + 13 / 14 * (ABS(C12 - C13) + 13 / 14 * (ABS(C13 - C14) + 13 / 14 * (ABS(C14 - C15) + 13 / 14 * (ABS(C15 - C16) + 13 / 14 * (ABS(C16 - C17) + 13 / 14 * (ABS(C17 - C18) + 13 / 14 * (ABS(C18 - C19) + 13 / 14 * (ABS(C19 - C20) + 13 / 14 * (ABS(C20 - C21) + 13 / 14 * (ABS(C21 - C22) + 13 / 14 * (ABS(C22 - C23) + 13 / 14 * (ABS(C23 - C24) + 13 / 14 * (ABS(C24 - C25) + 13 / 14 * (ABS(C25 - C26) + 13 / 14 * (ABS(C26 - C27) + 13 / 14 * (ABS(C27 - C28) + 13 / 14 * (ABS(C28 - C29) + 13 / 14 * (ABS(C29 - C30) + 13 / 14 * (ABS(C30 - C31) + 13 / 14 * (ABS(C31 - C32) + 13 / 14 * (ABS(C32 - C33) + 13 / 14 * (ABS(C33 - C34) + 13 / 14 * (ABS(C34 - C35) + 13 / 14 * (ABS(C35 - C36) + 13 / 14 * (ABS(C36 - C37) + 13 / 14 * (ABS(C37 - C38) + 13 / 14 * (ABS(C38 - C39) + 13 / 14 * (ABS(C39 - C40) + 13 / 14 * (ABS(C40 - C41) + 13 / 14 * (ABS(C41 - C42) + 13 / 14 * (ABS(C42 - C43) + 13 / 14 * (ABS(C43 - C44) + 13 / 14 * (ABS(C44 - C45) + 13 / 14 * (ABS(C45 - C46) + 13 / 14 * (ABS(C46 - C47) + 13 / 14 * (ABS(C47 - C48) + 13 / 14 * (ABS(C48 - C49) + 13 / 14 * (ABS(C49 - C50) + 13 / 14 * (ABS(C50 - C51) + 13 / 14 * (ABS(C51 - C52) + 13 / 14 * (ABS(C52 - C53) + 13 / 14 * (ABS(C53 - C54) + 13 / 14 * (ABS(C54 - C55) + 13 / 14 * (ABS(C55 - C56) + 13 / 14 * (ABS(C56 - C57) + 13 / 14 * (ABS(C57 - C58) + 13 / 14 * (ABS(C58 - C59) + 13 / 14 * (ABS(C59 - C60) + 13 / 14 * (ABS(C60 - C61) + 13 / 14 * (ABS(C61 - C62) + 13 / 14 * (ABS(C62 - C63) + 13 / 14 * (ABS(C63 - C64) + 13 / 14 * (ABS(C64 - C65) + 13 / 14 * (ABS(C65 - C66) + 13 / 14 * (ABS(C66 - C67) + 13 / 14 * (ABS(C67 - C68) + 13 / 14 * (ABS(C68 - C69) + 13 / 14 * (ABS(C69 - C70) + 13 / 14 * (ABS(C70 - C71) + 13 / 14 * (ABS(C71 - C72) + 13 / 14 * (ABS(C72 - C73) + 13 / 14 * (ABS(C73 - C74) + 13 / 14 * (ABS(C74 - C75) + 13 / 14 * (ABS(C75 - C76) + 13 / 14 * (ABS(C76 - C77) + 13 / 14 * (ABS(C77 - C78) + 13 / 14 * (ABS(C78 - C79) + 13 / 14 * (ABS(C79 - C80) + 13 / 14 * (ABS(C80 - C81) + 13 / 14 * (ABS(C81 - C82) + 13 / 14 * (ABS(C82 - C83) + 13 / 14 * (ABS(C83 - C84) + 13 / 14 * (ABS(C84 - C85) + 13 / 14 * (ABS(C85 - C86) + 13 / 14 * (ABS(C86 - C87) + 13 / 14 * (ABS(C87 - C88) + 13 / 14 * (ABS(C88 - C89) + 13 / 14 * (ABS(C89 - C90) + 13 / 14 * (ABS(C90 - C91) + 13 / 14 * (ABS(C91 - C92) + 13 / 14 * (ABS(C92 - C93) + 13 / 14 * (ABS(C93 - C94) + 13 / 14 * (ABS(C94 - C95) + 13 / 14 * (ABS(C95 - C96) + 13 / 14 * (ABS(C96 - C97) + 13 / 14 * (ABS(C97 - C98) + 13 / 14 * (ABS(C98 - C99) + 13 / 14 * (ABS(C99 - C100) + 13 / 14 * (ABS(C100 - C101) + 13 / 14 * (ABS(C101 - C102) + 13 / 14 * (ABS(C102 - C103) + 13 / 14 * (ABS(C103 - C104) + 13 / 14 * (ABS(C104 - C105) + 13 / 14 * (ABS(C105 - C106) + 13 / 14 * (ABS(C106 - C107) + 13 / 14 * (ABS(C107 - C108) + 13 / 14 * (ABS(C108 - C109) + 13 / 14 * (ABS(C109 - C110) + 13 / 14 * (ABS(C110 - C111) + 13 / 14 * (ABS(C111 - C112) + 13 / 14 * (ABS(C112 - C113) + 13 / 14 * (ABS(C113 - C114) + 13 / 14 * (ABS(C114 - C115) + 13 / 14 * (ABS(C115 - C116) + 13 / 14 * (ABS(C116 - C117) + 13 / 14 * (ABS(C117 - C118) + 13 / 14 * (ABS(C118 - C119) + 13 / 14 * (ABS(C119 - C120) + 13 / 14 * (ABS(C120 - C121) + 13 / 14 * (ABS(C121 - C122) + 13 / 14 * (ABS(C122 - C123) + 13 / 14 * (ABS(C123 - C124) + 13 / 14 * (ABS(C124 - C125))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))) + 50

And that is using an Exponential Moving Average of Price to eliminate the manual expansion in the numerator. Only the denominator has been fully expanded manually.

Thanks Bruce.

Hi Bruce,

just need few more explanations...

I am exporting datas to excel and i want to apply the formula to calculate the wilder RSI value:

For 14 days if i use this formula to calculate average:

Wilder's smoothing = Previous MA value + (1/n periods * (Close - Previous MA))

I don't get the same value from the software.

Can you summarize the formula the software use to calculate the exponential average

to get the same value of the software so that i can apply  rsi = 100 + 100 / (1 + AVG/AVL).

Thanks

To get something similar to what the software is using, you need to use a lot more than 14 bars of data. The approximation given in my Wednesday, May 02, 2012 12:33:22 PM ET post uses 126 bars of data in its calculations.

I've attached an spreadsheet. It uses recent closing prices from SNX to calculate an RSI manually in column G while displaying the RSI as exported from TC2000 version 12.2 in column H.

Attachments:
RSI Example_SNX.xlsx - 74 KB, downloaded 796 time(s).

Thank you very much.

Can i send you attachments files because it works for RSI 14 but not for RSI 9 ?

We don't provide support or training for third party products. We are not going to troubleshoot somebody else's spreadsheet or provide Excel training.

That said, while a spreadsheet with calculations on a 14-Period RSI is certainly not going to match a 9-Period RSI, adjusting the calculations to account for the change in period gets things incredibly close again.

Attachments:
RSI9_Example_IBM.xlsx - 72 KB, downloaded 714 time(s).

Thanks Bruce for your time.

It works well for day data but when i try with intraday value i still

don't get good values.

Can you try with with 1 min data for a stock with RSI 9 ?

Your help is greatly appreciated.

It works just fine with the 1-Minute data as well.

Attachments:
RSI9_1MinExample_GOOG.xlsx - 106 KB, downloaded 657 time(s).

I am using freestockcharts to export data. Is it possible that the values obtain are different

from TC2000 ?

Thanks

Gabriel