Please see the Connors RSI (CRSI) topic for improved syntax in TC2000 v17+.

Larry Connors has just disclosed a new propreitary momentum oscillator from Connors Research used for swing trading.  The indicator combines Wilder's RSI plus duration of an up / down trend and relative magnitude of price change into a single oscillator number.  Connor's research shows that the combination of the three elements is superior to analyzing the elements individually.  

I have included a link to TradingMarkets.com that will allow you to download the formula information free of charge:  http://www.tradingmarkets.com/about/tradingmarkets/our-latest-release-i-an-introduction-to-b-connorsrsi-b-i-1581332.html

Could you please explore writing a custom indicator for the ConnorsRSI.  I would also highly recommend that Worden explore including ConnorsRSI as a standard indicator within TC2000.

Thanks

It's early on a Sunday morning and I haven't had my coffee yet, but two thoughts ...

• I just downloaded the PDF so I haven't looked at the ConnorsRSI calculations yet, but if possible a PCF would be great.

• I know Worden has had Larry Connors do a Tuesday evening webinar or two in the past - perhaps Larry can do one on ConnorsRSI.

Time for coffee! :)

An Indicator Formula for RSI(Close,3) isn't necessarily simple, but we've posted about how to do this previously:

50 * (C - XAVGC5.1) / 3 / (.333363037576155 * (ABS(C - C1) + 2 / 3 * (ABS(C1 - C2) + 2 / 3 * (ABS(C2 - C3) + 2 / 3 * (ABS(C3 - C4) + 2 / 3 * (ABS(C4 - C5) + 2 / 3 * (ABS(C5 - C6) + 2 / 3 * (ABS(C6 - C7) + 2 / 3 * (ABS(C7 - C8) + 2 / 3 * (ABS(C8 - C9) + 2 / 3 * (ABS(C9 - C10) + 2 / 3 * (ABS(C10 - C11) + 2 / 3 * (ABS(C11 - C12) + 2 / 3 * (ABS(C12 - C13) + 2 / 3 * (ABS(C13 - C14) + 2 / 3 * (ABS(C14 - C15) + 2 / 3 * (ABS(C15 - C16) + 2 / 3 * (ABS(C16 - C17) + 2 / 3 * (ABS(C17 - C18) + 2 / 3 * (ABS(C18 - C19) + 2 / 3 * (ABS(C19 - C20) + 2 / 3 * (ABS(C20 - C21) + 2 / 3 * (ABS(C21 - C22) + 2 / 3 * (ABS(C22 - C23))))))))))))))))))))))))) + 50

An Indicator for the PercentRank(% Change, 100) is fairly long, but not particularly complicated:

50 * (SGN(C / C1 - C1 / C2) + SGN(C / C1 - C2 / C3) + SGN(C / C1 - C3 / C4) + SGN(C / C1 - C4 / C5) + SGN(C / C1 - C5 / C6) + SGN(C / C1 - C6 / C7) + SGN(C / C1 - C7 / C8) + SGN(C / C1 - C8 / C9) + SGN(C / C1 - C9 / C10) + SGN(C / C1 - C10 / C11) + SGN(C / C1 - C11 / C12) + SGN(C / C1 - C12 / C13) + SGN(C / C1 - C13 / C14) + SGN(C / C1 - C14 / C15) + SGN(C / C1 - C15 / C16) + SGN(C / C1 - C16 / C17) + SGN(C / C1 - C17 / C18) + SGN(C / C1 - C18 / C19) + SGN(C / C1 - C19 / C20) + SGN(C / C1 - C20 / C21) + SGN(C / C1 - C21 / C22) + SGN(C / C1 - C22 / C23) + SGN(C / C1 - C23 / C24) + SGN(C / C1 - C24 / C25) + SGN(C / C1 - C25 / C26) + SGN(C / C1 - C26 / C27) + SGN(C / C1 - C27 / C28) + SGN(C / C1 - C28 / C29) + SGN(C / C1 - C29 / C30) + SGN(C / C1 - C30 / C31) + SGN(C / C1 - C31 / C32) + SGN(C / C1 - C32 / C33) + SGN(C / C1 - C33 / C34) + SGN(C / C1 - C34 / C35) + SGN(C / C1 - C35 / C36) + SGN(C / C1 - C36 / C37) + SGN(C / C1 - C37 / C38) + SGN(C / C1 - C38 / C39) + SGN(C / C1 - C39 / C40) + SGN(C / C1 - C40 / C41) + SGN(C / C1 - C41 / C42) + SGN(C / C1 - C42 / C43) + SGN(C / C1 - C43 / C44) + SGN(C / C1 - C44 / C45) + SGN(C / C1 - C45 / C46) + SGN(C / C1 - C46 / C47) + SGN(C / C1 - C47 / C48) + SGN(C / C1 - C48 / C49) + SGN(C / C1 - C49 / C50) + SGN(C / C1 - C50 / C51) + SGN(C / C1 - C51 / C52) + SGN(C / C1 - C52 / C53) + SGN(C / C1 - C53 / C54) + SGN(C / C1 - C54 / C55) + SGN(C / C1 - C55 / C56) + SGN(C / C1 - C56 / C57) + SGN(C / C1 - C57 / C58) + SGN(C / C1 - C58 / C59) + SGN(C / C1 - C59 / C60) + SGN(C / C1 - C60 / C61) + SGN(C / C1 - C61 / C62) + SGN(C / C1 - C62 / C63) + SGN(C / C1 - C63 / C64) + SGN(C / C1 - C64 / C65) + SGN(C / C1 - C65 / C66) + SGN(C / C1 - C66 / C67) + SGN(C / C1 - C67 / C68) + SGN(C / C1 - C68 / C69) + SGN(C / C1 - C69 / C70) + SGN(C / C1 - C70 / C71) + SGN(C / C1 - C71 / C72) + SGN(C / C1 - C72 / C73) + SGN(C / C1 - C73 / C74) + SGN(C / C1 - C74 / C75) + SGN(C / C1 - C75 / C76) + SGN(C / C1 - C76 / C77) + SGN(C / C1 - C77 / C78) + SGN(C / C1 - C78 / C79) + SGN(C / C1 - C79 / C80) + SGN(C / C1 - C80 / C81) + SGN(C / C1 - C81 / C82) + SGN(C / C1 - C82 / C83) + SGN(C / C1 - C83 / C84) + SGN(C / C1 - C84 / C85) + SGN(C / C1 - C85 / C86) + SGN(C / C1 - C86 / C87) + SGN(C / C1 - C87 / C88) + SGN(C / C1 - C88 / C89) + SGN(C / C1 - C89 / C90) + SGN(C / C1 - C90 / C91) + SGN(C / C1 - C91 / C92) + SGN(C / C1 - C92 / C93) + SGN(C / C1 - C93 / C94) + SGN(C / C1 - C94 / C95) + SGN(C / C1 - C95 / C96) + SGN(C / C1 - C96 / C97) + SGN(C / C1 - C97 / C98) + SGN(C / C1 - C98 / C99) + SGN(C / C1 - C99 / C100) + SGN(C / C1 - C100 / C101)) / 100 + 50

RSI(Streak,2) is a problem however. We can create an Indicator Formula for Streak as long as we have a lookback which would bound the high absolute value of Streak available. If we used 10 for the lookback (so Streak can only range from -10 to 10), we would get a streak formula of:

ABS(C > C1) * (ABS(C1 <= C2) + ABS(C1 > C2) * (ABS(C2 <= C3) * 2 + ABS(C2 > C3) * (ABS(C3 <= C4) * 3 + ABS(C3 > C4) * (ABS(C4 <= C5) * 4 + ABS(C4 > C5) * (ABS(C5 <= C6) * 5 + ABS(C5 > C6) * (ABS(C6 <= C7) * 6 + ABS(C6 > C7) * (ABS(C7 <= C8) * 7 + ABS(C7 > C8) * (ABS(C8 <= C9) * 8 + ABS(C8 > C9) * (ABS(C9 <= C10) * 9 + ABS(C9 > C10) * 10))))))))) - (ABS(C < C1) * (ABS(C1 >= C2) + ABS(C1 < C2) * (ABS(C2 >= C3) * 2 + ABS(C2 < C3) * (ABS(C3 >= C4) * 3 + ABS(C3 < C4) * (ABS(C4 >= C5) * 4 + ABS(C4 < C5) * (ABS(C5 >= C6) * 5 + ABS(C5 < C6) * (ABS(C6 >= C7) * 6 + ABS(C6 < C7) * (ABS(C7 >= C8) * 7 + ABS(C7 < C8) * (ABS(C8 >= C9) * 8 + ABS(C8 < C9) * (ABS(C9 >= C10) * 9 + ABS(C9 < C10) * 10))))))))))

The problem is that this is too long to substitute for C into a 2-Period Wilder's RSI formula and get a formula which is short and fast enough to be practical or post in the forums. So we can only reasonably create two of the three components necessary to create the Connors RSI in TC2000 (it is possible to do this relatively easily in StockFinder however).

Bruce, thank you for your efforts!  Is it possible in the future that the ConnorsRSI could be programmed into TC2000 as a standard indicator?

I could not say if it will possible to add ConnorsRSI to TC2000 as a standard indicator. It would certainly seem to be possible from a technical standpoint, but it is a proprietary indicator.

Thank you for your suggestion. It has been assigned case number 9373.

I made a ConnorsRSI in StockFinder as follows

1. Create UpDown indicator as follows

Is it possible that this can be made without dragging and keeping all the indicators?

Please try the following RealCode Indicator. It will differ slightly from maljester's version, but the differences should not be too significant.

'|******************************************************************
'|*** StockFinder RealCode Indicator - Version 5.0 www.worden.com 
'|*** Copy and paste this header and code into StockFinder *********
'|*** Indicator:ConnorsRSI
'|******************************************************************
'# Cumulative
Static sumWeight(1) As Single
Static termRatio(1) As Single
Static weight As Single
Static num(1) As Single
Static den(1) As Single
Static ratio As New List(Of Single)
Static streak As Integer
If isFirstBar Then
	For i As Integer = 0 To 1
		sumWeight(i) = 1
		num(i) = 0
		den(i) = 0
	Next
	termRatio(0) = 2 / 3
	termRatio(1) = 1 / 2
	ratio.Clear
	streak = 0
	Plot = 50
Else
	Dim output As Single = 0
	Dim net As Single = Price.Last - Price.Last(1)
	weight = 1 / sumWeight(0)
	num(0) = num(0) * (1 - weight) + weight * System.Math.Max(0, net)
	den(0) = den(0) * (1 - weight) + weight * System.Math.Abs(net)
	sumWeight(0) = sumWeight(0) * termRatio(0) + 1
	weight = 1 / sumWeight(1)
	net = System.Math.Sign(net)
	If System.Math.Sign(streak) = net Then
		num(1) = num(1) * (1 - weight) + weight * System.Math.Max(0, net)
		den(1) = den(1) * (1 - weight) + weight * System.Math.Abs(net)
		streak += net
	Else
		num(1) = num(1) * (1 - weight) + weight * System.Math.Max(0, net - streak)
		den(1) = den(1) * (1 - weight) + weight * System.Math.Abs(net - streak)
		streak = net
	End If
	sumWeight(1) = sumWeight(1) * termRatio(1) + 1
	For i As Integer = 0 To 1
		If den(i) > 0 Then
			output += num(i) / den(i)
		Else
			output += .5
		End If
	Next
	Dim rank As Integer
	ratio.Insert(0, Price.Last / Price.Last(1))
	If ratio.Count > 1 Then
		For i As Integer = 1 To ratio.Count - 1
			rank += System.Math.Sign(ratio(0) - ratio(i))
		Next
		If ratio.Count = 101 Then
			ratio.RemoveAt(ratio.Count - 1)
		End If
		output += rank / ratio.Count / 2 + .5
	Else
		output += .5
	End If
	Plot = 100 * output / 3
End If
If isLastBar Then
	ratio.Clear
End If

Chart values are close to my version but filters run more than twice as fast with Bruce's version.  Great job Bruce.  Thanks so much for the RealCode.

Bruce

Does your second PCF percent rank

match the StockFinder calculaton?

I was looking at symbol PKI

on Dec 7 2012.

The max per up is 9.79

The min per down is neg 3.44

The Dec 7 close is neg .36

That would give a 100 day Stoc of about 23.3

Your PCF shows 35.35 on Dec 7 for PKI.

Thanks

It is a percentile rank (or a Wilder's Stochastic), not a normal Stochastic. It doesn't matter really where it falls in the range of the highest percent change to the lowest percent change. It matters how many percentage changes were lower, higher or the same.

If the 35.35 is correct what value do you get for the streak RSI and

the final Connors RSI for PKI on Dec 7?

Thanks

I get a 3-Period Wilder's RSI of Price of 15.56, a 2-Period Wilder's RSI of Streak of 3.26 and 100-Period Percentile Rank of 1-Period Price Percent Change of 35.35 for PKI on December 7, 2012.

Note that there are changes in the RealCode given in myTuesday, December 18, 2012 1:03:30 PM ET post since it was first posted. You may want to re-copy and paste the current version into StockFinder if you are having issues (the streak variable was not getting changed at each bar in the original version).

I saw a Connors video where they give

PKI 18.27

LOW 21.46

VFC 21.98

ABB 82.15

as the Connors RSI values for Dec 7 2012

and was trying to zero in on the numbers.

Technically the PKI close has 35 closes worse than it.

Im guessing the denominator changes to 99

because one of the 100 days was unchanged?

35 divided by 99 would give you 35.35

or is it because of something else?

Can you explain whats happening in your PCF

I dont use SGN that often.

Thanks

The Percent Change is higher than 35, the same as 1 (itself) and less than 64. I got my value for Percentile Rank by replacing:

Plot = 100 * output / 3

With:

Plot = 100 * (rank / ratio.Count / 2 + .5)

This was a mistake as ratio.Count would have been 100 when it was added to output, but had been reduced to 99 by the time of the Plot statement. The actual value at the time of the calculation would have been:

35.5

The .5 is a result of the tie with itself (an item that is part of the set is included in the percentile rank of that set).

SGN(C / C1 - C1 / C2)

Will return 1 if C / C1 > C1 / C2, 0 if C/ C1 = C1 / C2 and -1 if C / C1 < C1 / C2. We are doing this for 100 items (actually, the PCF only does so for 99 because it doesn't actually make the comparison with itself). The value will range from -99 to 99 (but could range from -100 to 100 if C / C1 could somehow <> C / C1).

Since we have a range of 200 points when we only want a range of 100 points, we divide the result by two to get a range of -50 to 50 instead of -100 to 100. Since we want the result to from 0 to 100 instead of -50 to 50, we add 50.

Well Bruce

you lost me with the tie with itself part and the could somehow part.

Sounds like you are always using 99?

I read the Connors PDF again and realized I made a mistake.

I had used the Dec 7 per change as part of the 100 days.

It says compared to the pervious 100.

Bringing the July 16 closing percent into the picture

makes my count 36.

36 closing percents of the previous 100 were lower than the Dec 7 percent.

Which for PKI gives me

RSIStreak 3.26

RSIPercent 36

RSI3 15.56

That comes out to 18.27 which matches Connors number exactly.

While 35.35 isnt much error as its one third of the final I feel better hitting the Connors number exactly.

I didnt expect this to have enough rounding error to miss by much.

I apologize if this was difficult to read as I get all types of gibberish when I use certain symbols.

Im trying to type as raw as possible.

Thanks

Well since the goal in to replicate ConnorsRSI and not necessarily to calculate Percentile Ranks correctly, I can certianly make those changes (which should be reflected in the PCFs and RealCode given above).

Bruce

this has nothing to do with Connors RSI

but can I have 3 more versions of your first RSI3 PCF

for 3 days back in time?

Based on C1 C2 and C3.

Thanks

3-Period Wilder's Smoothed RSI from 1-bar ago:

50 * (C1 - XAVGC5.2) / 3 / (.333363037576155 * (ABS(C1 - C2) + 2 / 3 * (ABS(C2 - C3) + 2 / 3 * (ABS(C3 - C4) + 2 / 3 * (ABS(C4 - C5) + 2 / 3 * (ABS(C5 - C6) + 2 / 3 * (ABS(C6 - C7) + 2 / 3 * (ABS(C7 - C8) + 2 / 3 * (ABS(C8 - C9) + 2 / 3 * (ABS(C9 - C10) + 2 / 3 * (ABS(C10 - C11) + 2 / 3 * (ABS(C11 - C12) + 2 / 3 * (ABS(C12 - C13) + 2 / 3 * (ABS(C13 - C14) + 2 / 3 * (ABS(C14 - C15) + 2 / 3 * (ABS(C15 - C16) + 2 / 3 * (ABS(C16 - C17) + 2 / 3 * (ABS(C17 - C18) + 2 / 3 * (ABS(C18 - C19) + 2 / 3 * (ABS(C19 - C20) + 2 / 3 * (ABS(C20 - C21) + 2 / 3 * (ABS(C21 - C22) + 2 / 3 * (ABS(C22 - C23) + 2 / 3 * (ABS(C23 - C24))))))))))))))))))))))))) + 50

3-Period Wilder's Smoothed RSI from 2-bars ago:

50 * (C2 - XAVGC5.3) / 3 / (.333363037576155 * (ABS(C2 - C3) + 2 / 3 * (ABS(C3 - C4) + 2 / 3 * (ABS(C4 - C5) + 2 / 3 * (ABS(C5 - C6) + 2 / 3 * (ABS(C6 - C7) + 2 / 3 * (ABS(C7 - C8) + 2 / 3 * (ABS(C8 - C9) + 2 / 3 * (ABS(C9 - C10) + 2 / 3 * (ABS(C10 - C11) + 2 / 3 * (ABS(C11 - C12) + 2 / 3 * (ABS(C12 - C13) + 2 / 3 * (ABS(C13 - C14) + 2 / 3 * (ABS(C14 - C15) + 2 / 3 * (ABS(C15 - C16) + 2 / 3 * (ABS(C16 - C17) + 2 / 3 * (ABS(C17 - C18) + 2 / 3 * (ABS(C18 - C19) + 2 / 3 * (ABS(C19 - C20) + 2 / 3 * (ABS(C20 - C21) + 2 / 3 * (ABS(C21 - C22) + 2 / 3 * (ABS(C22 - C23) + 2 / 3 * (ABS(C23 - C24) + 2 / 3 * (ABS(C24 - C25))))))))))))))))))))))))) + 50

3-Period Wilder's Smoothed RSI from 3-bars ago:

50 * (C3 - XAVGC5.4) / 3 / (.333363037576155 * (ABS(C3 - C4) + 2 / 3 * (ABS(C4 - C5) + 2 / 3 * (ABS(C5 - C6) + 2 / 3 * (ABS(C6 - C7) + 2 / 3 * (ABS(C7 - C8) + 2 / 3 * (ABS(C8 - C9) + 2 / 3 * (ABS(C9 - C10) + 2 / 3 * (ABS(C10 - C11) + 2 / 3 * (ABS(C11 - C12) + 2 / 3 * (ABS(C12 - C13) + 2 / 3 * (ABS(C13 - C14) + 2 / 3 * (ABS(C14 - C15) + 2 / 3 * (ABS(C15 - C16) + 2 / 3 * (ABS(C16 - C17) + 2 / 3 * (ABS(C17 - C18) + 2 / 3 * (ABS(C18 - C19) + 2 / 3 * (ABS(C19 - C20) + 2 / 3 * (ABS(C20 - C21) + 2 / 3 * (ABS(C21 - C22) + 2 / 3 * (ABS(C22 - C23) + 2 / 3 * (ABS(C23 - C24) + 2 / 3 * (ABS(C24 - C25) + 2 / 3 * (ABS(C25 - C26))))))))))))))))))))))))) + 50

Hi Everyone,

I'm not familiar with RealCode or Stock Finder, but if I can assist your efforts by supplying some ConnorsRSI values (or the values of the three components), please let me know.  Also note that you can always get current ConnorsRSI values by entering a ticker symbol on the free Analytics page on the Trading Markets web site: http://analytics.tradingmarkets.com/.

Regards,

Matt Radtke

Senior Researcher, Connors Research

Looks like folks have made progress in "StockFinder-izing" ConnorsRSI.

I've pretty much made the transition to TC2000 v12.3, though ... and it would sure be nice if ConnorsRSI would be available in the TC2000 Standard Indicator Library.

Recently ConnorsRSI has become available for AmiBroker, TradeStation, and Thinkorswim.

Surely the winner of S&C Reader's Choice for 20 years running can figure out a way to add it to their toolbox.

I've contacted LarryConnors in the past and he's not only okay with Worden adding ConnorsRSI to the TC2000 library, he even offered to do a webinar (Larry has done Worden webinars in the past).

Maybe Matt Radtke (above) can help with the calculation. Now that v12.3 is live, I'm sure the programmers are just sitting around with nothing to do :)

We are planning on adding ConnorsRSI to TC2000 version 12.3, I just could not say when it will happen.

Sweet!!

I'm doing okay without ConnorsRSI, but I'm always looking for ways to improve my P&L.

Tell you what Bruce ... I already circled April 26th & 27th on my calendar (when Julia, Gary, and Tal are coming to Chicago) ... put in a good word with the programmers and I'll not only buy Julia, Gary, and Tal a drink, I'll give them a little something to bring back for you ... just name it: Wine? Scotch? Maybe some cigars??

;)

Any update as to when Connors RSI will be added to TC2000 12.3?

I know we are planning on adding it, but I do not have a timetable on when exactly it will happen.

Any estimate for adding ConnersRSI?

I was told there were plans to add ConnersRSI, but I do not know when it will happen. The programmers have been busy with other stuff.

Just a suggestion...

Order some pizza for the programmers, lock them in a room, and don't let them out until the "opening 15 minutes of the trading day" bug is fixed.

I might even be willing to pop for some Chicago pizza. Everyone knows we have the best! Or I can treat Julia to some pizza when she's in town April 26th & 27th, and we can ship some back to Worden for the programmers :)

Hi Bruce,

I would appreciate it if you would supply the complete RSI(streak,2) formula representing the third component of the ConnorsRSI formula.

I already indicated in my Monday, November 26, 2012 12:46:12 PM ET post that this is not possible.

I understood from your post that the formula would be too long or practical to supply.  I would accept an email.  :-)  

I will not be sending the formula by email either. The formula is about 42000 characters long and will not even fit into the standard Windows clipboard (which is limited ot 32768 characters). While it could probably be shortened somewhat, I cannot think of a way to optimize it enough.

Wow!

I didn't realize that it would be that large.  Thanks for sharing your thoughts Bruce.  I'll wait for the coders to include it in TC2000.

Bruce:

on the 18th of dec. at 1:03 you posted the real code for connorsRSI use in Stockfinder. 

on the 20th of dec at 2:19 you posted some further/additional real code for connorsRSI

Can you please elaborate on the realtionship between these two postings?  Do they go together?

thanks.  kd

The RealCode in my Wednesday, December 19, 2012 3:09:07 PM ET was not intended to be used. It just explained how the percentile values from my Wednesday, December 19, 2012 11:45:43 AM ET post were extracted from the RealCode Indicator contained in my Tuesday, December 18, 2012 1:03:30 PM ET post.

Dear Bruce.

Since November 2012, many users are claiming Worden to includes "Connors RSI" as an standard indicator on TC2000.  Please try to do this. 

Thanks

Pat

Thank you for your suggestion.

Here's another vote for Worden to add ConnorsRSI.

I've been using various Connors' strategies for 5 years. Some of their strategies are a bit stale, but ConnorsRSI is hot now. And TC2000 users are late to the party.

As Putt4Dough pointed out almost 5 months ago, if we were using AmiBroker, TradeStation, thinkorswim (and others by now, now doubt), we wouldn't be standing around on the sidelines, watching other people improve their strategies and make money with ConnorsRSI.

We can stop harrassing Bruce about this. He gets it. I wonder if Worden management gets it, too.

Hope ConnorsRSI shows up in 12.4. 

Has it only been 5 months? Seems like forever :)

Fortunately my trading plan doesn't rely on ConnorsRSI, or the first few bars at the open.

If it's a matter of manpower, and Worden needs additional Software Ninjas, I might be available ... would love to golf w/ Tal in North Carolina :)

I have been using ConnorsRSI in SF5 for months.  Works great.