Jump to content

Welcome to the new Traders Laboratory! Please bear with us as we finish the migration over the next few days. If you find any issues, want to leave feedback, get in touch with us, or offer suggestions please post to the Support forum here.

  • Welcome Guests

    Welcome. You are currently viewing the forum as a guest which does not give you access to all the great features at Traders Laboratory such as interacting with members, access to all forums, downloading attachments, and eligibility to win free giveaways. Registration is fast, simple and absolutely free. Create a FREE Traders Laboratory account here.

dbntina

VWAP Indicator with 1SD and 2SD bands

Recommended Posts

Guys I am not a TS coding guru by any means...still new at it but wanted to share this code if anyone else was interested. Just wrote this code for myself to keep up with JPERL's threads on Market Statistics. It plots the VWAP, and 1st and 2nd Standard Deviation bands. It lines up with his numbers within a tick or two so I think it is working correctly.

 

Any other TS coders out there please take a look because it can probably be improved upon.

 

This is my first post of a file so if I screwed something up...be gentle!

 

Hope this helps,

 

dbntina

boxmeister :)

DBVWAP_SD.ELD

Share this post


Link to post
Share on other sites

Good post mate. I've been wanting to do this since reading Jpearls threads. I am however using ninja trader, so (if you dont mind) Im going to try and convert your code to ninja script.

 

I dont think there are too many ninja users out there - everybody seems to use esignal or tradestation!

 

Cheers, Jay.

Share this post


Link to post
Share on other sites

Hi DbnTina,

I am not sure if the calculation for vi is correct.

In your EL you have used

variance_i = ((UpTicks[Value1]+DownTicks[Value1])/ShareW) * (Square(AvgPrice-VolWAPValue));

 

where (UpTicks+DownTicks) represents volume of the bar ... I think in a probability calculation one should use total volume at price i, not just volume of the bar i. Lets ask Jerry for clarification.

Share this post


Link to post
Share on other sites

Hi DbnTina,

I am not sure if the calculation for vi is correct.

In your EL you have used

variance_i = ((UpTicks[Value1]+DownTicks[Value1])/ShareW) * (Square(AvgPrice-VolWAPValue));

 

where (UpTicks+DownTicks) represents volume of the bar ... I think in a probability calculation one should use total volume at price i, not just volume of the bar i. Lets ask Jerry for clarification.

 

 

 

 

NickM001,

 

You are correct. In the code, ShareW is always the total share volume up to the current time from your start. You divide the volume of the current bar by the total volume and use this "normalized volume" to multiply with according to Jerry. This should be correct if I am understanding the formula correctly. It also seems to line up with Jerry's bands so I think it is correct.

 

Let me know...I can correct if I am not suppossed to normalize this way...

 

Thanks for taking a look at the code if it is incorrect we want to find out now.

 

dbntina :)

Share this post


Link to post
Share on other sites

Hi dbntina,

 

Thank you for your contribution. If you happen to need any updating on the code, threads here can be edited at anytime so please feel free to do so. Also, would it be possible to add a screenshot of the indicator in action? It would help traders give a visual clue on what the indicator is all about. Thanks.

Share this post


Link to post
Share on other sites

OK let me try to explain again my understanding of the formula. Let quote what Jerry wrote in his explanation:

QUOTE#######################################

While we won't address all these questions in one thread their answers can be obtained by analysis of the volume distribution function. To do so requires that we introduce a third property of the volume distribution function called the Standard Deviation of the VWAP, SD for short. SD is computed from the following equations:

 

NOTE : Formula did not copy ... see http://www.traderslaboratory.com/forums/f6/trading-with-market-statistics-iv-standard-2101.html

 

where the summation subscript i, runs over all prices in the volume distribution

pi = ith price in the volume distribution

Pi = vi/V is the probability of occurrence of price pi

vi = the volume traded at price pi from the volume distribution

V = total volume for the entire distribution

#################################

 

Note that factor vi is volume traded at price pi. If I understand your code, you have used volume of the bar, or volume at time , not as described in the formula.

 

To get proper value for vi, you would have to keep track of all the volume distribution at EACH price level of the range in the array and use the value of volume array element that corresponds to price pi. It can be done, but it would take more then few lines of code, unfortunately.

 

For some odd reason, it does not seem to make much difference even if you set vi/V = 1, for the few sample charts I looked at. So for all practical purposes, SD value is usable...

Share this post


Link to post
Share on other sites

I think it's considered an 'acceptable' trade of computational complexity/speed against accuracy.

 

You can control the accuracy by putting a second series of lower timeframe bars and applying the indicator to that perhaps? (hide the second series). A 1 tick series should be completely accurate?

 

Just a thought.

 

Oh and thanks Dbntina!

Share this post


Link to post
Share on other sites

Nick,

 

You are exactly correct...I know what you are getting at for sure. Blowfish is exactly right...the code I wrote on a 10M bar will do what you are getting at but only at 10M increments. A 5M bar will be more accurate...then a 3M bar will be even more accurate...then a 2M bar will be even more accurate...etc. Jerry is content using a 2M bar. Technically you would have to run it how you are describing to get it exact but I am not sure that it is necessary according Jerry he thinks running it on a 2M bar is close enough.

 

I understand though what you are saying because if I remember my match classes correctly it is a continuous function not a discrete function? I am approximating (so is Jerry I believe) using calculations at every 2 minutes (discrete) when the true value would be evaluated constantly like a continuous function.

 

We will get more and more accurate as will decrease our bar interval. But knowing that the bands are truly at .38 on the S&P as oppossed to having our numbers 1 to 2 ticks off (I don't know if this is the case but I am just giving an example) I think it is close enough on a 2M according to Jerry.

 

Let me know if I am misunderstanding what you are getting at...the important thing is that the code is doing the same thing as Jerry's code on a 2 minute chart...that was all I was trying to accomplish so I could follow Jerry's logic and trade management.

 

dbntina :)

Share this post


Link to post
Share on other sites

I don't want to beat this to death, since the code is performing close enough to what Jerry has. However, it is only important if you decide to expand the code to include volume distribution. At that point you will realize that assumptions made in this code are not valid in all cases. It is accurate observation that the smaller the time frame is, the more accurate calculation becomes, simply because there is an underlying assumption that the volume was constant within the bar interval ( we know the total volume of the bar, and range of the bar). Calculation will be accurate as long as the price does not hit the same level more then once. Now suppose that we have a consolidation day when the price moves back and forth several times thru the range. Every time it comes to the price pi, you will have to add volume of that instance to previous volume at the same price and build volume histogram bar for the price pi. Now when you do calculation for the SD, you would take sum of the volume at price pi and divide it with total volume for the session . I hope that this explanation is better then previous attempts.

Share this post


Link to post
Share on other sites

Nick,

 

I hope I didn't mislead anyone. I should have been more clear. The indicator I posted does not give the exact VWAP or SD bands. It most definitely is an approximation. I am only trying to create the identical indicator Jerry uses on his 2min charts for TS users.

 

It should be the exact way that Jerry told us to caculate it. That is all I am trying to accomplish. He said to take the open/hi/low/close and divide by 4, for the price bar...that in itself already throws some level of accuracy out the window. However I am normalizing the volume according to the way he said to do it I think.

 

I would like people to check the code for accuracy according to the way Jerry does it...not for exact accuracy of VWAP and SD bands...I already know it is not to that level of accuracy.

 

Hopefully everyone is clear on what the indicator is doing now. But I agree with you it is not accurate the way you are talking about...that was not my intention. My intention was to create Jerry's indicator for TS users.

 

Hope that clears it up,

 

Thanks,

 

dbntina :)

Share this post


Link to post
Share on other sites
Soultrader,

 

I have the screenshot. Can you point me to the instructions on how to add the image to the eld?

 

Thanks,

 

dbntina

 

Hi dbntina,

 

You should be able to edit your first post and then edit the attachment manager. Pm me if you are still having problems.

Share this post


Link to post
Share on other sites

Soultrader, I still can't find how to edit my original post...having said that...

 

I have coded the VWAP, SD bands and PVP on a tick by tick basis on a 1 Tick chart. I was thinking I could simply hide the 1 tick chart and plot the indicators on a 2 minute chart and everything would be cool...problem is that it gives me an error "Tick and Volume intervals cannot be used in a mult-symbol chart".

 

Does anyone know of a way around this to plot the data gathered from the 1Tick to plot on a minute chart? There has to be a way around this. Maybe creating a function and access...I don't know just thinking out loud...like I said I am new to the functionality of TS and need help.

 

I will post for all TS users for free if we can get this thing figured out.

 

Any help is appreciated. This should be the fix that NICK was pointing out for accuracy in the indicator thread.

 

dbntina ;)

Share this post


Link to post
Share on other sites

You would put the 1 tick series on the chart add the indicator on that series then, then hide the initial 1 tick series.

 

Simply add a second 2 5 10 15 minute data series of the same instrument (whatever you want to trade from) to the chart.

 

Cheers,

 

PS. have you posted the PvP yet I could not see it

PPS. Great work!! Thanks you (if I had not already thanked you).

Share this post


Link to post
Share on other sites
Guest cooter
You would put the 1 tick series on the chart add the indicator on that series then, then hide the initial 1 tick series.

 

Simply add a second 2 5 10 15 minute data series of the same instrument (whatever you want to trade from) to the chart.

 

Cheers,

 

PS. have you posted the PvP yet I could not see it

PPS. Great work!! Thanks you (if I had not already thanked you).

 

That doesn't work on Tradestation as you can't mix and match tick and minute data series on the same chart.

 

Try ADE (all data anywhere) as a workaround. You'll find more info about it in the Tradestation.com forums.

Share this post


Link to post
Share on other sites

Does anyone here use Amibroker for their charts? The programming language seems very flexible, I just wish I knew how to program! If someone could convert these for that program I would be eternally grateful. Specifically the standard deviations for the VWAP. Thanks!

Share this post


Link to post
Share on other sites

Thanks Cooter and Blowfish for your help...finally got ADE working and I can see the PVP on my minute charts with the VWAP and SD bands...appreciate the knowledge on this board!

 

Testing it out now for a few days.

 

dbntina

Share this post


Link to post
Share on other sites

Could someone tell me which of these SD bands look correct (yellow or blue). Actually I cant work out why they differ as the actual formula I use is the same in both, I just am summing the bars slightly differently.

 

Be really great full if someone could compare with ensigns finishing values. Note that the YM chart ends at 3.15pm central

 

Cheers.

Nick

SDBands.thumb.png.8098b835c0573511cf6875db1834124e.png

SDBands1.thumb.png.f879a537f3f61126099248dc3f015ef0.png

Share this post


Link to post
Share on other sites

OK figured out what was wrong with the last two indicators but have come up with something that I think is OK.

 

The reason for this is that there are some performance issues with the indicator here as it has to scan the series back to the days start when there is a new tick. Applying it to a tick chart (especially at the end of the day) it slows right down.

 

Anyway I figured there should be a way to do it using running statistics and indeed there is. There is a side benefit to doing it this way as you don't get cumulative rounding errors (according to the signal processing guys). This does not get rid of the sampling inaccuracy that Nick mentions earlier in the thread.

 

I'd welcome comments? I wonder if rounding errors would explain the 2 or 3 ticks difference?

SDBands2.thumb.png.de3c7b969a32149f625f725f3d8a61dc.png

Share this post


Link to post
Share on other sites

One last chart a 1 tick of the same period as the last post (yesterdays ER2). Looks good to me? It is much much quicker this way and its plausible to run it on 1 tick charts (meaning complete accuracy is attainable if desired). If it looks good enough I'll upload it.

 

Incidentally watching PvP for the last week or two and comparing tick precise with 1 & 2 minutes I think I might have changed my mind - tick precise would have kept you out of a couple of bad trades due to quicker 'flipping'

 

Cheers

SDBands3.thumb.png.e2771bb7f91a4cf90df76e235f6580ac.png

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

×

Important Information

By using this site, you agree to our Terms of Use.