Trading with Market Statistics. IV Standard Deviation

[darthtrader 10]

That's correct Darth. NEWBIE would have not taken any trades today based on his present knowledge. But NEWBIE is about to graduate from trading 101 in the next thread. Based on the knowledge he is going to receive there, he would have made a killing today.

 

haha, nice...cha ching.

when your done with this thread you really need to consider the value of someone who would travel an hour to make your coffee, do your laundry and walk the dog in exchange for watching you trade

 

Dogpile, as someone who is trying to learn from Dalton basically at the same time as seeing this information I would be interested in what you think are the benefits of viewing TPO's vs this way of thinking. Honestly, I don't get it at all. Why a half hour vs 29 minutes? I think Nick pointed out in a response that Steidlmayer probly chose that because of computational concerns 20 years ago. Now that I'm starting to get this stuff, Mind Over Markets seems like a waste of time.

[Dogpile 10]

<<Now that I'm starting to get this stuff, Mind Over Markets seems like a waste of time.>>

 

Use the book for its concepts - it is excellent at that. For specific trade set-ups, Mind Over Markets is no good -- but that is not what it is trying to do -- MoM is about adding some 'context' to your existing processes.

 

Look at todays profile based on 30-min charts. What you see is what Dalton refers to as a 'b' profile. This indicates that the market is no longer 'trending' down -- longer-term buyers stepped in today. The market will generally auction the opposite direction following a 'b'... watch for this tomorrow. If it doesn't, then that might be due to new information and a new trending profile down or that might offer asymmetric location for a great trade back up... we will just have to see. A low above todays low would be idael location for a big move back up. If it just auctions straight up tomorrow, then we will have to find good spots on pullbacks to go long.

 

The media will be talking about how terrible today was -- but the 'b' profile is speaking to you if you are able to listen. It is very dangerous to be short at this point, in my opinion...

[jperl 11]

I caught this thread couple days ago and I cant stop reading and re-reading. I'm learning market Profile myself at the moment and you really have me jperl. This stuff looks like dynamite. But please, please dont tease me any more - I need part V

 

Keep it up - I'm loving it!

Jay.

 

Glad your learning something about market statistics Jay. Be careful however. Do not confuse Market Profile with what is being discussed here. They are not the same.

[Dogpile 10]

Market did it again today... Forms coil/balance/equilibrium around the VWAP level of 1482.25 -- notice how the std dev bands compress while price trades around the 82.25 pivot over and over. (sidenote: the PVP = VWAP here).

 

This coil-break however offered no clean bear flag/pullback pattern to enter on. Instead, it just dogpiled away from 1482.25.....

 

http://bp1.blogger.com/_5h-SWVGx6Ms/RqpaDuvXa9I/AAAAAAAAAWU/IjXqkPo2rH4/s1600-h/July+27+VWAP+Coil+Break.bmp

[walterw 10]

Market did it again today... Forms coil/balance/equilibrium around the VWAP level of 1482.25 -- notice how the std dev bands compress while price trades around the 82.25 pivot over and over. (sidenote: the PVP = VWAP here).

 

This coil-break however offered no clean bear flag/pullback pattern to enter on. Instead, it just dogpiled away from 1482.25.....

 

http://bp1.blogger.com/_5h-SWVGx6Ms/RqpaDuvXa9I/AAAAAAAAAWU/IjXqkPo2rH4/s1600-h/July+27+VWAP+Coil+Break.bmp

 

I like this bands contractions and expansions from vwap... Dogpile did you enter the break of the lower band ? could you remind me what formula you are using for this bands ? thanks Walter.

[Dogpile 10]

<<Dogpile did you enter the break of the lower band ?>>

 

I didn't, I was watching closely for a bear flag or some kind of readable pullback to form but it didn't happen. In retrospect, I should have put something on (partial) at that trendline break at least.... I did have a set-up on NQ trigger on a very similar set-up that I have been trading for a while though. However the range has been so huge lately, I didn't have the balls to take it. I should have done it in reduced size but froze up because it had already dropped so much at the point of entry and didn't want to go 'in the hole' 5-10 pts simply if a bear flag did form. What can you do? Sometimes the market just doesn't let you in at a reasonable point with relatively low risk....

 

 

<<could you remind me what formula you are using for this bands ? thanks Walter.>>

 

here is how it is written now... (note, I run it on a 2-min chart with 'custom session' start time of 5am. 45 bars of 2-mins gives you 90 minutes of time so I set the program to start drawing at 7am California Time -- 1/2 hour into trading -- so 1 hour of pre-market and 30-mins of actual trading starts the chart -- but I am not really using it for beginning of day -- btw, I am still learning this set-up)

 

value1=

square(c-vwap_h)+

square(c[1]-vwap_h)+

square(c[2]-vwap_h)+

square(c[3]-vwap_h)+

square(c[4]-vwap_h)+

square(c[5]-vwap_h)+

square(c[6]-vwap_h)+

square(c[7]-vwap_h)+

square(c[8]-vwap_h)+

square(c[9]-vwap_h)+

square(c[10]-vwap_h)+

square(c[11]-vwap_h)+

square(c[12]-vwap_h)+

square(c[13]-vwap_h)+

square(c[14]-vwap_h)+

square(c[15]-vwap_h)+

square(c[16]-vwap_h)+

square(c[17]-vwap_h)+

square(c[18]-vwap_h)+

square(c[19]-vwap_h)+

square(c[20]-vwap_h)+

square(c[21]-vwap_h)+

square(c[22]-vwap_h)+

square(c[23]-vwap_h)+

square(c[24]-vwap_h)+

square(c[25]-vwap_h)+

square(c[26]-vwap_h)+

square(c[27]-vwap_h)+

square(c[28]-vwap_h)+

square(c[29]-vwap_h)+

square(c[30]-vwap_h)+

square(c[31]-vwap_h)+

square(c[32]-vwap_h)+

square(c[33]-vwap_h)+

square(c[34]-vwap_h)+

square(c[35]-vwap_h)+

square(c[36]-vwap_h)+

square(c[37]-vwap_h)+

square(c[38]-vwap_h)+

square(c[39]-vwap_h)+

square(c[40]-vwap_h)+

square(c[41]-vwap_h)+

square(c[42]-vwap_h)+

square(c[43]-vwap_h)+

square(c[44]-vwap_h)+

square(c[45]-vwap_h);

 

value2=squareroot(value1/45);

value3=vwap_h+value2;

value4=vwap_h-value2;

value5=vwap_h+(value2*2);

value6=vwap_h-(value2*2);

 

 

 

if time>659 then begin

 

Plot1(value3, "UpperBand" ) ;

Plot2(value4, "LowerBand" ) ;

Plot3(value5, "UpperBand2" ) ;

Plot4(value6, "LowerBand2" ) ;

Plot5(vwap_h, "MidLine" ) ;

 

end;

[Dogpile 10]

btw, here was the one from the other day that I did take walter:

 

http://bp0.blogger.com/_5h-SWVGx6Ms/RqfSmOvXa6I/AAAAAAAAAV8/W9MY6-aw6wk/s1600-h/July+25+ES+Coil+Break.bmp

[walterw 10]

Thanks Dogpile for the formula... it looks like these bands have a very good S&R performance... cheers Walter.

[Dogpile 10]

Right now, I am using them strictly to help define times when the market maybe ready to begin to trend after a period of consolidation. As price moves away from VWAP -- the bands will widen. If the bands are starting from a tight spot -- then you may be able to catch a trend early-on and ride it for a bit. If the bands aren't tight, then maybe there are just no 'asymmetric' opportunities available and you whoul wait for a higher timeframe pattern to set-up.

[walterw 10]

Right now, I am using them strictly to help define times when the market maybe ready to begin to trend after a period of consolidation. As price moves away from VWAP -- the bands will widen. If the bands are starting from a tight spot -- then you may be able to catch a trend early-on and ride it for a bit. If the bands aren't tight, then maybe there are just no 'asymmetric' opportunities available and you whoul wait for a higher timeframe pattern to set-up.

 

 

Yes I like that squeeze... it creates oportunities... cheers Walter.

[nelo 10]

Darth,

 

you are not alone, I read the MoM and I am not much smarter from it. It is quite difficult for me to grasp just a little piece of useful information from it. What is useful can be certainly written on one-two pages, the rest is pure alchemy for me (yet). General MP theory as given by Dalton is without strong experience like rubber, you can exactly described every move by it, but only after it happened.

 

From my short experience with Jerry's method, the first SD from VWAP position is very close to VA in MP. Since it counts trades instead of time it is much more relevant to market nature (market is not moved or confirmed by seconds but by volume). One of the main difference is that MP is based on psychology while VWAP trading is based on statistics (which mirrors psychology of course).

 

Actually, Dogpile, today was a great selling day - the least expected happened. I guess because we are in a strong downside move.

[LS_Chad 10]

Here's the chart w/ definition for Investor/RT or MarketDelta users, including the VWAP along w/ the 1 & 2 Std Dev bands.

 

http://www.charthub.com/images/2007/08/09/VWAP_Bands_2.png

 

FWIW, I only read the first 15 or so posts of this thread...during which it was stated that it used the Standard Deviation of the VWAP when computing the bands. However, my best reconstruction (pretty well to the tick) of the chart in the video, was using a standard deviation of the 2-min closing prices. That probably has been mentioned in the posts I haven't yet read but wanted to make sure.

 

Regards,

Chad

[BlowFish 10]

Hi Jerry (and other standard deviants )

 

With respect to the SD bands. If the VWAP changes do you re-calculate the whole series with respect to the new VWAP or do you calculate the SD with the current point and the current VWAP?

 

i.e. is it the sum of the squares of price(n)-VWAP(n) or price(n)-VWAP(z) (where z is the last element of the series).

 

I think that's what the difference between the two methods I used here :-

 

http://www.traderslaboratory.com/forums/46/vwap-indicator-with-1sd-and-2sd-2175-3.html#post17254

[jperl 11]

Hi Jerry (and other standard deviants )

 

With respect to the SD bands. If the VWAP changes do you re-calculate the whole series with respect to the new VWAP or do you calculate the SD with the current point and the current VWAP?

 

i.e. is it the sum of the squares of price(n)-VWAP(n) or price(n)-VWAP(z) (where z is the last element of the series).

 

Nick, the new SD is computed using the new VWAP. See post 14694 for an example.

[BlowFish 10]

Thanks, I figured that was the case just wanted to be sure. I have used and alternate algorithm from the signal processing world. I'll post in DBtinas thread but would welcome your expert opinion. It has a couple of advantages.

 

Cheers,

Nick.

[BlowFish 10]

OK I've been tearing my hair out with this. Basically I coded all 4 different text book algorithms for calculating Variance. See

 

http://en.wikipedia.org/wiki/Algorithms_for_calculating_variance

 

I added in the volume weighting (using a couple of methods) but could make none match exactly with method one. I hesitate to say this but I wonder if multiplying the squares by the barvolume/total sample volume is mathematically 'valid'? The more pressing question is should the ratio Bar Vol/Tot Vol use the complete sample size volume for its calculation I guess the answer is yes again. Does it make sense (math wise) to multiply the Square by price too?

 

I favour the last algorithm however the bands seem slightly narrower than looping through the sample each time. X is the price and mean is the VWAP which plots perfectly.

 

n = 0

mean = 0

S = 0

 

foreach x in data:...//i.e each time we get a new bar

..n = n + 1

..delta = x - mean

..mean = mean + delta * Weight

..S = S + delta*(x - mean) ........... // Note This expression uses the new value of mean

end for

 

variance = S/(n - 1)

 

Anyone have any ideas? BTW I have lots of great code but am loath to post it until I resolve this issue. BTW above Weight is Volume/TotalVolume. BUT Total Volume is the total for the series to that point. I am convinced if it is mathematically valid there must be a way to calculate variance correctly as each bar arrives without scaning the whole series. I can do that for the VWAP and a regular SD its just the weighting that causes issues.

 

Cheers,

Nick.

 

Cheers.

[jperl 11]

I hesitate to say this but I wonder if multiplying the squares by the barvolume/total sample volume is mathematically 'valid'?

If you want to weight the variance computation by the volume, that's exactly what you would do as described in [thread=2101]Part IV[/thread].

Think of it this way. A 10 contract trade can be thought of as ten 1-contract trades. To compute the variance you would have to include all 10 contract trades in the variance computation. That's identical to multiplying the square by 10.

 

 

The more pressing question is should the ratio Bar Vol/Tot Vol use the complete sample size volume for its calculation I guess the answer is yes again.

correct

 

Does it make sense (math wise) to multiply the Square by price too?

no

[BlowFish 10]

Hi jerry,

 

The last question was meant to be "does it make sense to multiply the square by volume as well". Multiplying the price by normalised volume makes sense as you illustrated above it is the whole basis for the VWAP. You can see what it actually is and why its done. Multiplying the square by the normalised volume I can't get my head round. The square is just an intermediate number used in the calculation. In fact if you are multiplying the sum by vol and the sum of the squares when you take the difference they largely cancel out (but not completely). Why not divide the sum by normalised volume or multiply and then square?

 

The other thing is that all of the ways of calculating variance (except just re totalling the series when you get a new data point) seem to produce a slightly different (but consistent result). I presume there are proofs for all these formula adding in the weighting seems to break them all. Unless I am doing it wrong which is highly likely.

 

Still kinda tearing my hair out.

 

EDIT: Another thought wherever you normalise price with the VWAP (Price-VWAP) You are taking into account the volume as it is already factored into the VWAP. I wish my maths was better to actually work through the proofs.

[jperl 11]

Hi jerry,

 

Multiplying the square by the normalised volume I can't get my head round. The square is just an intermediate number used in the calculation. In fact if you are multiplying the sum by vol and the sum of the squares when you take the difference they largely cancel out (but not completely). Why not divide the sum by normalised volume or multiply and then square?

 

I looked at some of those algorithms. With an unweighted variance, they are simple. The computation with a weighted variance is a messy business and not worth the effort. I don't think they will improve computation speed.

[BlowFish 10]

Jerry,

 

Thanks for your indulgence on this. I appreciate it. On the plus side I understand a lot more about the math behind things! (Clearly still not enough). I am hunting for an arithmetician as I type.

 

Let me just say how I got started on this. It was precisely because of performance issues. Basically using the code by Dbtina and Nick the code was too slow when loading on a tick chart. It takes about 10 minutes to process a few days of ticks. Once running its OK but for every tick it needs to process every bar back to the start of the day. Having re-done the PvP code (that used to have the same speed issues) and getting pretty massive performance gains, I figured I would try this. A lot tougher!

 

Actually on YM the various results are close enough and I think the difference can be attributed to rounding errors (again the non-iterative code is superior in this respect). I'll post a couple of screens and challenge anyone to detect which is which. I would have said I was done...except....

 

The DAX subjectively looks 'different'. The lines track but sometimes end up a handful of ticks apart. Its perplexing.

 

I think I'll go ahead and 'publish' I reckon its solid code and there are a couple of neat bits.

 

Next thing a super efficient histogram (again the plan is to make it tick precise and perform OK) I have a couple of ideas how to approach that without iteration too.

 

So which is the real SD blue plot or yellow plot?:) Actually I think I can spot them apart but i have been staring at these things for 2 days now.

[BlowFish 10]

Here is the dax seems the difference is more pronounced. (gives the game away) I am not sure why.

[dbntina 10]

Blowfish,

 

How did you get the volume histogram...is that multicharts or something else?

 

dbntina

[BlowFish 10]

Yes it is. Like most things in multicharts its clunky or not all there. For example you can only select nnnn bars, whole series, or screen for the bars to factor into your calc (no dates). So for todays values you have to constantly mess around re scaling and scrunching up bars. Seems like they half implemented it so they could tick the box then gave up on. This seems a common philosophy over at TSSuport (the supplier).

[dbntina 10]

Oh..Okay I almost got really excited!

 

dbntina

[BlowFish 10]

No takers on the which plot is which? I thought you might have a go DBNTina as you must know this stuff quite well Do they both have merit?

 

Cheers.