Trading with Market Statistics X. Position Trading

[TheAnalyst13 10]

  jperl said:

The most likely problem is the approximation used to compute the VWAP as well as the histogram. Ideally, the VWAP should be computed from data for every trade. In practice this is not done. As a good approximation, the VWAP is computed from the price of every bar and its volume. The error in this approximation gets larger the larger the time frame of the chart. As the time frame of the chart gets smaller, the error diminishes. So if you compare a 30 min, with a 5 min and 1 min chart you should see the VWAP converging to the "correct" value.

 

1) DO you mean then the 1 min VWAP should be the most correct one then?

2) What about the PVP?

 

Thanks, again Jerry you are my hero. Since your threads I have been trading futures again combined with market internals very successfully.

[TheAnalyst13 10]

Sorry, I forgot one more question. What is VPOC, VWTPO, VOC? I just switch using Ninjatrader and it has an indicator avaiable that displays these 3 and TPOs. I was wondering which one could be Volume Profile.

 

Thanks, Jerry.

[BlowFish 10]

  TheAnalyst13 said:

Jerry, thanks for your wonderful thread. i have been using market statistics and I am learning everyday since I started reading your thread. I have a few problems though that I was wondering if you can answer me.

I use Zenfire with Ninjatrader and there are two VWAP indicators that have been programmed and are avaiable for everybody. My problem is when I change the interval on the chart from let's say 30 min to 5 min, or 1 min that the VWAP changes value and as far as I undestand that should not be the case right? The second thing is that my PVP changes as well when I change the interval. So I just wanted to confirm with you that this should not be the case and I got a problem then and cannot trade with this tools right?

 

Thanks, M.

 

As Jerry says the inconsistencies come from averaging the volume in a bar to an average price for that bar. It's a data sampling issue if you like (a bar is a sample).

 

A further observation, with the VWAP the difference is not to significant. After a day there may be a tick or two difference between a 1 min chart and a 10 min chart. With the PVP the issue can be much greater. If you look at how the PVP is calculated you can see that it will jump when there is a new peak. You may get a new peak when there is not one due to averaging or you might not get a new peak when there should be one due to averaging. Keeping an eye on the volume profile might alert you to this possibilities (two peaks of similar magnitude).

 

If you are bothered about accuracy use a small time frame chart and transfer the lines to your trading chart. I always have a 1 tick chart scrunched up in the background to know precise values. Sadly when I tried this with NT it choked as this can be processor intensive.

[jperl 11]

  TheAnalyst13 said:

Sorry, I forgot one more question. What is VPOC, VWTPO, VOC? I just switch using Ninjatrader and it has an indicator avaiable that displays these 3 and TPOs. I was wondering which one could be Volume Profile.

 

Thanks, Jerry.

 

Well, since you are a Ninjatrader user, you will find discussions of these on the Ninjatrader forums here:

 

Calculate Value Area of volume at price - NinjaTrader Support Forum

 

and

 

 

here:

 

Enthios live index futures trading

[Head2k 10]

  BlowFish said:

If you are bothered about accuracy use a small time frame chart and transfer the lines to your trading chart. I always have a 1 tick chart scrunched up in the background to know precise values. Sadly when I tried this with NT it choked as this can be processor intensive.

With 1R (1 tick range) chart you get almost the same precision calculated from much less data points than if you use 1 tick chart.

 

I have observed that running computation on time charts or larger range charts is generally less precise than using CVB or tick charts. The reason is obvious. Bars on CVB and tick charts form in relation to activity, while in time bars the activity within the bar is much less likely to be evenly distributed.

So if one wants to calculate developing value areas from less data points but still with high level of precision, he should use CVB (or tick) charts.

[TheAnalyst13 10]

  Head2k said:

With 1R (1 tick range) chart you get almost the same precision calculated from much less data points than if you use 1 tick chart.

 

I have observed that running computation on time charts or larger range charts is generally less precise than using CVB or tick charts. The reason is obvious. Bars on CVB and tick charts form in relation to activity, while in time bars the activity within the bar is much less likely to be evenly distributed.

So if one wants to calculate developing value areas from less data points but still with high level of precision, he should use CVB (or tick) charts.

 

What do you mean with CVB charts?

 

Thanks

[BlowFish 10]

  TheAnalyst13 said:

What do you mean with CVB charts?

 

Thanks

 

Constant Volume Bars. Each bar is nn contracts.

[TheAnalyst13 10]

Thanks blowfish.

 

Another questions: Does anybody use these entry (VWAP & PVP) techniques for stocks?

[TheAnalyst13 10]

  jperl said:

Position Trading is generally described as a trade which you enter and expect to hold for a considerable period of time during the day. Such a trade can be entered at any time after the open. My personal preference for a position trade is at the beginning of the trading day using market statistics from the previous day as my guide for determining entry, profit target, stoploss and scale in points if necessary. The direction of the trade is based on interpretation given in the last 9 "Trading with Market Statistics" threads but using the previous days statistics as the starting point. Position trading is thus no different than any other type of trading that I have previously described.

 

Here is the idea:

 

a)Set up a chart with yesterdays volume histogram, PVP, VWAP and SD's on it. Leave sufficient room to the right of yesterdays close so that at the open you can continue to add to the statistical data as todays market begins to unfold. In effect you are continuing to update yesterdays volume distribution as more data is added to the chart.

 

b)Before the open, decide on your trading plan. Pick a direction for the trade, an entry point, profit target and stoploss based on what you see in the volume distribution function. It will help to reread the previous threads to determine what you should be looking for.

 

c)When the market opens, execute the plan.

 

In the following video on trading the ER2 (Emini Russell 2000), you will see that the previous days volume distribution ended the day in a symmetric state with the VWAP = PVP. I then concluded that I should look for a countertrend trade back toward the VWAP as described in [thread=2285]"Trading with Market Statistics Part VIII"[/thread].

 

Watch the video to see what I did on September 06, 2007.

 

ER2PostionTradeSep06

 

This trade was a good position trade which would have been even better if I had traded more than one contract. After having climbed up to the 2nd SD above the VWAP, the price action continued on down below the VWAP to the 1st SD and then evenutally to the 2nd SD, a very typical signature of a symmetric distribution.

 

I have several questions here:

 

1) You said that you setup a chart with yesterdays Volume Distribution function and PVP, SDs. So when today the market opens is the data added to yesterday's function (sorry english is not my first language, if the answer is already in your post). If this is not possible with Ninjatrader, can I just use yesterdays PVP, VWAP and SDs static points and extend these point as S/R lines as you showed in your chart?

 

2) So just to be sure if I understood it right. For position trading the rules are:

 

• When yesterday's VWAP = yesterday's PVP = no skew, I look for countertrend trades towards the yesterday's VWAP at yesterday's Standard Deviations. Or also I can breakout trades as previously posted in a thread.
  
• today's price open above y. VWAP and y. VWAP > y. PVP: take long trades or
  
• today's price action is below y. VWAP and y. VWAP < PVP: take short trades
  

 

3) If you do add today's data to yesterday's distribution when do you switch over to use only today's volume distribution function.

 

Thanks, M.

[jperl 11]

  TheAnalyst13 said:

 

1) You said that you setup a chart with yesterdays Volume Distribution function and PVP, SDs. So when today the market opens is the data added to yesterday's function (sorry english is not my first language, if the answer is already in your post). If this is not possible with Ninjatrader, can I just use yesterdays PVP, VWAP and SDs static points and extend these point as S/R lines as you showed in your chart?

 

If the question is "Can you", the answer is yes. If the question is "should you", the answer is no. Yesterdays VWAP and SD's will change dynamically as you add more data for today. Using yesterdays data as a static VWAP isn't going to help you much.

  TheAnalyst13 said:

2) So just to be sure if I understood it right. For position trading the rules are:

 

• When yesterday's VWAP = yesterday's PVP = no skew, I look for countertrend trades towards the yesterday's VWAP at yesterday's Standard Deviations. Or also I can breakout trades as previously posted in a thread.
  
• today's price open above y. VWAP and y. VWAP > y. PVP: take long trades or
  
• today's price action is below y. VWAP and y. VWAP < PVP: take short trades
  

• You have it about right. These are however not rules. Simply one way of interpreting the data.
  

  TheAnalyst13 said:

3) If you do add today's data to yesterday's distribution when do you switch over to use only today's volume distribution function

That's your choice. You don't have to switch. If you are trading against yesterdays data + todays, you are looking for market moves of 1 SD within that data set. If you are trading against todays data only, you are looking for market moves against todays data set. You choose.

[BlowFish 10]

I am sure Jerry will correct me if I am wrong but yesterdays static VWAP SD's etc. can act as HuP's? They certainly appear to.

[TheAnalyst13 10]

  jperl said:

If the question is "Can you", the answer is yes. If the question is "should you", the answer is no. Yesterdays VWAP and SD's will change dynamically as you add more data for today. Using yesterdays data as a static VWAP isn't going to help you much.

 

• You have it about right. These are however not rules. Simply one way of interpreting the data.
  

 

That's your choice. You don't have to switch. If you are trading against yesterdays data + todays, you are looking for market moves of 1 SD within that data set. If you are trading against todays data only, you are looking for market moves against todays data set. You choose.

 

Thank you Jerry.

In your video, did VWAP etc chantge at all? Or were they static?

[BlowFish 10]

  TheAnalyst13 said:

Thank you Jerry.

In your video, did VWAP etc chantge at all? Or were they static?

 

I think you will find they are dynamic lines from a 2 day sample (yesterday with todays being added to the sample). The horizontal lines might be a little confusing (suggesting static) but they actually are 'projections' of the unfolding VWAP/SDs. You will see they (the horizontal lines) have moved at the end of the video from where they started at the beginning of the day. They correspond to the current values of a 2 day VWAP/SD set. Jerry also mentions moving the target up as the VWAP develops. I seem to have been overly verbose:)

[BlowFish 10]

  BlowFish said:

I am sure Jerry will correct me if I am wrong but yesterdays static VWAP SD's etc. can act as HuP's? They certainly appear to.

 

I think I may well be mistaken. Reviewing Market Statistics XI maybe Jerry would not classify previous days static lines as HuP's. Having said that they often do appear to hold up price. Of course early in the day they will be close to the dynamic ones until new data is added to the set.

[BlowFish 10]

  jperl said:

That's your choice. You don't have to switch. If you are trading against yesterdays data + todays, you are looking for market moves of 1 SD within that data set. If you are trading against todays data only, you are looking for market moves against todays data set. You choose.

 

Hi Jerry,

 

Presumably you can use any of the principles described in earlier threads with a two day data set?

 

Presumably you can use even larger data sets (5 days for example) provided you are comfortable with the risk?

 

Finally do you have favourite data set periods and favourite trades? I must admit that originally I was attracted towards the scalp type trades but now find the larger data sets interesting. Maybe it's because volatility died down somewhat.

 

I'm still impressed with how elegant and flexible what you have presented is.

 

Edit: One other thing occurs to me your statement above seems to imply that you only ever consider one set at a time? Is that set in stone? I must admit that I like to see things 'aligned' though can lead to indecision when they are not.

[jperl 11]

  BlowFish said:

 

Presumably you can use any of the principles described in earlier threads with a two day data set?

Yes of course

  BlowFish said:

 

Presumably you can use even larger data sets (5 days for example) provided you are comfortable with the risk?

Yes. The longer the time frame the larger the SD and therefore the greater the risk.

 

  BlowFish said:

Finally do you have favourite data set periods and favourite trades?

I trade the Russell 2000 index futures almost exclusively. I will always have yesterdays data set up and todays.

 

 

  BlowFish said:

One other thing occurs to me your statement above seems to imply that you only ever consider one set at a time? Is that set in stone? I must admit that I like to see things 'aligned' though can lead to indecision when they are not.

 

The question of how many HUPS to put on your trading data set is subjective. Too many and you can get confused; too few and you may miss important ones. If I'm trading just today, I will always have yesterdays data set HUPS on my chart( which will continue to change dynamically as todays data gets added to it). But I may also include additional VWAP data from 1 week, 1 month, 1 year ago if they happen to be nearby the price action.

Hope that helps.

[TheAnalyst13 10]

Blowfish or Jperl,

 

I was wondering if any could help me out with which type of Profile I should choose in order to replicate Jperl's approach to markets correctly. I use Ninjatrader and there are various Profile indicators based on Volume, TPOs etc.

 

Here are the ones I can choose from that a programmer wrote in the NT Forum:

 

VOC - Volume at Close. Only maps volume at the close tick per bar.

VTPO - Volume Time Price Opertunity. Evenly distributes Volume accross the ticks per bar. (TPO is realy just another phrase for POC in this context).

VWTPO - Volume Weighted. Distributes The Volume accross the ticks of a bar on a weighted basis (larger bars get less volume per tick, vs. smaller bars with the same volume.

TPO - only price is used.

 

Thanks for the help.

[BlowFish 10]

  TheAnalyst13 said:

Blowfish or Jperl,

 

I was wondering if any could help me out with which type of Profile I should choose in order to replicate Jperl's approach to markets correctly. I use Ninjatrader and there are various Profile indicators based on Volume, TPOs etc.

 

Here are the ones I can choose from that a programmer wrote in the NT Forum:

 

VOC - Volume at Close. Only maps volume at the close tick per bar.

VTPO - Volume Time Price Opertunity. Evenly distributes Volume accross the ticks per bar. (TPO is realy just another phrase for POC in this context).

VWTPO - Volume Weighted. Distributes The Volume accross the ticks of a bar on a weighted basis (larger bars get less volume per tick, vs. smaller bars with the same volume.

TPO - only price is used.

 

Thanks for the help.

 

I am not familiar with these indicators having said that the second (despite sounding like misnomer) behaves like Ensign (if it does what it says on the can).

 

If you look closely at Jerrys videos you can see the volume averaged across each bar at its close.....the profile for the range of the bar increases.

[BlowFish 10]

  jperl said:

The question of how many HUPS to put on your trading data set is subjective. Too many and you can get confused; too few and you may miss important ones. If I'm trading just today, I will always have yesterdays data set HUPS on my chart( which will continue to change dynamically as todays data gets added to it). But I may also include additional VWAP data from 1 week, 1 month, 1 year ago if they happen to be nearby the price action.

Hope that helps.

 

Yes it does. Just interested in your own personal preference .

[BlowFish 10]

  BlowFish said:

Yes it does. Just interested in your own personal preference .

 

Actually can I recant that? . Let me describe a senario and possible way of approaching it and ask for your comments.

 

Lets say a weekly dataset has a positive skew and that price has moved up from the VWAP some way towards the SD1 (1/3rd or middway, whatever). We might anticipate that price would continue to SD1 the minimum expected movement.

 

Lets Imagine the daily profile (our chosen trading set) develops a downward skew. We might choose to skip a VWAP trade against the context of the larger dataset perhaps favouring a breakout (in accord with the larger dataset) or for waiting for the daily data set to flip.

 

Nothing really revolutionary just a way of considering the context of the bigger picture (data set). In the HUP section you described how they (HUPs) might hold up price but it seems to me they can provide more information too.

 

Of course you have to be careful not to information overload and to remain clear what you are actually trading. From previous posts I doubt that you would use the data like that but would still be interested in your comments.

[tradepro 10]

Hi Jerry,

 

Today I traded HD ( home depot) on a 1 minute chart and got 1 trade out of it. I am fully aware that the market can and will do whatever it wants. That being said HD was in an up trend all day but it was skewed to the negative.I took a text book "Jerry" short at 37.74 from the 1st SD to the VWAP at 12:36 to 12:53, covered, made a profit. It then went balistic because of the Beige book report. Before the report the slope of the VWAP was rising toward the PVP but still negatively skewed with price action way above the PVP and after the report it finally crossed over. My question to you is, if I see that the VWAP slope is converging towards the PVP, and price action is way above, could I enter a long trade at obvious HUP areas or would I be breaking any rules?

 

I am using ENSIGN and was wondering if I have the correct parameters on my price histogram:

 

range% 70

 

minutes: 1

 

restart: 930

 

boxsize: 1

 

initial: 0

 

Your help or anyone out there would be appreciated

 

Thanks,

 

 

Trade Pro

[jperl 11]

  BlowFish said:

Actually can I recant that? . Let me describe a senario and possible way of approaching it and ask for your comments.

 

Lets say a weekly dataset has a positive skew and that price has moved up from the VWAP some way towards the SD1 (1/3rd or middway, whatever). We might anticipate that price would continue to SD1 the minimum expected movement.

 

Lets Imagine the daily profile (our chosen trading set) develops a downward skew. We might choose to skip a VWAP trade against the context of the larger dataset perhaps favouring a breakout (in accord with the larger dataset) or for waiting for the daily data set to flip.

 

Nothing really revolutionary just a way of considering the context of the bigger picture (data set). In the HUP section you described how they (HUPs) might hold up price but it seems to me they can provide more information too.

 

Of course you have to be careful not to information overload and to remain clear what you are actually trading. From previous posts I doubt that you would use the data like that but would still be interested in your comments.

 

Your question comes down to the situation of what do you do when one data set shows a positive skew and another shows a negative skew; which way do you take the trade.?

 

This situation occurs all the time. The answer is not unique. It depends on what type of trade you are looking to enter. If you are trading against the weekly data with positive skew, it may take a whole week for you to see a positive long trade. If you are trading against the daily data with negative skew then you may want to go short keeping in mind that the longer time bias is long.

[BlowFish 10]

Hi Jerry, thanks for your reply. My question was not really meant to be a general 'what to do' question, more a request for a comment on the general principle. The general principle being using a larger data set as 'context' for a smaller dataset trade. Context could mean a directional bias or possibly a filter.

 

As you say there are many way's to interpret two data sets, the question is more along the lines of is there any advantage doing so? I think so, on the flip side one has added complexity to deal with.

 

As an example (again not asking for a comment on this specific case), if the weekly is moving from VWAP to SD1 (which will likely take some number of days) one might want to concentrate on daily trades in the same direction (these will be smaller, quicker trades) they could be VWAP, SD1, or even break outs as long as they are in the direction of the weekly movement (or context if you like). This particular example is similar to the traditional concept of going with the greater trend.

[jperl 11]

  BlowFish said:

Hi Jerry, thanks for your reply. My question was not really meant to be a general 'what to do' question, more a request for a comment on the general principle. The general principle being using a larger data set as 'context' for a smaller dataset trade. Context could mean a directional bias or possibly a filter.

 

As you say there are many way's to interpret two data sets, the question is more along the lines of is there any advantage doing so? I think so, on the flip side one has added complexity to deal with.

 

As an example (again not asking for a comment on this specific case), if the weekly is moving from VWAP to SD1 (which will likely take some number of days) one might want to concentrate on daily trades in the same direction (these will be smaller, quicker trades) they could be VWAP, SD1, or even break outs as long as they are in the direction of the weekly movement (or context if you like). This particular example is similar to the traditional concept of going with the greater trend.

 

Blowfish your concept of using the longer data set to decide on trade direction for the shorter data set would work fine. In fact it should work fine for just about any type of technical analysis one uses: eg. long time moveing average with short time moving average, long time regression analysis with short time, long time stochs with short time stochs. The key as you state is trading the short time in the direction of the long time trend. The only problem I have with it is you will miss about half the good trades.

[BlowFish 10]

I could see no reason why it would not be valid. As you say hardly a new concept. I guess the question is (and it is as much a rhetorical question as one directed at you ) is whether it causes one to miss more bad trades than good trades. It might be useful to give confidence in some of the riskier trades (break out springs to mind) if it is in the direction of the larger data set.

 

Another thought occurs to me (again nothing new) and that is using a trade setup in a smaller data set as a 'trigger' for a larger data set trade. Similar sort of deal really the main difference is the focus of the trade is on the larger data set.