Login Page - Create Account

Support Board


Date/Time: Sun, 26 May 2019 16:31:08 +0000



[User Discussion] - Trade Intensity/Pace of Tape indicator for SierraChart?

Support Request:
[2013-04-05 16:49:11]
pe1111112 - Posts: 35
There have been some long threads about this indicator on other forums so I will try to keep this short.

Basically this is an indicator that measures the speed at which orders are filled. Here's a description from the original thread about this indicator (link at bottom).

In most markets and certainly in the equity index futures markets there is a small set of very big traders whose entries into the market often have an immediate and very tradeable impact on price.

These traders go to great lengths to disquise their trade - they have automated routines to send a series of small transactions to the market at intervals close to 1000th of a second.

Because many of these traders operate from price based models they must get their execution within a very narrow price/time range which means they must execute as much of the volume they want as fast as they can while price is in that range. This results in a huge spike in what we call the intensity of trade.

See the bottom study on this image:

[img]http://www.tradepointtechnologies.com/postpics/tradeintensity2.jpg[/img]

Note the time at the bottom. That entire chart is for about one minute of data.

These spikes often signal reversals:

[img]http://www.tradepointtechnologies.com/postpics/tradeintensity1.jpg[/img]


Something similar called "Pace of Tape" is available for NinjaTrader. The code can be found on this page and is pretty short:
http://www.tradewiththeflow.com/2010/05/17/pace-of-tape-indicator/


Here is the a thread about it:
http://www.traderslaboratory.com/forums/technical-analysis/5277-trade-intensity.html

A thread about the NT version called "Pace of Tape":
http://www.trade2win.com/boards/technical-analysis/94424-pace-tape-measure-trade-intensity-detect-reversals.html

Is this indicator possible to code for SierraChart? Is the NT code posted on that site helpful?

edit - how does one post images on this forum? I tried [img] tags but it didn't work.
[/img]
Date Time Of Last Edit: 2013-04-05 16:49:41
[2013-04-05 18:10:11]
tomgilb - Posts: 2162
Take a look at this:
http://www.bigmiketrading.com/sierra-chart-programming/22945-speed-tape.html
[2013-04-05 18:26:31]
pe1111112 - Posts: 35
Hmm, interesting. I was trying to do it with the "bar time duration" and tick charts. Unfortunately, the smallest time value SC seems to be able to process is 0.0000115740695036948 which it represents as 1.15741E-05. That is not decimals of a second, but is just part of how SC displays time in the worksheets.

The goal was to use a fast tick chart and display the time per bar, as bars appearing faster = more trades = lower time per bar, and then modify that data in a worksheet to display when the time per tick bar was under a certain level. However, the lowest value possible is the number I mentioned above, and any bars that take less time than that show a time of 0. Because of this I was unable to get the necessary sensitivity (ie. all the values that exist between 0.0000115740695036948 and 0), and in order to not get a ton of false signals, a higher tick chart value must be used, and the higher the tick chart value used, the less accurate the signal (eg. a 30 tick chart will be better than a 200 tick chart, but there are already a lot of false signals on the 200 tick chart).

I'd post example charts but I can't figure out how to post images here.

Question for SierraChart staff: what unit of time does a value of 1.15741E-05 represent in the spreadsheets? Is that one second?
[2013-04-05 19:50:51]
pe1111112 - Posts: 35
Here's the best I could come up with. The speed seems to be limited to the value I mentioned above which means below a certain bar threshold you'll get more signals than you want since SierraChart isn't able to calculate units of time less than what I believe is one second (represented by 0.0000115740695036948 in the spreadsheets).

You can see the attached 250 tick chart starts to show too many signals, and any faster chart (meaning less than 250 ticks) is going to show even more unless we can use a lower value than 1 second.
imagepace of tape 500 tick es.gif / V - Attached On 2013-04-05 19:49:51 UTC - Size: 31.21 KB - 1810 views
imagepace of tape 2000 volume es.gif / V - Attached On 2013-04-05 19:49:56 UTC - Size: 30.68 KB - 1475 views
imagepace of tape 250 tick es.gif / V - Attached On 2013-04-05 19:50:00 UTC - Size: 29.55 KB - 1227 views
[2013-04-05 21:11:32]
tomgilb - Posts: 2162
In Excel time format, one second equals 1/86400 because there are 86400 seconds in 24 hours.

1/86400 = 1.157407E-05
[2013-04-07 03:56:09]
pe1111112 - Posts: 35
Ok that makes sense then.

In that case, the indicator I posted is as accurate as it can get. I believe the original one was able to look at smaller than one second at a time.
[2013-04-07 04:22:51]
ganz - Posts: 944
hi

this is "Trades per Second" study.

also we are waiting for the subsecond DB by SC Team in the nearest future

gd lck

[img=s23.postimg.org/v90nexca3/7073975.png][/img]
Date Time Of Last Edit: 2013-04-07 04:24:34
attachmentsot.dll - Attached On 2013-04-07 04:22:43 UTC - Size: 66 KB - 598 views
image7073975.png / V - Attached On 2013-04-07 04:24:20 UTC - Size: 98.11 KB - 1447 views
[2013-04-07 04:57:22]
M5amhan - Posts: 468
When I used to track the number of trades going through I would use a 10 second chart with "number of trades" study and time and sales with different filters for different markets .. it sounds like you are trying to get more complicated than this though.

Just throwing that out there
[2013-04-07 05:01:47]
M5amhan - Posts: 468
I would also use "color bar based on alert condition" to differenciate when trading was slow, fast, or very fast
Date Time Of Last Edit: 2013-04-07 05:01:59
[2013-04-07 06:20:09]
Sierra Chart Engineering - Posts: 72155
0.0000115740695036948 is equivalent to 1 second.
Sierra Chart Support - Engineering Level

Your definitive source for support. Other responses are from users. If possible please keep your questions brief and to the point. Please be aware of support policy:
https://www.sierrachart.com/index.php?l=PostingInformation.php#GeneralInformation

If your question/request has been answered and you do not have anything further, then it is easiest for us if you do not reply again to say thank you.
[2013-04-07 16:19:15]
pe1111112 - Posts: 35
hi

this is "Trades per Second" study.

also we are waiting for the subsecond DB by SC Team in the nearest future

gd lck

s23.postimg.org/v90nexca3/7073975.png

Where/how do I install that .dll file? I'm new to Sierra Chart.

Thanks.
[2013-04-07 16:26:36]
tomgilb - Posts: 2162
Put it in your Sierra Chart \Data folder, then click on Analysis >> Studies >> Add Custom Study...
[2013-04-07 19:26:59]
pe1111112 - Posts: 35
Cool, thanks. What exactly is that "speed of tape" study measuring? I noticed the histogram is almost exactly the same as the volume histogram on a 5 min chart.

edit - it seems if the time on a bar is 0 seconds, the speed of tape indicator returns a value of zero, too. This value should be a spike instead since it means the orders were coming so fast that the bar took less than 1 second to form.
Date Time Of Last Edit: 2013-04-07 19:46:59
image5 min es march 12 2013.gif / V - Attached On 2013-04-07 19:26:55 UTC - Size: 20.7 KB - 1056 views
[2013-04-07 19:59:27]
tomgilb - Posts: 2162
My suggestion at the link in post#2 was only an attempt to emulate the pic of the original poster. It only measures the # of trades per bar, then applies an MA to it to create a threshold for an alert/color. Admittedly, it's not very sophisticated.

There is a close (but not exact) correlation between volume and # of trades, so histograms of each will look similar.
[2013-04-08 00:49:33]
ganz - Posts: 944
hi

tomgilb, Tom, thanks a lot for the assistance

pe1111112

What exactly is that "speed of tape" study measuring?

SoT = NumbersOfTrades / Bar Time Duration

it seems if the time on a bar is 0 seconds, the speed of tape indicator returns a value of zero, too. This value should be a spike instead since it means the orders were coming so fast that the bar took less than 1 second to form.

it uses DB w time stamps as low as 1 second

i'm sorry but i don't understand how you are able to create a bar w 0 seconds resolution - using SC it will be always w 1 second or more by design at this moment, imho

ps: try to use non-time based bar charts w it

gd lck
Date Time Of Last Edit: 2013-04-08 01:09:03
[2013-04-08 01:27:59]
pe1111112 - Posts: 35
it uses DB w time stamps as low as 1 second

i'm sorry but i don't understand how you are able to create a bar w 0 seconds resolution - using SC it will be always w 1 second or more by design at this moment, imho

When two bars print so quickly that the time between them registers as 0 seconds (ie. they have the same timestamp down to the second).
[2013-04-08 01:54:03]
ganz - Posts: 944
pe1111112,

there was the discussion about it http://www.sierrachart.com/supportboard/showthread.php?t=39831

there are no chances to divide some ticks w the same timestamp right now
[2013-04-10 17:22:53]
pe1111112 - Posts: 35
I've attached an image showing the problem with the Pace of Tape indicator.

Pace of Tape shows taller bars for faster candles.

This works correctly unless the time of a candle is 0 in which case Pace of Tape shows no bar when it should actually be showing the tallest bar possible.
imagepace of tape vs bar time duration.gif / V - Attached On 2013-04-10 17:22:26 UTC - Size: 40.36 KB - 1466 views
[2013-04-11 05:01:30]
ganz - Posts: 944
hi

Bar Time Duration = Last Time of a Current Bar - Last Time of a Previous Bar

I've got no an idea to segregate bars/ticks w the same /1sec/ time stamp as was mentioned above

The time differential of these bars will be always zero and SoT will paint it as zero.

So the study has just a kind of indication of the situation.

gd lck
[2013-04-12 07:24:26]
crazybears - Posts: 314
Hi Pe1111112 :)

would you try to measure the velocity at which a bar is formed regerdlless the number of trades of each bar
or also you want calculate the # of trades ?

I think first post here : http://www.traderslaboratory.com/forums/technical-analysis/5277-trade-intensity.html measure something like the # of trade or bid and ask volume and a moving average.
[2013-04-12 07:46:05]
Marmany - Posts: 207
Spent time on another platform looking into high speed trading. Based in UK with highest speed broadband available it was taking 200 msec to receive CME data from Chicago. If you plan to follow these trades and act on them then it is necessary to have your software hosted on a cpu in the exchange.
My solution was to place stop market entries directly on exchange (via TransAct).
This either gets you ahead of the HS traders or you simply pull the order if they fail to strike.
[2013-04-13 14:22:14]
T44 - Posts: 179

In most markets and certainly in the equity index futures markets there is a small set of very big traders whose entries into the market often have an immediate and very tradeable impact on price.

These traders go to great lengths to disquise their trade - they have automated routines to send a series of small transactions to the market at intervals close to 1000th of a second.

Because many of these traders operate from price based models they must get their execution within a very narrow price/time range which means they must execute as much of the volume they want as fast as they can while price is in that range. This results in a huge spike in what we call the intensity of trade.

These spikes often signal reversals.

A lot of assumptions being made here.

-that these trades represent "entries" (new positions not closing out existing positions)
-that these trades are directional (not a hedge, an arb against an ETF or basket, a basis trade etc)
-that you can quantify a "big trade" accurately - a "big trade" is one which creates enough market impact to make it "tradeable" (the amount of size which will halt a market differs according to the market, time of day, condition in force, and other factors which you do not mention and probably cannot calculate)
-that these "big traders" transacting at a point of high liquidity to complete some of their business in a certain price/time spread are going to continue to influence price in the direction they have traded, enough for you to profit from

The biggest and most obvious examples you are missing is the time slicing of orders and partial covering of large positions. Both of these may trade aggressively at a point of high liquidity where X contracts can be filled quickly and in a tight spread, but neither indicates the market will move up (down) after these buy (sell) orders are filled.

If you're off down the wrong path, spending the time to create an indicator (graphical representation of a condition) is a waste of time if you have not proved to your satisfaction that a certain outcome is more likely to occur after this condition is shown. Prove that the assumptions underlying the theory are valid before deploying resources to make a computer implementation, be it a Sierra study or an execution algorithm to be colocated at the exchange.
[2013-04-14 05:41:12]
joshtrader - Posts: 318
The assumptions given above cannot be proven or disproven for a general case, so that alone would be a waste of time.

If the original poster can get a graphical representation of what he wants to see, this itself can be used to prove whether a certain outcome is more likely to occur, which is what you recommend. In other words, he can't very well do research easily on this until he has a tool which can give him the data, which is what this can be used for.

I agree that we cannot assume the above, and perhaps the original poster has, which would be a mistake, but whether the assumptions hold or not is irrelevant; what is relevant is to find something that can yield an edge.
[2013-04-14 06:06:52]
crazybears - Posts: 314
Hi

What does Pace of Tape measure?
[2013-04-14 08:35:32]
T44 - Posts: 179
Joshtrader, surely if the assumptions cannot be or are not proven then it is dangerous to continue a theory through the point of allocating capital without testing these assumptions, and if the assumptions are not testable....

Perhaps having a graphical representation would be of assistance not to prove the theory but to quickly falsify it so the aspiring trader can move on to investigating areas which may be more productive. I've looked quickly at the bigmike, traderslab, and trade2win threads about this indicator and the screenshots provided there should be enough to "falsify" the hypothesis that this technique in isolation is of any assistance whatsoever in picking reversals.

The great thing about programs such as Sierra and the support provided on this board is that it allows non-technical/non-programmer people to express themselves in ways which would not otherwise be possible. The potential downside of this is that there are so many theories and ideas put forward that a lifetime could be wasted testing each in turn to see if it contains the holy grail, at the cost of neglecting more productive approaches. It is for the individual to discern what is relevant and those who can do this will be richly rewarded.

The marketplace is competitive and therefore there are significant barriers both to discovering a successful strategy and to implementing it in size. If this study alone did provide an edge, too many people would discover or copy the strategy and hammer out the edge. Trying to use the same pace of trade / tick volume indicator as a mass of other people is almost guaranteed to remove any edge there may have been. So if anyone here does design the "magic indicator" it would be wise to trade it up to the liquidity limits of the instrument and timeline being used (for as long as the opportunity lasts) and not to share it on the Internet.

Of course human nature being as it is, people will continue to try to systemize in ways which are unlikely to work, and further argue that they want this indicator not because they can prove it to work but because it somehow 'supports' or 'gives confluence to' their other (untested, unproven) entry criteria. All one can do is suggest some questions and try to point in an alternative direction, but the human condition being as it is few will find a path to consistent success in trading. This is how it has to be.

To bring this thread back to a slightly more relevant point: surely how the exchange matches and reports trade volumes is going to be a significant input in deciding which instruments this study will be of any use on. For example, some years ago the CME changed how their time and sales feed was composed. Previously, the size of the aggressor order was reported, and currently the price and size of all trades are reported. So in the prior case, by intent of the aggressor order, a market order sell 50 in ES would have reported a single entry of 50 lots in the time and sales, whereas now what will be reported is the contracts at best bid who filled the 50, which can be either a single bid of 50, 50 bids of 1, or anything in between. This is an issue affecting anyone who previously used tick based charting in that instrument.

So this proposed study will show different results depending on how the exchange reports time and sales data. Is the OP and others who use the strategy aware of these concepts and how it would affect their results? There is no easy money in this business, and if there is you're getting paid short odds.
[2013-04-15 05:47:24]
pe1111112 - Posts: 35
The general idea is that the faster orders are being filled, the more likely you are to be at a key point.

If you watch the orders throughout the day sometimes there is a few seconds between fills. No one really cares what is going on there. And other times many orders get filled nearly at the same time because something important is happening.

If you do not have a way to measure how fast trades are taking place, instead what you have to do is use a constant volume bar (or tick bar) and measure how quickly each bar forms. This is a sort of alternate way to measure kind of the same thing.

If, for example, two constant volume chart candles appear with the same timestamp, it's because many orders were just filled in a short period of time (less than one second). Compare that against two constant volume candles with 2 minutes difference in their timestamp in which case orders were filled much more slowly.

The Bar Time Duration study lets you measure this. Due to SierraChart's time limitations, however, you can only be accurate to one second (or zero seconds). So for your signal you have to look for candles where the Bar Time Duration shows a time of zero, or one second, and that must be matched with an appropriate volume or tick candle. For example, if you use 5,000 constant volume candles on the ES, there will almost never be any signals because it will always take more than 1 second for the bars to complete. So 5,000 constant volume candles are too slow. If you use 50 constant volume candles, almost every bar will have a signal, because it won't take more than one second for most of the candles to form, so this chart is too fast.

All I did in the screenshot I posted was use a worksheet to put a point on the candle when the Bar Time Duration was less than one second (which would be zero seconds), and then I played around with the chart settings until I found one that wasn't too fast (signal on every bar) nor too slow.
[2013-04-15 13:08:54]
joshtrader - Posts: 318
and the screenshots provided there should be enough to "falsify" the hypothesis that this technique in isolation is of any assistance whatsoever in picking reversals.

If this study alone did provide an edge

Maybe some will want to use this in isolation, but nothing in isolation that is systematic in nature will work, because it will quickly be taken advantage of, as you say. But I certainly didn't propose using it in isolation.

they want this indicator not because they can prove it to work but because it somehow 'supports' or 'gives confluence to' their other (untested, unproven) entry criteria.

Yes, absolutely. "Entry criteria" is very interesting because some who are very systematic in nature (who don't really trade, but who do math instead) need fixed entry criteria. Others can look at a DOM and trade effectively this way, and use context and other "unprovable" criteria (in the sense that it is not mathematically quantifiable, it cannot be programmed and cannot be tested or proved in the general sense). The indicator proposed here may provide a visual help for some who trade this way, as one possibility.

surely how the exchange matches and reports trade volumes is going to be a significant input in deciding which instruments this study will be of any use on.

Of course, which is why a tick chart is far inferior in this case to a volume chart, which does not take into account orders but volume only, so that changes such as the one you mentioned, which affected quite a few people, would not affect someone who does not care about what a "trade" is but rather about traded volume. I do not use this "pace of tape" indicator, and in fact all one needs to really see this is to put up a small time-based chart with a volume histogram, or program (as I did) a simple indicator to be used on a volume chart that shows the same thing.

Is the OP and others who use the strategy aware of these concepts and how it would affect their results? There is no easy money in this business, and if there is you're getting paid short odds.

I would hope the OP is aware -- most traders know about the CME changes you referenced, and it's not rocket science.

Absolutely there is no easy money, in the sense that profitable "dumb" strategies which do not rely on unique human expression of thought and which are quantifiable in nature are few and far between, and don't work for long without being modified. In contrast, using logic, intuition, and simple tools can often be quite rewarding.
Date Time Of Last Edit: 2013-04-15 13:09:45
[2019-03-27 19:52:37]
whats1thingnow - Posts: 233
in fact all one needs to really see this is to put up a small time-based chart with a volume histogram, or program (as I did) a simple indicator to be used on a volume chart that shows the same thing


regardless of tick or volume chart

add volume study
add bar duration study
add study subgraphs divide/ratio (input study 1: volume & input study 2: duration)

this will effectively give you volume/sec (better than trades/sec imo)

To post a message in this thread, you need to login with your Sierra Chart account:

Login


Login Page - Create Account