Very Strong Coordinated Technical Commercial Selling in ES, YM & NQ

[UrmaBlume 10]

Todays trade saw unusually strong, perfectly coordinated selling by commercials in all major US equity markets.

 

At EXACTLY 0828 PST, 1128 EST there was very strong, computer based technical selling in the S&P Futures, The Dow Future and the NASDAQ 100 future. I can verify that this selling was coordinated down to the millisecond.

 

Here are the 3 screenshots that tell the story:

 

 

 

Edited September 17, 2009 by UrmaBlume

[UrmaBlume 10]

The same selling noted in the previous post happened again at 1224 PST, and again in all 3 markets:

 

In addition this indicator game up 3 great trades today (09/17) which can be seen here

 

 

 

Edited September 17, 2009 by UrmaBlume

[BlowFish 10]

 

I can verify that this selling was coordinated down to the millisecond.

 

 

I think it was your very post here that I quizzed you about your infrastructure. I thought you where using Tradestation data? Or are you getting a data feed that is time stamped at the exchange? I am going to go out on a limb here and call hyperbole. That's not to say that this is not a pretty interesting post but as I've said a million times (hehe) milisecond accuracy isn't plausible. Transactional latency on the CME systems is likely to be up to an order of magnitude greater than milliseconds. I hope I am not sounding pedantic but it might shape other peoples expectations.

[UrmaBlume 10]

I think it was your very post here that I quizzed you about your infrastructure. I thought you where using Tradestation data? Or are you getting a data feed that is time stamped at the exchange? I am going to go out on a limb here and call hyperbole. That's not to say that this is not a pretty interesting post but as I've said a million times (hehe) milisecond accuracy isn't plausible. Transactional latency on the CME systems is likely to be up to an order of magnitude greater than milliseconds. I hope I am not sounding pedantic but it might shape other peoples expectations.

 

Easy BlowFish. The point of interest here is that precisely coordinated selling at an abnormal intensity in 3 related markets got knocked off.

 

Of course you are right about millisecond accuracy. Everybody with experience at that level of operation knows that. The point is that the code is written to millisecond or even microsecond time stamps so that is the time frame most practioners use as a reference. Also, there are, indeed, ways to handle Trade Station data FROM inside Trade Station with millisecond granularity. Not as good as thousand dollar a day data sources that put you cheek to cheek with the exchange but certainly close enough for practical application.

 

For everybody but the Goldman's of the world and from a practical standpoint the issue has always been one of method rather than of latency.

 

Whether I work with milliseconds, microseconds or half-hours is not nearly so to the point as the fact that in the sub 1 minute time frame this indicator spotted this very significant anomaly in commercial trade in 3 different markets.

 

Certainly the signals were timely engouh to allow entry at very advantageous prices and after all shouldn't accurate, timely indications of profitable Trade Points be the most valuable product of any Trader's Laboratory?

 

Call the millisecond, micorsecond, half-second point poetic and or technical license and liberty and take a broader look at a very rare view of a rare occurrence.

Edited September 19, 2009 by UrmaBlume

[BlowFish 10]

Absolutely. Interesting post. I was in two minds whether to say anything but I figured it might have some value to people that are following your work (and that is quite a few if the PM's I have had about code are anything to go by). Certainly it wasn't just to be argumentative and I apologise if it seemed overly critical. Some people perhaps don't have a good idea of 'scale' and perhaps it might be helpful to them (particularly if trying to come up with their own indicators, that level of precision is not necessary and probably unattainable in windows on PC hardware).

 

For example when an app requests data from disk it can take 10 or even 20 milliseconds before the first byte is delivered. DSL local loop delay can be 30 ms. Light will get about 300km in a vaccum in a bit over a millisecond.

[ZOSO 10]

So what?

 

You have a great indicator and the rest of the world doesn't. Good for you and bad for us. Why do you get such pleasure rubbing it in our faces? We recognize your superiority and mastery over the markets. Poor retail schlebs like us aren't as blessed or gifted.

 

We all understand your theory and how the indicator supports it, but give it a rest already. Unless you open up the kimono (which you said you won't), there is nothing more to be learned by plastering the forum with these charts day after day.

 

What your doing here is exactly the same thing you did at ET and you got banned for it.

[Frank 10]

have a question,

 

in most of these screen shots you post there is a simple pattern -- a ton of block orders after a swing in one direction followed by a moving average crossover in the other direction, which seems to confirm it was buying or selling. (I am not talking trade set-up here, just recognizing commercial activity). if you see a situation that doesn't qualify for this, large block orders with a MA crossover against last swing -- can you discuss perhaps why that would not be a good way to track commercial activity? if there is no edge past the opening 2 minutes of this occurence, then I would say its just a short-term scalp and so there is nothing in these screenshots that will ever be useful to anyone here. If that is the answer, then so be it -- just curious.

Edited September 20, 2009 by Frank

[UrmaBlume 10]

Absolutely. Interesting post. I was in two minds whether to say anything but I figured it might have some value to people that are following your work (and that is quite a few if the PM's I have had about code are anything to go by). Certainly it wasn't just to be argumentative and I apologise if it seemed overly critical. Some people perhaps don't have a good idea of 'scale' and perhaps it might be helpful to them (particularly if trying to come up with their own indicators, that level of precision is not necessary and probably unattainable in windows on PC hardware).For example when an app requests data from disk it can take 10 or even 20 milliseconds before the first byte is delivered. DSL local loop delay can be 30 ms. Light will get about 300km in a vaccum in a bit over a millisecond.

 

BlowFish,

 

Thanks for the kind words.

 

I too have had many PMs & emails about this and other or our indicators and I am very appreciative that the vast marjority of them have been so complimentary and suppotive.

 

I especially applaud those with the intellectual capital, market understanding and technical ability that report a significant increase in the efficacy of their trading as a result of their efforts to build their own versions of our work.

 

Long-term successful trading is not about instinct, divine inspiration or spontaneous intellectual combustion. It is about intelligent data processing and sound method and those willing to think for themselves and do their own work will usually lead the pack in any endeavor.

 

Thanks again for the kind words and your many contributions to this community.

 

cheers

[UrmaBlume 10]

have a question, in most of these screen shots you post there is a simple pattern -- a ton of block orders after a swing in one direction followed by a moving average crossover in the other direction, which seems to confirm it was buying or selling. (I am not talking trade set-up here, just recognizing commercial activity). if you see a situation that doesn't qualify for this, large block orders with a MA crossover against last swing -- can you discuss perhaps why that would not be a good way to track commercial activity? if there is no edge past the opening 2 minutes of this occurence, then I would say its just a short-term scalp and so there is nothing in these screenshots that will ever be useful to anyone here. If that is the answer, then so be it -- just curious.

 

Frank,

 

Thank you for your query.

 

You are indeed correct when you note that big blocks mark commercial activity, 1-10 lot traders NEVER trade 100's or bigger. The fact of the big blocks is sometimes a warning that it is NOT the KIND of commercail activity that forms local extremes. It all depends on how you measure the metrics of the order flow.

 

Block trades and MA crossovers have nothing to do with the calculation of this indicator or how it should be used. On reason the method you discuss might lack efficacy is that those block trades could be premium arbitrage and the fact of a block trade says little as to whether it is buying or selling.

 

The calculation of this indicator involves much more that just volume over time. "When taken in combination, the acceleration and deceleration of buying and selling volumes, total volume and the velocity/rate of change in the balance of trade reveal a certain dynamic that we find present at many, if not most, intra-session extremes."

 

For those with greater than retail market understanding that are also technically astute the value in these screenshots and the accompaying text has been, or so they report, two fold 1) the screenshots document that the indication occurs at almost all session extremes and many intra-session local extremes. 2) the value of the accompanying text to those same informed, technically competent traders is that they have been able to gleen from that text enough informaiton to build their own. The value of the ability to locate session highs and lows as or slightly BEFORE they occur speaks for itself.

 

Speaking generally - of course those are the traders with above averge intellect and ability. The lowest intellectual common denomiator has never done well in this game and most likely never will. It has always been a very small percentage of the players that win at this game and what separates them from the masses is not their size but is their understanding and ability.

 

Some of what is so wonderful about this game we both play is that to win you don't have to have a license, you don't have to have anyone's approval, you don't have to be a member of anything and you don't need tremendous capital - all you have to be is right and if you are right you wont be little for long. There is and always has been unlimited backing for those that can demonstrate either great efficacy in their personal trading or the understanding and ability to construct profitable automated systems.

 

cheers

[Frank 10]

ok ---

 

I point this out for others and for my own reminder.

 

there are all these threads like 'trade intensity' and this one is ' Very Strong Coordinated Technical Commercial Selling in ES, YM & NQ' --- but the indicator in the charts is actually an index of indicators measuring a number of things on different timeframes. so the size of the bar in the indicator pane --- the one in the histogram --- is building up when the INDEX adds up. There is likely a perception here that the histogram is a summation of some kind of filtered volume for large block orders --- because of the title of these threads makes it sound like this is a volume indicator -- like figuring out if that last 2k contract trade was a buy or a sell. but the histogram in the indicator pane is actually an index that is stacking up biases in combination and just because there are large block orders passing doesn't mean the index will build.

 

for example, (this is structure of the indicator only, done intentionally in oversimplified 'if/then' branching programming statements for structure only)

 

1) if trade is strong given this time of day then +1

2) if the velocity of trade on X timeframe has DECREASED then +1

3) if the velocity of trade on Y timeframe has DECREASED then +1

 

so the index would be +3 in this example.

 

if block orders ARE passing without the above conditions, then it is possible that the index may be 0 or whatever, big block orders without a signal

 

I was not implying that the MA crossover was a part of the indicator -- I was just suggesting that action just after a signal would confirm the signal and this is quite easily seen in the MA crossover.

 

 

edited

Edited September 20, 2009 by Frank

[UrmaBlume 10]

ok --- I point this out for others and for my own reminder.there are all these threads like 'trade intensity' and this one is ' Very Strong Coordinated Technical Commercial Selling in ES, YM & NQ' --- but the indicator in the charts is actually an index of indicators measuring a number of things on different timeframes. so the size of the bar in the indicator pane --- the one in the histogram --- is building up when the INDEX adds up (right?) There is likely a perception here that the histogram is a summation of some kind of filtered volume for large block orders --- because of the title of these threads makes it sound like this is a volume indicator -- like figuring out if that last 2k contract trade was a buy or a sell. but the histogram in the indicator pane is actually an index that is stacking up biases in combination and just because there are large block orders passing doesn't mean the index will build.for example, (this is structure of the indicator only, done intentionally in oversimplified 'if/then' branching programming statements for structure only)1) if trade is strong given this time of day then +12) if the velocity of trade on X timeframe has DECREASED then +1) if the velocity of trade on Y timeframe has DECREASED then +1 so the index would be +3 in this example.if block orders ARE passing without the above conditions, then it is possible that the index may be 0 or whatever...I was not implying that the MA crossover was a part of the indicator -- I was just suggesting that action just after a signal would confirm the signal and this is quite easily seen in the MA crossover.edited

 

 

You analysis of this indicator mixes in waaay too much from the thread I started as a primer on the construction of an index of weighted biases. That is not the way this indicator is calculated plus almost all of the inputs for this indicator are preprocessed from a single time/volume frame and that is waaaaay sub 1 minute.

 

Again to get closer to the essence of this dynamic I recommend that you reduce your focus on blocks - they are not to the point here.

 

While slightly off-topic yet still related and another clue for the AI folks here - Does anyone else here use MARS (Multivariate Adaptive Regression Splines) from Salford systems as part of their final layer of processing?

[Frank 10]

almost all of the inputs for this indicator are preprocessed from a single time/volume frame and that is waaaaay sub 1 minute

 

ok, thanks - that is what I was wondering. but one thing still remains unclear. the indicator exists on a sub 1 minute level --- but it is unclear on the timeframe for how long an extreme reading remains relevant. I ask because this is central --- while the INDICATOR exists on this timeframe --- is it only the high-speed execution of this signal what accounts for profits or losses? if you are saying that it often marks the time of the 'day session extreme' -- then certainly this is a signal that is good on a higher level --- something that does not require an application written in machine language and a server within 10 feet of a major internet backbone for sub-second execution.

[UrmaBlume 10]

ok, thanks - that is what I was wondering. but one thing still remains unclear. the indicator exists on a sub 1 minute level --- but it is unclear on the timeframe for how long an extreme reading remains relevant. I ask because this is central --- while the INDICATOR exists on this timeframe --- is it only the high-speed execution of this signal what accounts for profits or losses? if you are saying that it often marks the time of the 'day session extreme' -- then certainly this is a signal that is good on a higher level --- something that does not require an application written in machine language and a server within 10 feet of a major internet backbone for sub-second execution.

 

Very astute. Yes indeed the indicator is calculated with sub-second granularity but no we do not depend on sub-second execution except in our very high frequency tick systems.

 

You are also right about not needing to be cheek to cheek w/exchange data to operate. From a practical perspective for all but the Goldman's of the world the difference between a thousand dollar a day connection and feeds such as ZenFire are not meaningful.

 

We have done this work in both TradeStation and Ninja and we find no practical advantage to either. While we know that ZenFire is faster we have ways to add much more granularity to Trade Station data than their 1 minute time stamp allows. The net is that for our purposes we find no practical difference so we stay with Trade Station because we are very familiar with the platfrom and that over the years we have written several thousand indicators, paint bars, show me, activity bars, functions, strategies and dlls based on Easy Language plus we find Ninja to be very young in its presentation and cumbersome to use.

 

Day after day I post here and in certain reports the occurence of this indication on session extremes for a variety of markets and yes indeed we treat and trade this as a "longer" term intra-session indicator and execute, sometimes more than a few times in the same direction from a single such indication. How many of these entries and how long we stay i them is taken from indications of order flow and the rate of change in the balance of trade on volume charts that range in time frame from about 20 second bars to 3 min bars.

 

BTW I used to live in your neck of the woods. Lived in HIllsborough, just to the South, and had offices on New Montgomery next to the Sheraton Palace. Did many a drunken biz lunch at the long gone House of Shields accross the street, great food and evern better bar.

 

cheers

[Frank 10]

thanks

 

yes, I remember the House of Shields, I worked within 2 blocks of there from 1993 - 2003 and went to many conferences at the Sheraton Palace. Ate at Henry Hunans many times too -- best chinese food ever -- New Montgomer and Natoma St.

 

So you are in Vegas, you play any World Series events this year?

[UrmaBlume 10]

thanksyes, I remember the House of Shields, I worked within 2 blocks of there from 1993 - 2003 and went to many conferences at the Sheraton Palace. Ate at Henry Hunans many times too -- best chinese food ever -- New Montgomer and Natoma St.So you are in Vegas, you play any World Series events this year?

 

No WSOP for me for the last several years. My last big events were a couple of the Party Poker Million cruises that I won online. They were really great parties. The whole ship chartered for a week's trip to Mexico and nothing but poker players onboard - tequila, lobster, poker and a shot at a million what a week and I while I didn't cash in either event I won both $14k entries by moving up from a $25 satellite.

 

Party was great back then. They had just moved to Gibraltar and flew me to Gib to do some consulting for them.

 

As to the House of Shields, San Francisco has always been a great "eating" town. If you play poker maybe you should visit the Garden City poker room in San Jose. Not the best poker in the bay area but certainly the best food of any poker room in the bay area.

 

While you can certainly eat well here in Vegas it is almost always in one of the "name" joints where dinner costs a grand and that's only if you don't drink a lot. I miss the bay area for all its small, out of the way, hole in the wall resturants that are soooo good across so many cultures. Two exceptions to that here in Vegas are the King and I on Trop for Thai and Mamacitas on Fremont for Cuban/Mexican.

 

cheers

 

PS - While we are waaaay off topic here is a shot of the Miraflores Locks in the Panama Canal that updates everytime it is loaded or refreshed, in my days as a sailor I made a dozen passages of the canal and all of them were fascinating.

 

Edited September 20, 2009 by UrmaBlume

[MightyMouse 12]

Todays trade saw unusually strong, perfectly coordinated selling by commercials in all major US equity markets.

 

At EXACTLY 0828 PST, 1128 EST there was very strong, computer based technical selling in the S&P Futures, The Dow Future and the NASDAQ 100 future. I can verify that this selling was coordinated down to the millisecond.

 

Here are the 3 screenshots that tell the story:

 

 

 

 

Clearly, you can imply from the screenshot that the concerted effort to "push" down price in all US markets worked and, erroneously, you can conclude that each time they attempt to push prices they will succeed. I think the more interesting screenshots would be the other 7 out of 10 screenshots that show them failing or actually getting their arse handed to them because price didn’t react as they expected and now need to liquidate.

 

It looks like another good way to overtrade your account.