CME tag 37 changes, anyone have thoughts on this?

[Zzzz1]

Very good point. I know some guys who model order positioning in the queue based on order IDs and exchange member IDs (only applicable in certain markets and with certain feeds).

There is a document on the exchange website that details which changes to an order will retain and which will lose queue priority though I strongly assume any market practitioner should know about that.

Think "queue position".

 

[Lee-]

Very good point. I know some guys who model order positioning in the queue based on order IDs and exchange member IDs (only applicable in certain markets and with certain feeds).

There is a document on the exchange website that details which changes to an order will retain and which will lose queue priority though I strongly assume any market practitioner should know about that.

Certainly you can estimate queue position based on price levels, arrival times, changes, what effects queue positions, etc, but it's merely an estimate. What I mean is this:

Say 4 orders come in at the same price (orders A, B, C, D) and that none of these orders are identified by unique order ids. Order C is your order. B and D are the same size, but A and C are not the same size as B and D. Now you see an order cancellation that is the size of the B and D order. How can you be sure which order was affected? You can statistically model the likelihood that it was B or D, but you don't know, so at this point it becomes less certain than if you are given the corresponding order IDs.

Knowing exactly which orders ids provides additional transparency, however not all players are going to benefit equally from this additional transparency.

It also changes the types of games algos can play against each other.

 

[garachen]

Interesting. I look at this field quite a bit. I have anonymous=false for my own orders currently.

From my last observation it appeared that about 50% of users were anonymous and 50% not.

It's a two way street though. IF there were any way to figure out which orders belonged to whom it would also be possible for HFT players to identify each other which most HFT would consider not worth the risk.

I'll take a look at the field today (since that's what I'm working on currently anyways) and will respond privately if I find anything of concern as a thanks for giving me the heads up.

Something about this sets off the paranoia and I'm curious if I'm reading it right:

http://www.cmegroup.com/notices/electronic-trading/2017/03/20170313.html#tso
http://www.cmegroup.com/confluence/...ummary#MDP3.0-TradeSummary-TradeSummaryOpt-In

View attachment 171926

View attachment 171927

View attachment 171928

Whenever I see something like "increase transparency" I usually translate it to "benefit HFT" - so what's the angle here, if any? Or am I just reading into something inconsequential?

CC: @garachen

 

[garachen]

There's no identifying information in the OrderID field. It's useless for gaining insights into anyone else's trading. One could argue that the current setup where 50% of the orders are anonymised you would actually get more information based on analyzing the ratio of anonymous vs non-anonymous executions.

An example of this field: 844075693525 with next fill of 844075693531

*Maybe* then you could deduce that there were 5 orders between these that were canceled before the fill but I don't know how that will help given that this is a post-fill analysis in the public feed.

 

[Zzzz1]

It would help if all order IDs would have to be displayed to the public (I mean on the feed) for everyone. One could indeed glean a lot of valuable information about the order book and queue positioning.

There's no identifying information in the OrderID field. It's useless for gaining insights into anyone else's trading. One could argue that the current setup where 50% of the orders are anonymised you would actually get more information based on analyzing the ratio of anonymous vs non-anonymous executions.

An example of this field: 844075693525 with next fill of 844075693531

*Maybe* then you could deduce that there were 5 orders between these that were canceled before the fill but I don't know how that will help given that this is a post-fill analysis in the public feed.

 

[jjw]

It would help if all order IDs would have to be displayed to the public (I mean on the feed) for everyone. One could indeed glean a lot of valuable information about the order book and queue positioning.

All orders for contracts traded on COMEX and on NYMEX have already had their order ids visible for several months. Along with this visibility comes the display of a full order book, not just 10 levels deep. Check out GC and CL to see depth of more than 100 levels on each side. By the end of May, it is planned that this visibility will be complete for all orders on the CME and on the CBOT. Along with the order id is information from which position in queue can be derived. In the near future, we expect to release a version of R | Trader Pro that will show position in queue for a trader's orders in the order book window and in Microsoft Excel spreadsheets.

 

[garachen]

That's good information. But it should come with the disclaimer that it has nothing to do with the OPs question. You are talking about OrderIDs in the MBO feed. He's asking about the visibility of orderIDs in the trade summary message. It's the same ID but they serve very different purposes.

 

[Zzzz1]

I don't know those details as I don't trade those contracts. All i was saying that with order IDs visible and an itch feed one can build a complete order book and know exactly where you are in the queue on each level.

All orders for contracts traded on COMEX and on NYMEX have already had their order ids visible for several months. Along with this visibility comes the display of a full order book, not just 10 levels deep. Check out GC and CL to see depth of more than 100 levels on each side. By the end of May, it is planned that this visibility will be complete for all orders on the CME and on the CBOT. Along with the order id is information from which position in queue can be derived. In the near future, we expect to release a version of R | Trader Pro that will show position in queue for a trader's orders in the order book window and in Microsoft Excel spreadsheets.

 

[i960]

https://ftalphaville.ft.com/2017/05/19/2188959/consider-avoiding-icebergs-this-summer/

Consider avoiding icebergs this summer

MAY 19, 2017
By: Alexandra Scaggs

Information is a funny thing: More is usually better. But once there’s enough data out there, only specialists have the time and ability to parse it. At that point, those specialists are the only ones who can turn information — YOUR information, usually — into profit.

That’s true for the internet, and it’s true for electronic financial markets as well, since they’re basically ultra-fast standardised platforms for information exchange. Really, markets even had their own a version of the “fake news” panic.

Anyway, once CME Group rolls out a new data feed on its US exchanges this weekend, investors trying to conceal large trades could be at an informational disadvantage. First of all, the new offering involves a much higher volume of data, industry sources say, which requires up-to-date infrastructure and extra computing power to process and analyse.

Another key point: Each individual order will have a distinct ID number that everyone can see.

CME says it’s introducing the feed “to increase transparency and customer choice,” because this type of data has been offered for a while by other exchanges.

But if you’ve been following along, you know why this might be a problem. When investors need to make a large trade, they try to conceal their intentions from the broader market. Otherwise, they’re basically walking into a BMW dealership and announcing they MUST buy a new 6-Series today. (It should be said that the car dealer wouldn’t be evil or bad for giving them a higher price — he’d be drawing a conclusion about their demand and pricing the vehicle accordingly.)

Traders on CME’s exchanges have historically been able to conceal sizable trades fairly easily, by using an “iceberg” order qualifier offered by the exchange. They’re called “icebergs” because they only show part of the total size of a trade at any given time. Yes, like the tip of the iceberg.

But with this new data, iceberg orders offered by CME will have the same identification number through the entire trade, according to two industry sources.

This is clearly not ideal for a trader who doesn’t have the budget to hire programmers or a technology vendor. While it won’t be clear how large the total trade is, it will be obvious that there’s more demand than meets the eye. Both sources said they expect high-frequency trading shops to find some way to take advantage.

The problem is pretty easily solvable, of course. A trader could either write an algorithm that splits up the order before sending it to the exchange, or use one from a vendor.

But it’s unclear that the traders who rely on CME-provided iceberg orders — probably ones without huge technology budgets — are even aware a change might be necessary. The exchange announced the new data feed, of course, but like most things in market structure, the industry-provided wording is hard to decipher.

We asked CME if they have alerted traders about the fact the iceberg orders will be displayed this way, and will update with any response.

 

[jjw]

It is my understanding that iceberg orders on the Globex exchanges are no longer supported.

With respect to:

Traders on CME’s exchanges have historically been able to conceal sizable trades fairly easily, by using an “iceberg” order qualifier offered by the exchange. They’re called “icebergs” because they only show part of the total size of a trade at any given time. Yes, like the tip of the iceberg.​

Though such an order did conceal its total size to the market, when the tip of the iceberg was filled, only then a new tip of the iceberg would appear, but at the back of the queue. It is not the case that the rest of the order had to be filled before other orders could get filled. The beauty of the iceberg was that the next quantity visible from the order would become visible as fast as possible, much faster than breaking up an order into smaller ones and submitting the next small order once the trader was notified that the previous order was filled. And an order was identified as an iceberg simply by submitting it with a minimum display quantity indicated, so special (expensive) technology was not needed.

Perhaps the CME could restore icebergs and show the non visible portion of the order as part of the implied quantity for the contract at the specified price?