Jump to content

Welcome to the new Traders Laboratory! Please bear with us as we finish the migration over the next few days. If you find any issues, want to leave feedback, get in touch with us, or offer suggestions please post to the Support forum here.

  • Welcome Guests

    Welcome. You are currently viewing the forum as a guest which does not give you access to all the great features at Traders Laboratory such as interacting with members, access to all forums, downloading attachments, and eligibility to win free giveaways. Registration is fast, simple and absolutely free. Create a FREE Traders Laboratory account here.

TheNegotiator

Delta Volume in Intraday Trading

Recommended Posts

Hi,

 

This is a new thread to discuss all things Delta - referring to the difference in trades placed at the bid compared with the offer. So for example if 10000 contracts trade at one price over a 5 minute period in the E-Mini S&P 500 contract, 6000 bought at the offer and 4000 sold at the bid, the delta of that price would be +2000. The premise is that the traders who do their business at market are the more aggressive participant. Another example would be in the same 5 minute period, regardless of price, 50000 contracts are traded. Let's say 20000 were bought at the offer and 30000 were sold at the bid. That 5 minutes would be said to have a delta of -10000.

 

For me, delta is a great visualisation of one way action that may lead to a decent move during the day. You can see if you are watching say a 15 or 30 minute chart the imbalance of trade building up and it can be a great indication of continuation in a particular direction.

 

Another way in which I have more recently found delta useful is by watching cumulative delta compared to price movement. Cumulative delta is just like a price chart but with delta. So let's say you have 5 minute bars and you have deltas over the course of 15 minutes of +7500, -1500 and + 4000. You would get a delta plot of +7500, +6000 and finally +10000. So I have found that for example when the cumulative delta is pushing down and overall price is drifting up, there is an implication that there is an underlying bid in the market which is strong enough to absorb that aggressive trade and as such often the market can then break in the direction of price. This is divergence.

 

I also know that with tools from certain vendors, it is possible to look at delta action in much greater detail. Traders might look at say delta at extreme prices when the market is retesting these extremes to judge better the effort of the market than simply looking at volume.

 

There are many ways in which delta can be applied and interpreted and I feel it could be useful to discuss them here. I am happy to answer any questions on the subject, although there will be many who know more than I do!

 

TheNegotiator.

Share this post


Link to post
Share on other sites

Thanks for opening this thread Negotiator.

 

Very timely for me as I have recently started looking at these metrics myself.

One of ways I am trying using Delta is to identify trapped retail, as this is my preferred setup.

 

What Delta tracking tools are you using?

 

On what platforms/feeds?

 

Are there any you have tried and rejected for some reason?

 

 

Thanks and look forward to a good discussion and sharing research on this topic.

 

 

JohnBly

Share this post


Link to post
Share on other sites

Currently I am using Ninja/zenfire. There is an add on which is pretty good for a monthly subscription but I don't use volume breakdown at price at the moment at least. I have used market delta in the past with CQG and this is very good. I did feel though it was somewhat a waste considering the cost and the basic way I used it.

 

How are you looking for trapped retail? With the way CME unbundled some of the tick data, isn't it easier to have commercial appear as retail? I'm sure it is possible but I've never really gone into such depth.

 

Cheers,

 

TheNegotiator.

Share this post


Link to post
Share on other sites

[quote name=

With the way CME unbundled some of the tick data' date=' isn't it easier to have commercial appear as retail?

[/quote]

 

Yes the rub isn't it?

 

 

Here is a description of the new CME reporting I ran into:

 

"As of last Monday( early 2009), the CME changed the way that they disseminate trade information for equity index futures. Due to the changes, there are now 2.5 times more reported trades in the e-mini SP500. Here are the highlights and potential ramifications:

 

There used to be, on average, 1800 trades a day of 199 contracts or more. After the change, we are averaging about 400 trades a day of 199 contracts or more. In addition, the average trade size used to be 12 contracts. It is now between 3-4 contracts. Needless to say, it has become very difficult to follow "the smart money" (large buyers and sellers). This is because trade reporting is based on counterparties. Prior to the changes, if I placed an order to buy 100 contracts at market, the tape (time and sales) would show a single 100 contract market order executed, or possibly one 25 executed market order and a 75 executed market order (or something to that extent). After the CME changes, my 100 contract market order will now be reported based off of each counterparty. Therefore, if it takes 70 different counterparties to fill my 100 market order, the tape will reflect all 70 trades individually Keep in mind that it could be a single counterparty executing my trade, they could just be offering out 70 one and two lot offers as opposed to a single 100 lot offer (I called the CME, this is how they explained it to me)".

 

As for tracking retail vs block under these new conditions, I need to understand the exact CME trade reporting algos better before attempting to create a tool.

 

One basic concept I have is to track the mkt order volume over very small timeframes. If a 2000 lot block mkt order is being bled out in small lots, they are going to want to get it done fast so as not to telegraph, so one would expect flurry of trades at bid/ask over small time period (ei 1 sec). So perhaps this kind of thing could be filtered for.

 

If anyone feels they have a good detailed grasp of the CME trade reporting situation maybe they can post. Thanks.

Share this post


Link to post
Share on other sites

Hi,

 

Prior to the prop shop I currently work with, I worked for a decent sized prop shop in Chicago. They never put through a large order on the CME, they always "shredded" their orders, in fact, they had algorithms just for breaking up the large orders into randomly sized smaller orders (1-3 lots).

 

They also stacked the book in such a way as to always be near the top with some orders and yet have others sprinkled throughout. This allowed them to quickly pull an order from the front of the book if the algo "didn't like" the order flow but still retain some position in the queue in case the order flow returned to something that could be traded.

 

One major question - do you know of any completely unfiltered, non-coaleseced data feeds that are not institutional in terms of cost? If you are only getting snapshots, regardless of how frequent, I don't know of any way to guarantee the trade is lined up with the actual inside bid/ask of the book.

 

Without unfiltered data you don't really have an accurate delta because even though it looks like a trade went off at the ask, it may have been the bid when it was initiated.

 

Best Regards,

Scott

Share this post


Link to post
Share on other sites

they always "shredded" their orders, in fact, they had algorithms just for breaking up the large orders into randomly sized smaller orders (1-3 lots).

 

Scott, wondering how "Shredding" was used exactly.

Do you mean market orders to open a position were bled out discreetly in this way to hide their footprints?

 

One major question - do you know of any completely unfiltered, non-coaleseced data feeds that are not institutional in terms of cost?

Without unfiltered data you don't really have an accurate delta because even though it looks like a trade went off at the ask, it may have been the bid when it was initiated.

 

Researching this myself.

Someone I know who relies on Delta work swears by DTN.IQ.

I have ZenFire, CQG and Photon here so I will be comparing them over the next few weeks.

But part of the problem with this matter is that apparently the exchanges themselves bundle the data. I believe this is true primarily when order flow is high. So Time & Sales may be getting bundled when things are busy. This may mean misreporting whether hits are at bid or ask, and also bundling several small lot trades into single bigger trades. Perhaps someone who understands this better can comment.

Share this post


Link to post
Share on other sites
Currently I am using Ninja/zenfire. There is an add on which is pretty good for a monthly subscription but I don't use volume breakdown at price at the moment at least. I have used market delta in the past with CQG and this is very good. I did feel though it was somewhat a waste considering the cost and the basic way I used it.

 

I was using eSignal when I became interested in market delta analysis. The problem was that eSignal would not backfill bid/ask volume plots. I opened a demo account to try out Ninja/zenfire but it seemed to have the same problem. It sounds like you're alluding to the Final Market Balance add-on but all I need is the basic delta (bid/ask/difference) plot and a plot of the cumulative delta. Are you getting backfilled delta volume plots with out-of-the-box Ninja Trader/Zenfire?

Share this post


Link to post
Share on other sites

Imo, the issue about bundling is only a problem if you are looking for specific traders. When I started on eurex, it used to be obvious when paper came in. Then they made it tough. I decided that it had become best guess and it was better to look at the context of trade occurrence rather than specific size. You can still get a feel generally if you are decent at DOM reading, but I never rely on this alone. So for me, it's about delta with context.

Share this post


Link to post
Share on other sites

John, shredding was used for a couple of reasons - to disguise their intentions and to maintain a predictable position in the queue at a given price.

 

Let's say the ES is 800 on the inside bid; the algo would put in 50 orders in those 1-3 lots sizes, wait a few milliseconds to a couple of seconds and do it again. They do this over and over until they have 300 - 500 orders scattered through the 800 (which by now might be 2000 depending on what other players are doing.

 

Here is the part that is difficult to explain: they didn't necessarily want those first orders, they were just placeholders. They would then cancel order and see if other algos followed suit thus allowing the orders further back to move forward. They would cancel orders/add new ones until they felt they were in just the right place in the book and then let them rest to execute or get pulled if the market moved the wrong direction.

 

The goal is to be at the top of the queue with a large number of orders behind you. Because their commissions were less than $.50 a round turn, scratching a trade meant nothing. So if you are filled and there are a lot of orders behind you if you want out, you just hit that side and scratch the trade.

 

Because they are doing this on both the bid and ask, the ideal trade is to be near the top of the book on the opposite side and once they get the first side filled, they get out virtually immediately on the opposite side with a 1 tick profit. They do this literally hundreds, if not thousands, of times per day.

 

I may not be doing a very good job of explaining this but it is a random walk trade. If you have the speed, commission structure and capital to have literally thousands of orders outstanding at any given time, it is the lowest risk trade in the universe.

 

As to the other question; thanks I will check those out. The exchanges can burst data out but it will be consistent at the timestamp level regardless of when it is actually received. Again, there was an entire group of programmers whose only task was to write data integrity/alignment algorithms.

 

Scott

Share this post


Link to post
Share on other sites

Scott,

Thanks for describing that.

Pretty clever system they were running, with very low risk it seems.

But couldn't do it without that $.50 commission...

 

When that firm had market orders to place, did they shred those too?

Share this post


Link to post
Share on other sites

They didn't normally use market orders or stops because they always had orders on both sides of the market. They shredded all orders.

 

There are literally dozens of things like the one I described. They combine to make it very difficult to compete in the micro scalping (essentially market making) arena.

 

However, with any directional component involved the playing field gets leveled pretty quickly. Once you are going for more than a couple of ticks things get a lot move even.

 

There are still things you can do with an ultra fast data feed that a normal retail trader can't - think of all the times you have been stopped out because a small order (sometimes a 1 lot) hits your stop price with a gazillion orders at that level.

 

Your stop is triggered but it was one of a handful of trades at that price. With a fast enough feed, an algorithm can monitor the number of contracts and if it gets below a threshold you hit the market, otherwise you let it ride. Often you end up stopping out anyway but for the high frequency player, that technique alone saves thousands over the course of a month.

 

Scott

Share this post


Link to post
Share on other sites
They didn't normally use market orders or stops because they always had orders on both sides of the market. They shredded all orders.

 

There are literally dozens of things like the one I described. They combine to make it very difficult to compete in the micro scalping (essentially market making) arena.

 

However, with any directional component involved the playing field gets leveled pretty quickly. Once you are going for more than a couple of ticks things get a lot move even.

 

There are still things you can do with an ultra fast data feed that a normal retail trader can't - think of all the times you have been stopped out because a small order (sometimes a 1 lot) hits your stop price with a gazillion orders at that level.

 

Your stop is triggered but it was one of a handful of trades at that price. With a fast enough feed, an algorithm can monitor the number of contracts and if it gets below a threshold you hit the market, otherwise you let it ride. Often you end up stopping out anyway but for the high frequency player, that technique alone saves thousands over the course of a month.

 

Scott

 

Scott,

 

Fascinating insight into some of the inner technical workings that make up the market. In this case though, it doesn't seem as though it is directly affecting the delta. Correct me if I am wrong though.

Share this post


Link to post
Share on other sites

I always found Delta to be most useful in context. If you constantly try to read it, it will drive you nuts, but if you are trading pivot points, for example, it can provide a look at order flow at the pivot.

 

One thing you can look for is extreme deltas with failure to break the level significantly. Example: Price declines to a support level. You look at the delta, which shows a 5000 selling delta one tick below the pivot. You also notice that prices have stalled. You could take this to mean that passive buyers (limit orders) are absorbing all that aggressive momo selling. If you are in a mean reverting instrument like Spoo, you want to be on the lookout for a pop. This can help you exit your short profitably, and/or put on a long with some confidence.

 

It doesn't always happen this way of course, but there's an example of how I think Delta should be used. The important thing to remember is that screen time and context are everything as usual.

Share this post


Link to post
Share on other sites
Thanks sdoma. I definitely think that this is the sort of idea that is very useful to intraday traders. What are your thoughts on cumulative delta?

 

I will preface this by saying that I haven't done as much back testing on cumulative delta as I'd like. From what I remember, I wondered what information it gave you that simply observing price action couldn't. I would have to guess that you'd want to look for divergences between cumulative delta and price, but again, could a CCI give you that?

 

To sum up, from my recollections I wasn't sure if I saw an edge or not. I will come back to it eventually to test it.

Share this post


Link to post
Share on other sites
Please elaborate on the signs that gave a hint about "paper coming in".

 

"Paper" means institutional order flow. In the pit, it was obvious since the broker was standing right next to you as he did the orders. On the screen, you could see it because the orders were larger. Also, initially I think there were names attached to orders in time and sales.

 

Now, institutional order flow is chopped up to make it harder to see. Institutions don't want other participants jumping on and affecting the quality of their fills. So, instead of 2000 in the S&P it will be done in ones, twos, and threes by an algorithm.

Share this post


Link to post
Share on other sites
"Paper" means institutional order flow. In the pit, it was obvious since the broker was standing right next to you as he did the orders. On the screen, you could see it because the orders were larger. Also, initially I think there were names attached to orders in time and sales.

 

Now, institutional order flow is chopped up to make it harder to see. Institutions don't want other participants jumping on and affecting the quality of their fills. So, instead of 2000 in the S&P it will be done in ones, twos, and threes by an algorithm.

 

Exactly. I think you are right about the names in t/s but I never saw that when I first started. I was told it was separated into different groups though rather than specific names.

Share this post


Link to post
Share on other sites
Delta or cumulative delta is completely useless IMO there is no edge as most of the volume is done in a limit fashion very little percentage of the volume is done by aggression.

 

AuctionMarket_Trader,

 

I am glad you qualified your statement with the all important IMO, otherwise I would have to take you up on this statement. As it is though I have a question for you. When a market auctions from one price to another, what is the relative ratio of price moves which happen by pure movement of bid/ask order queue to the next price, to the number which happen due to the bid/ask being traded out? You say most volume is done by limit orders. How do you think those limit orders get filled? I would question the depth to which you are applying the delta concept and the level of context you use it with.

Share this post


Link to post
Share on other sites

looking at delta is like looking at a candle chart when price goes up so does delta so i dont see a point there, as far as delta divergence or anything like that their is no point because when price goes up in an aggressive fashion the orders are pulled out of the que so the delta will be magnified not showing the real intent or the actual quantity or strength as i said before better off analyzing responsive action. There are many resources and people who code the algos for major firms that agree to this for more info you can look at this papers which are very hard to read but you will get an idea http://www.tradingphysics.com/Resources/Papers/

Share this post


Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.


×
×
  • Create New...

Important Information

By using this site, you agree to our Terms of Use.