Market Delta by bolter

A lot is said about this, but I've yet to see a comprehensive comparison of IB and a full tick feed. My gut feeling is that it is 'good enough', though some advantage may possibly be had with a full tick feed. Some exchanges aggregate ticks so what constitutes a 'tick feed' may be a bit of a moot point.

From writing my own MD charting code and looking at a number of SIFs (Asia, Europe, US) IMHO the IB feed is far from useless for MD. I am talking about my own implemetation here, so I can't vouch for Realtick, but it would be pretty surprising if they weren't handling the IB feed correctly.

The charts you posted seem to be tick charts. Correct me if I am wrong about this. Tick charts will always look different (maybe a lot different) because of the different number of ticks and different starting points for each bar. A fairer comparison would be time charts, maybe in the range 1 - 5 min bars.

 

jimmy,
We're on the same page. I like to have at least one other positive confirmation before taking a divergence trade. Volume is usually key in a divergence trade. If we also happen to be approaching a key S/R - then we really are good to go.

Previously I used to use just raw volume numbers and/or volume bars, looking for a reduction in buying/selling pressure. However, with MD I get a much better view of the internals using the delta bars (VB). Case in point on the ESTX today (I've added raw volume for demo purposes) .....



At label 2 you can see a lower low in price with an oversold and diverging TSI, with decreased volume. Normally this would constitute a reasonable entry point. However, the delta bars (ignoring the wick) were not diverging at label 2. That is, there was more net selling on the second probe, despite less absolute volume, hence the divergence setup is not yet valid.

The market went on to make a lower just prior to label 3 where the delta bars did confirm the divergence. This was 3 bars before the actual low so you had plenty of forewarning. At that point you watch the footprint and look for evidence of selling drying up and/or buyers stepping up. As you can see 3996 @ 18:34 did attract buyers and sellers were done at that point. Perfect entry - low risk and no heat whatsoever.



Great demonstration of how MD's unqiue view of the action can put you 1 step ahead of the crowd.

One last point here - I can tell you that small traders were buying that entire move down from 4005 to 3995 so most of them got hurt. But we'll cover this type of analysis in a later post.

Bon chnace,
bolter

 

for comparison

 

for comparison

 

Bolter


Interesting Eurostoxx example

If you saw a very high sell off Delta at the end of the move would you consider this to negate the delta divergence.

 

hi jl12,

Au contraire mon ami. It would if anything improve the setup in my mind - a sign of potential capitulation. A bottom where selling just dries up may mean the sellers are not yet exhausted, and will wait to sell into the next pullback.

Please, if I may, no more esignal vs IB screenshots. We've got enough to make a comparison.

bolter

 

With regard to IB data and MD - Chad (software guru, MD/IRT developer and all round nice guy) has just conducted a study on this very issue. You may be interested in his findings:

http://www.linnsoft.com/qa/a/108.htm

 

I’m struggling to find time to devote to this thread at the moment so I’ll just post interesting MD related setups as they occur.

One of the really neat forms of analysis that MD enables is the ability to study what the big guys are doing versus the small guys. The VB indicator and the footprint charts allow you to filter trades by size (<,> etc), like most T&S windows.

Coventional wisdom suggests that one should “fade the retail guys because they are always wrong”. I find a certain amount of irony in the fact that this axiom is even widely held by retail guys themselves (eg: most ETers). Anyway, be that as it may, I think this is largely received wisdom.

One caveat to this sort of analysis however. It works well on liquid markets where decent size can be done. In thinner markets (eg: one that I trade sometimes the MSCI Taiwan on the SGX) settings filters sizes of say 50 lots to polarize the big guys is a futile exercise because nobody can do this size. There maybe 50 lots orders done in the market, but they will go across the tape as partials (5, 2, 6, 3, 8 etc.).

So here’s my thinking on the big vs small guys debate (refined with aid of MD):
1. In most markets only the big guys can really move the price meaningfully. That is, unless a move has “institutional participation/sponsorship” it ain’t going too far and probably should be faded.
2. The small guys are generally FIFO – ie: they are often early, entering and exiting.
3. A lot of moves are started by the small guys with the big guys coming late to the party (in fact they are the party in most markets).
4. In some markets the small guys are fairly good traders (eg: ESTX).
5. Sometimes the big guys and small guys are completely in sync (that’s when the market can really move), other times you can watch the big guys selling to the small guys etc.
6. You need to study what proportion of the total trade is done by the small guys (say <10 lots) in order to understand what impact they have on the market. For example, I would estimate 10% of the ES, 5% on the ESTX, and 50% on the Nikkei (SGX). Hence, you could filter out the small guys complete on the ESTX, but you need to watch them closely on the Nikkei.
7. But generally, you do want to be on the same side of the trade as the big guys (assuming your profit horizon is more than a few minutes).

Having said all that, there is a danger in overly generalising. Each market has it’s own characteristics and should be studied without preconceived notions.

One last point here. It is worthwhile to bear in mind that the market is comprised of many traders with different timeframes and objectives. An order to sell going across the tape is not necessarily an indication that that trader is anticipating the market heading lower in the 5 minutes, 5 hours, 5 days, whatever. This is especially true the larger the order is. A 1000 lot sell order in the future may be from a mutual fund who wishes to hedge the $50 million they have ploughed into the cash market this morning.

So how do you profit from all this wonderful data. Let me show you one example from today. I’ll post others examples going forward.

First some background. I’ve found that the cumulative delta (with the small guys filtered out) is a great measure for the “health” of a move, and that any divergences from price should get your attention. If you remember this is the blue connected line beneath my volume chart.

In the Nikkei on Monday we had a decent rally just after the open. It then made two legs higher during the course of the day but the big guys were not participating. This morning the market gapped open and continued to rally for the first hour, again without any interest from the big guys. Here’s the chart:



The divergence is obvious to a blind man what? Clearly you need to keep your longs fairly tight and look for an opportunity to get short and then hang on. Not surprisingly, without participation from the big guys, the market ran out of steam and headed south. Not a huge move to be sure, but the divergence has not fully resolved itself. So I would suggest in the next day or two we will see more downside or the big guys will be big net buyers to create a convergence.



This is not merely a well chosen example, when I see these divergences it gets my attention. So, I find it certainly pays to keep tabs on who's doing what.

As an afterthought:
On the nikkei my filter is set to >10. Don't just pick a number out of the air. Study a long term chart to find a filter setting that creates a good proxy for price. I exported the data and ran correls in excel. You should get better than 0.50.

Happy hunting.

bolter

 

I don't have a clue what size to filter with the es. I have delta bars with no filter plotted as a custom histogram and trades >99 plotted in the same pane as a hollow histogram so I can easily see the difference. There are times when the little guys are more active than the big ones and clues are provided. Some times the hollow histogram is bigger than the one showing the unfiltered trades. Not sure how this can happen.

I also have been plotting the filtered accumulated delta the last few days and haven't seen situations where this provided any clues. I changed it to show trades <50 and found many divergences that worked and breakouts/downs preceding price. I admit I'm new to all this and haven't been watching this very long but am confused with what I'm finding.

I've been using delta to trade for a little over a month. Recently started using a 30 second chart at the open and going with the first surge in the direction of the open in relation to the previous days MP vol. based POC, don't have footprint charts and imagine they would be better for this. An idea from Bolters MP thread using the open/POC relationship as a descretionary bias. This has worked quite often but only into the first s/r zone. Again this hasn't been tested or done very long but plan to continue following the idea. Today for eg. was a setup to look for a surge in delta to the downside but it didn't happen and the first significant sign of commitment was to the upside. Using delta may be a way to decide whether to fade or go with a gap open.
Sorry to dump so much at once on this thread but I'm also curious on peoples ideas of the delta bars vs. the delta buy-sell bars. Once a trend starts it looks as though the tails on the delta bars can help with entries on pullback. Any thoughts?
For reference most of this is being observed on a 2 min chart.