Volume Splitter

[daedalus 10]

Just wanted to share this chart of blowfish's code that I let run on the 1min es today with a 50 contract filter... (thanks again blowfish!).

 

Even in this "beta" state I can see some interesting interactions with volume and price. Particularly in high trending areas throughout the day.

 

All I know is this... I took three trades today, two of which were losers and both were against the volume histogram at the time and the one that did go onto work had the backing of the histogram.

 

Look at some of the divergence patterns you see as well as some of the cyclical timing. I think we (well mostly you, I haven't contributed much) are developing something that could be very helpful!

 

[bathrobe 10]

If anyone can code this in Ninja Trader and is willing to post it, Thanks in advance

[AgeKay 10]

I find it interesting that everyone here except BlowFish found it so hard to come up with a version even though the concept is really simple if you understand how trading works (i.e. order matching and difference in order types).

 

You should all read the two books BlowFish recommended before wasting more time on figuring out indicators you don't understand. Please don't flame me though for that comment. This is well-intentioned advice.

[BlowFish 10]

Daedulus glad you are finding something of interest in it. These sorts of indicators seem to make great divergence patterns if you fiddle with the smoothing. My guess is the original indicator is probably smoothed as without things can be a bit spiky.

 

AK, apart from the book references there is other good information in this thread which people seem to have missed, it can be kind of frustrating when you provide facts that would dispel a lot of the confusion if carefully read.

 

Just one more thing (as Columbo might say). These sorts of indicators are all well and good but order flow (which is essentially what they are trying to depict) can turn on a dime just as price can. There is no 'magic' indicator, any one of the many delta indicators (with suitable block filters) will do the job. It is easy to become obsessed by the indicator rather than the application of the indicator to make trades. In fact generally it is better to work backwards from a trading concept to an indicator rather than the other way round.

 

Edit: should add a couple of :), again meant as well intentioned advice.

[RichardTodd 10]

wow, I just want to say I'm flattered that this indicator gets so much attention in these forums, and the ninjatrader forums, and elsewhere. These clone attempts seem to focus on the amplitude/shape of the splitter, when that's only half the story it's telling.

 

Best of luck to you!

Edited May 12, 2009 by stanlyd
promoting service

[swansjr 10]

  RichardTodd said:

wow, I just want to say I'm flattered that this indicator gets so much attention in these forums, and the ninjatrader forums, and elsewhere. We have several other unique indicators, so if page after page of clone attempts earns me an ad, I'd suggest you all check out eotprolive.com. I mean... For some of you, I'm sure trying to figure it out is reward enough, but maybe others would rather be our clients, who have been actually using our splitter to make money since July of 2008. Plus, all these clone attempts seem to focus on the amplitude/shape of the splitter, when that's only half the story it's telling.

 

Best of luck to you!

 

Yeah, the other half is the number of "big" traders participating in a given move, right? Anyway, EOT does have some great indicators from what I’ve seen.

 

You're right about coding this up as being part of the fun. So, now I have to ask, are we having fun yet?

[Tasuki 10]

I just got an email from Bill Dennis. It appears he's using Multicharts, but says that the indicator will work on Tradestation. If TS won't give us access to Time and Sales data for Easylanguage, I'm not sure how that's possible, but some of you seem to think that "clever programming" will do the trick. Gosh I hope so, because the EOTPROLIVE indicator does seem to have real value.

 

BTW, does anyone have the link for the thread on the TS forum for voting on the necessity of having T&S data for EL?

[RichardTodd 10]

  Tasuki said:

I just got an email from Bill Dennis. It appears he's using Multicharts, but says that the indicator will work on Tradestation.

 

I have coded the indicator for Tradestation, Multicharts, Ninja, and eSignal. Platform differences meant changes to some of the details of the calculations, but we have essentially the same indicator on each one. In fact, I did the tradestation version first.

[BlowFish 10]

  Tasuki said:

I just got an email from Bill Dennis. It appears he's using Multicharts, but says that the indicator will work on Tradestation. If TS won't give us access to Time and Sales data for Easylanguage, I'm not sure how that's possible, but some of you seem to think that "clever programming" will do the trick. Gosh I hope so, because the EOTPROLIVE indicator does seem to have real value.

 

BTW, does anyone have the link for the thread on the TS forum for voting on the necessity of having T&S data for EL?

 

Tasuki you can. If you look at the code I posted it shows you how. It may or may not bear any resemblance to Richards indicator, I honestly don't know, but it demonstrates the principle. If you are not so good at EL I also explained in words a few pages back how you can achieve it.

[bgtrader 10]

Hi Guys,

I'm new here, and I've been pondering the volume splitter and wanted to add an idea. First, thank you all for upholding such a refreshingly good vibe sharing/learning space. I believe another important variable in making a volume splitter useful is how it would "attempt" to calculate whether the big traders are "net long" or "net short"...so, I believe that it might be good to include a type of "cumulative" function. For example, you could code it to show the sum of buy/sell orders starting at market open...so as the indicator line progresses, it "attempts" to show whether folks are net long throughout the trading session. another cumulative idea, would be to show the total buy/sell volume during each pivot cycle, so it would have a kind of adaptive average accumulation of buy/sell volume...and another idea would be to cleverly include both ideas. I'm not a coder yet, just a trader and student of market techniques. Looking forward to you folk's input. -best wishes, B

[RichardTodd 10]

I am going to try to ward off additional moderation by stating that this link is intended to help you all. I've been watching people on forums try to duplicate this for almost a year now, and so far every time they've forgotten half of the story in my opinion. I apologize in advance if that sounds "arrogant" to anyone.

 

This is one of my earliest blog posts on the splitter while it was under development (we called it "volume splatter" back then!), which describes how the dot sizes it produces differ, and why I think that's so important to the story:

 

Volume-Splatter Indicator: Don’t Follow the Small Traders :: Move the Markets :: Entries ::

 

The fact is, if you want to follow the big traders from a pure delta perspective, splitting out their volume on the e-minis is hardly necessary... it's common knowledge that the big trade volume is usually like 70%+ of the total picture.. so if you compare the overall volume to the split-out big traders, the shape of the histogram looks very similar on most markets. Figuring out the story that the market's trying to tell you from their cycles of activity and inactivity is in my opinion much more powerful.

[AgeKay 10]

RichardTodd, I never understand how people can say that big traders win over small traders. You say at least 70% of volume is caused by big traders. At most 50% of traded contracts can be winning. Let's clarify this and break it down. 70% of volume by large traders leaves at least 20% losing (70% total - 50% winners = 20%). But you're only tracking those that are using market orders. Someone has to match these with limit orders which are apparantely all losing. So 70% of large limit order traders have to be losing since only 30% of small market order traders are losing to them. So what you get in the end is that about half of large traders are losers.

 

Maybe you could clarify?

[RichardTodd 10]

  AgeKay said:

RichardTodd, I never understand how people can say that big traders win over small traders. You say at least 70% of volume is caused by big traders. At most 50% of traded contracts can be winning. Let's clarify this and break it down. 70% of volume by large traders leaves at least 20% losing (70% total - 50% winners = 20%). But you're only tracking those that are using market orders. Someone has to match these with limit orders which are apparantely all losing. So 70% of large limit order traders have to be losing since only 30% of small market order traders are losing to them. So what you get in the end is that about half of large traders are losers.

 

Maybe you could clarify?

 

I think you are trying to get to a tidy mathematical answer with too little available information. If you assume all the small traders lose, then that leaves 50/70ths of the big trades winning and 20/70ths of them losing. But like you said that's only looking at the market half of the trades. There's not enough reliable limit-order data to suss out the rest, in my opinion, so that line of thinking is a dead-end. Too many games are played on the order book, even in stock land where they try to tempt you with book data. NOT TO MENTION a big trader can "lose" in the e-minis as a hedge against a huge gain another market, and vice-versa. And that's just one example of how much more complicated real life is.

 

People are always asking me "since there's a limit order on the other side how can you tell who's long or short?" and people are always asking me "how can you tell if they are buying or covering a short?" In both cases I don't actually care (though I would love to have real-time open interest to help me work out the second case). People that ask me this don't understand what the splitter is trying to show them.

 

Let me tell you my concept for why the splitter 'works', which some may disagree with but which satisfies me: Let's say you've got thousands of contracts to buy. Doesn't matter if it's to open or close a position. You've got thousands of contracts to buy, and you can't be bothered to try to hide it with limit orders and small buys here and there. Well then you must be pretty sure the market is about to go up. Otherwise you would never show your hand with a big display of market orders. If you do this too often and the market doesn't rise, then natural selection will make sure you aren't a large trader much longer! So, when the large traders are too worked up to put up smoke-screens, there tends to be a move in the making. The only other reason they come out buying in droves is in the face of big liquidity on the offer (like at major resistance points), because that's a time they can get off large orders without being slipped. If you are careful you can watch for this and differentiate the cases.

 

I actually take it as axiomatic that the big traders tend to win, ever since I read "the poker face of wall street" which I think makes the case for them pretty well. My reasoning above, plus months of good-looking charts, has me satisfied that the splitter identifies the case it is looking for well enough to be a useful part of my trading toolbox.

[bathrobe 10]

Wouldn't a large number of the bigger traders be other time frame traders and not at all concerned with their position on an intraday basis.

[RichardTodd 10]

  bathrobe said:

Wouldn't a large number of the bigger traders be other time frame traders and not at all concerned with their position on an intraday basis.

 

Once again my response is: "I don't care what time frame they trade on." Bottom line, if they are so anxious to grab their contracts that they won't bother to iceberg their order at support levels or otherwise hide their activity, then it follows that they think there is no time to waste at the present price level. Any traders that are careless enough to hit the open market with 1000's of contracts in market orders with no good reason seem to be in the minority. And anyway, they wouldn't all happen to carelessly act in confluence very often, which is the only way they'd be more than a blip on the screen.

 

That is my concept, anyway, as I described in an earlier post in this thread. If this board weren't so militant about me posting links I'd tell you where you could find lots more information about this kind of topic. Since it is, though, I leave you with google and my best wishes.

[daedalus 10]

You'll have to forgive some of the mods... but like most trading forums its a full time job keeping the spammers and "vendors" from coming in here and spouting advertisements for their wares.

 

Richard - I have attended many an EOT free thursday night webinars and did a trial run with the basic indicator set over a year ago and the entire EOT crew (you, bill, glen) is a truly standup and standout organization in the profession. The indicators ended up not being exactly for me at the time but I have (and continue) to recommend your site and service to people without reservation.

 

Keep up the good work!

[cooper59 10]

  BlowFish said:

This should do what is required.[HIGHLIGHT RED][/HIGHLIGHT RED]

Blowfish, thank you for all your work. For those of us who are easy language illerate, would it be to much trouble or work to ask you for an ELD?

[Tasuki 10]

Richard, so you're saying that the first part of the equation is whether the big players are supporting a move, and the second part is how many of them are involved, in other words how much pressure they're applying. Is that right?

[BlowFish 10]

  cooper59 said:

Blowfish, thank you for all your work. For those of us who are easy language illerate, would it be to much trouble or work to ask you for an ELD?

 

Hi there, I no longer have tradestation! You should simply be able to 'cut and paste' the code into tradestations editor, or maybe someone else will do it.

[BlowFish 10]

  bgtrader said:

Hi Guys,

....... I believe another important variable in making a volume splitter useful is how it would "attempt" to calculate whether the big traders are "net long" or "net short"...so, I believe that it might be good to include a type of "cumulative" function.......

 

Welcome to the forum bgtrader. I know a couple of traders who use this information. One of their key observations is that it requires a particular net absolute order flow before an intra day trend can develop. Compare the values on days market drift with those that trend. Pay particular attention to what sort of 'open interest' is required for a market to have a sustained break out of the opening range. Sadly this indicator will only run real time as it requires best bid/best ask information.

 

So here you go. Cumulative intraday order flow that accounts for net long or net short positions and large trader filter. I wonder how long before this shows up somewhere as a commercial offering?

 

inputs: 
UpColor(darkgreen), 
DownColor(red), 
DeltaBar(1), 
LargeBlockFilter(100),
ResetDeltaEachBar(0); 

variables: 
MyVol(0), 
color(yellow), 
intrabarpersist MyCurrentBar(0), 
intrabarpersist VolumeAtBid(0), 
intrabarpersist VolumeAtAsk(0), 
intrabarpersist BAVolRatio(0), 
intrabarpersist VolTmp(0), 
intrabarpersist Delta (0), 
intrabarpersist DeltaH (0), 
intrabarpersist DeltaL (0), 
intrabarpersist DeltaO (0); 

if LastBarOnChart then begin 
  	MyVol = Iff(BarType < 2, Ticks, Volume); 
if CurrentBar > MyCurrentBar then begin 
	VolumeAtBid = 0; 
	VolumeAtAsk = 0; 
	BAVolRatio = 0; 
	VolTmp = 0; 
	MyCurrentBar = CurrentBar; 
	if ResetDeltaEachbar = 1 then Delta =0;
	DeltaO = Delta; 
	DeltaH = Delta; 
	DeltaL = Delta; 
end; 
if absvalue(MyVol-VolTmp) <= LargeBlockFilter then begin 
	if Close <= InsideBid then
		Delta  = Delta - MyVol + VolTmp
	else if Close >= InsideAsk then 
		Delta = Delta + MyVol - VolTmp ;  
	VolTmp = MyVol ;
end; 
end ; 


DeltaH = maxlist(DeltaH, Delta); 
DeltaL = minlist(DeltaL, Delta); 


if Delta <= 0 then color = DownColor else color = UpColor; 

plot1(DeltaO, "DO"); 
Plot2(DeltaH, "DH"); 
Plot3(DeltaL, "DL"); 
plot4(Delta, "DC");	 

[agon 10]

Hi there, I no longer have tradestation! You should simply be able to 'cut and paste' the code into tradestations editor, or maybe someone else will do it.

 

Try of Volume Splitter by BlowFish :

!VOLUME_SPLITTER_BLOWFISH.ELDFetching info...

[RichardTodd 10]

  BlowFish said:

One of their key observations is that it requires a particular net absolute order flow before an intra day trend can develop.

 

I'm not sure you'd want a large trader filter for this purpose, though. A simple Cum(upticks-downticks) might be good enough for the cumulative task, and I believe you get history on tradestation if you go that route, though I don't have TS anymore and can't recall. I've also seen people look to the $UVOL vs $DVOL for similar purposes when trading the e-minis.

[BlowFish 10]

  RichardTodd said:

I'm not sure you'd want a large trader filter for this purpose, though. A simple Cum(upticks-downticks) might be good enough for the cumulative task, and I believe you get history on tradestation if you go that route, though I don't have TS anymore and can't recall. I've also seen people look to the $UVOL vs $DVOL for similar purposes when trading the e-minis.

 

I am not sure either to be honest. When I took a look at this some while ago I did not use a block filter.

 

I don't have TS either. Upticks Downticks are a decent 'proxy' for a lot of these sorts of things if you can get those historically. The way this was presented to me was "look at what happens when you get to a 80,000 (just an example) imbalance". Again no block filter was present when it was shown. Stuff seems to happen at absolute thresholds. I am not sure ticks would be a particularly good proxy for that.

 

Incidentally there are a couple of posts discussing this phenomena over at (ET) good luck finding them though! Even if I don't get in trouble for mentioning ET I will probably go to hell for it. :rofl:

[BlowFish 10]

The cumulative indicator I posted a while back has the filter discarding large blocks. Change the <= to >= to filter small blocks. I would recommend not filtering at all in the first instance and see how you get on. That'll teach me not to test code!

 

Oh and plot it as an OHLC bar.

[insideday 10]

  Tasuki said:

BTW, does anyone have the link for the thread on the TS forum for voting on the necessity of having T&S data for EL?

 

https://www.tradestation.com/Discussions/Topic.aspx?Topic_ID=24606