What is DEMAND/SUPPLY volume?

[mister ed]

Well, this is another thread I cannot keep up with!

Been some great points and ideas brought up, I think one of the reasons I cant keep up is there is so much to mull over. Thanks for the charts BlowFish. I too am experimenting with this stuff, there are many combinations and permutations to play with.

Can I return back to the opening discussion Jerry and I were having? I have had a few thoughts. In a nutshell. the definition I have been exclusively using (I wont define it again, its in the earlier posts) up until now is, to my mind, just a “pure� bid/ask volume. I don’t mean that in a value-judgement way, what I mean is it is just data without any interpretation placed on it. The bid was hit? That’s an aggressive sell. The ask was hit? That’s an aggressive buy. (Yeah, OK, "aggressive" is an interpretation, just call it a "buy" or "sell" if you prefer).

The uptick/downtick definition, on the other hand, places an interpretation on the data before it is presented. The bid was hit? Well, before we classify it, what was the most previous price movement, an up- or down-tick? A downtick? OK, well that bid getting hit is “supply� then. On the other hand, if the tick prior was an uptick and that bid getting hit does not represent a downtick then that bid getting hit is “demand�.

See the difference? I would suggest using the uptick/downtick to classify the buys and sells is one step removed from the data. Is there anything wrong with that? Well, actually, I think no, there isn’t; and in fact I think classifying the hits in this way can be extremely useful. DarthTrader has suggested another alternative way of presenting the data, this too looks interesting.

I have attached a chart showing yesterday’s ES (3-minute chart), with the panes below plotting, in order: total volume, straight bid/ask volume, uptick/downtick bid/ask volume. Close examination will show how the two definitions of bid/ask volume show the data differently.

[BlowFish]

I meant to post a couple of 'failures' or 'anomalies' I closed my workspace before I did a screen capture <doh>. I'm going to slow up and mull over a few things. I cant remember what I have said already!

Did I mention price seems to often pass the level that sees the influx of 'paper' by a few ticks before it actually makes the turn? That's interesting. I wonder if its like some kind of test.

Cheers.

Edit: absolutely agree with what you said above btw. I think Aggressive is fine to use if we agree to define this as "is prepared to hit the bid/ask" or "is prepared to trade at the market". Doesn't mean she wants to punch the counter party

[mister ed]

	Quote:	BlowFish
	Did I mention price seems to often pass the level that sees the influx of 'paper' by a few ticks before it actually makes the turn? That's interesting. I wonder if its like some kind of test.

Happens a lot - more to investigate.

	Quote:	BlowFish
	Doesn't mean she wants to punch the counter party

Well, maybe sometimes....

[zdo]

In TS, for the last week or so have been watching KlingerGoslin() and TickMoneyFlow() indicators plotted in the same subgraph using a TickBar data stream.
Scaling for both indicators is on NoAxis, Date range on screen.
KGI parameters are set fast (c,3,10,16,34,55, 13, 1, Darkgreen, Darkred).
(If you need I can post pictures or the version of the KGI I'm using)

Their movement is usually highly correlated but they also show some pretty neat divergences / disagreements, etc.
Is this a ‘squigly lines’ representation of what this topic is exploring?? Thx

[BlowFish]

	Quote:	zdo
	In TS, for the last week or so have been watching KlingerGoslin() and TickMoneyFlow() indicators plotted in the same subgraph using a TickBar data stream. Scaling for both indicators is on NoAxis, Date range on screen. KGI parameters are set fast (c,3,10,16,34,55, 13, 1, Darkgreen, Darkred). (If you need I can post pictures or the version of the KGI I'm using) Their movement is usually highly correlated but they also show some pretty neat divergences / disagreements, etc. Is this a ‘squigly lines’ representation of what this topic is exploring?? Thx

Most likely though it did not get me where I wanted to go. Maybe cause I just don't like squiggly lines! If you have a search over at ET for posts by 5pillars on delta you will get up to speed pretty quickly on delta divergence. I got the impression he's 'for real' though you never know for sure.

Cheers,

[zdo]

Thanks for the reference.

Aren't divergences btwn price and the TickMoneyFlow (standalone ie not considering KGI at all) equivalent to Delta divergences?

[mister ed]

	Quote:	zdo
	Thanks for the reference. Aren't divergences btwn price and the TickMoneyFlow (standalone ie not considering KGI at all) equivalent to Delta divergences?

Hi zdo - I dont use tradestation and am not up to speed with TickMoneyFlow, if you could describe what it is then I could compare it with Delta?

[zdo]

	Quote:	mister ed
	Hi zdo - I dont use tradestation and am not up to speed with TickMoneyFlow, if you could describe what it is then I could compare it with Delta?

MisterEd,

Here is the code I am using for TickMoneyFlow

{-------------------------John Hudson's TickMoneyFlow Indicator------------------------} {This code tries to keep up with how much money is actually entering or leaaving an issue in order to measure the buying or selling pressure on prices. It uses the snapshot Bid and Ask(CurrentBid, CurrentAsk) fields provided by TS to gauge whether a trade was initiated from the buy or the sell side. Although you can never be sure whether a trade hit the tape in proper sequence with the Bid/Ask data, in general trades initiated from the buy side will go off at the Ask and trades initiated from the sell side will go off at the Bid. In the case of a trade going off either between the Bid and Ask or outside the Bid and Ask, this code attempts to assign the trade to the most likely side of the market based on whether the trade was above or below the trade before it or, if the trade before it was at the same price, it determines what the status of the market was of the most recent trade in the past that was different from the current trade's price. This code will run on any time interval..... from 1 tick bars to daily bars. } [LegacyColorValue = true]; Inputs: Length1(2), Length2(89), ShowLinReg(false) ; Vars: IntrabarPersist TickFlag(0), IntrabarPersist OldPrice(0), IntrabarPersist TMF(0), IntrabarPersist NewTickVol(0), IntrabarPersist MyBarVolume(0) ; if BarNumber >= 1 then begin NewTickVol = Ticks - MyBarVolume; MyBarVolume = MyBarVolume + NewTickVol; if (C>=CurrentAsk) then begin TMF=TMF + (C * NewTickVol); TickFlag = 1; end; if (C<=CurrentBid) then begin TMF=TMF - (C * NewTickVol); TickFlag = 2; end; if (C<CurrentAsk and C>CurrentBid) and C>OldPrice then TMF=TMF + (C*NewTickVol); if (C<CurrentAsk and C>CurrentBid) and C<OldPrice then TMF=TMF - (C*NewTickVol); if (C<CurrentAsk and C>CurrentBid) and C=OldPrice and TickFlag = 1 then TMF=TMF + (C*NewTickVol); if (C<CurrentAsk and C>CurrentBid) and C=OldPrice and TickFlag = 2 then TMF=TMF - (C*NewTickVol); OldPrice = C; if BarStatus(1) = 2 then MyBarVolume = 0; end; Plot1(TMF, "TMF"); Plot2(ama(Plot1, Length1), ""); if ShowLinReg then begin Plot3(LinearRegValue(Plot1,Length1,0)); Plot4(LinearRegValue(Plot1,Length2,0)); end; //if ShowLinReg

The key logic is:
if (C<CurrentAsk and C>CurrentBid) and C>OldPrice then
TMF=TMF + (C*NewTickVol);
if (C<CurrentAsk and C>CurrentBid) and C<OldPrice then
TMF=TMF - (C*NewTickVol);
if (C<CurrentAsk and C>CurrentBid) and C=OldPrice and TickFlag = 1 then
TMF=TMF + (C*NewTickVol);
if (C<CurrentAsk and C>CurrentBid) and C=OldPrice and TickFlag = 2 then
TMF=TMF - (C*NewTickVol);


Screen shots are attached for some of today’s action in a 102 tick YM chart.
TickMoneyFlow1.jpg shows the first hour and a half of trading. Notice at around 10:10 the directional and angular congruency with KlingerGoslin (oscillator) started to fall apart – rally was snuffed and TMF led way downward.
TickMoneyFlow2.jpg shows angular divergence at about 14:45 (see white up arrow). Again rally was short lived. Over at around 15:02 the KlingerGoslin ‘rallied’, but it was unconfirmed in magnitude by TickMoneyFlow and further sell off ensued.
TickMoneyFlow3.jpg shows strong buying on the KlingerGoslin from 15:16 through 15:20 presaging the late rally but the ‘aggressive’ action shown on TickMoneyFlow delayed it until ‘aggressives’ capitulated / reversed at around 16:42. At around 15:46, after KlingerGoslin had faltered, the ‘aggressives’ carried the rally into the close…
…maybe

As I compare the code for TickMoneyFlow with MarketDelta code (which, in TS, is in arrested development now because the Bid Ask data is random snapshots / is inaccurate (ie has a median inaccuracy of about 2.5% compared to T&S prints - but can err much higher)) I see immediately that TickMoneyFlow uses CurrentBid and CurrentAsk while MD uses InsideBid and InsideAsk and upTicks and downTicks
But my same original questions hold:
Is TickMoneyFlow a ‘squigly line’ representation of MarketDelta?
Or is it a different perspective on / dimension of 'volume'? and
How does TickMoneyFlow relate to the original question of this thread?

Many thanks.

[mister ed]

I don’t know if I am going to be of much help with the question. The TickMoneyFlow looks to be trying to achieve the same thing as Volume at Bid/Ask Delta, but like you say there are many compromises in TMF, with random snapshots and inaccuracy. It appears that in principle, though, it trying to do the same thing.

As for TMF being just a “squiggly line� representation of the delta – on the chart I have posted delta is represented as a histogram, if it was represented as a line instead it would be a “squiggly line� too. There are many ways of representing the delta on the chart: can use a moving average of it, can apply an indicator to it, can chart its momentum, can sum it, and so on; it can be represented in its various forms as a histogram or line.

The one thing I would say that sets using the volume delta (or the TMF) apart is that it is not a “squiggly line� in respect that it is not derived from the price data, it is not an “indicator� as such. It is distinct data from the price data – sure the two are related, but a squiggly line indicator, like say the MACD or the RSI is derived from price whereas the volume delta, I would argue is not derived from volume, but is volume. Like the standard volume histogram found at the bottom of most standard bar charts is not derived from volume, but is volume. Of course, if you apply say an RSI on the volume delta (which you could do) then that is an indicator. I hope I am not stepping into a semantic minefield here… In short, I would say, as you ask, it is “a different perspective on / dimension of 'volume’�.

Hope that answers your questions zdo – and thanks for the charts and ideas.

[BlowFish]

Yup looks like a similar thing in a different can There are a couple of variations on this theme. There are a few different ways of plotting too.

I personally don't think it matters too much what flavour you use these are quite 'noisy' tools anyway (lets not even get started on smoothing!!). I use IB which aggregates ticks in fast markets anyway (so is inaccurate inherently). I think that maybe why my delta divergence research didn't come to much. As I say that didn't bother me as its not a direction I particularly want to go in just now.

I will carry on saving images of 1 tick charts as that is my particular interest. Trying to get a more visual overview (complete with history) of the tape. Trouble is I keep forgetting to save the images before shutting down!! As I have said (I think) this is to try to improve timing rather than for trade decisions.

Cheers.

EDIT PS the extra code looks like it is to deal with orders between the bid and ask but I only glanced at it rather than carefully examine the logic.