Welcome to NexusFi: the best trading community on the planet, with over 150,000 members Sign Up Now for Free
Genuine reviews from real traders, not fake reviews from stealth vendors
Quality education from leading professional traders
We are a friendly, helpful, and positive community
We do not tolerate rude behavior, trolling, or vendors advertising in posts
We are here to help, just let us know what you need
You'll need to register in order to view the content of the threads and start contributing to our community. It's free for basic access, or support us by becoming an Elite Member -- see if you qualify for a discount below.
-- Big Mike, Site Administrator
(If you already have an account, login at the top of the page)
Aha! It makes a perfect sense. I understand what you mean now. Thank you.
Regarding the tradeflow analysis, It seems to nicely pinpoint the areas of interest where MM is accumulating or distributing.
There's the obvious issue that once accumulation/distribution starts, there's no guarantee that it will stop (as seen in my previous snapshots).
There's no way to know in advance what the inventory level reached will be, but it helps to at least start trading in the direction of the MM after "she" has shown her hand.
In a way it complements one of my strategies, which is "mm-like". To sum briefly, the DAX respects technical levels to a fairly high degree. Min/Max/Pivots/Fibs/Major Trends and EMA/SMA lines are respected for a few ticks at least and can always be a turning point.
It looks like the level of inventory is partially connected to the probability of inversion. Minor waves are highly correlated but so seems to be the failure to follow up with a major wave generation (i.e. the end a major wave and the start a new one in the opposite direction). Those levels are good points to place your orders and size them accordingly in a way to put your "vwap" in a convenient place, or, those levels can be better skipped as the move is likely to continue, and thus a better position would be better for this vwap averaging. I hope what I'm writing is clear to you.
I looks like I really need to allocate time and energy to translate your code in python and see if I can confirm this in a more scientific way.
However, I need to make one thing perfectly clear. Trade Flow, or
Volume Analysis of Time and Sales,
(aka, in my world view, "Inventory Analysis") is only a piece of the
Analytical action...
Analysis of Market Maker's placements on Market Depth are the true
keys to "nailing" trend changes; but you've heard me claim that this
is eXtremely Difficult, and I really mean it.
Improving Inventory Analysis by using a "Fractal" Multi-Timeframe
"True" Inventory Analysis, that I've alluded to (frustratingly, I'm sure)
will yield a significantly Better result that the simple approximation we're
using here... But that's a Big Step, already a "bridge too far" for most...
...and then Depth of Market Analysis, for purposes of Predicting Trend,
is an order of magnitude more difficult than that; sorry to say...
I have no idea what's possible in Python, versus C# or Java which I
use for Multi-Threaded Object Oriented Analysis, etc....... so, I couldn't
help at all. But if you're delving into some C#/NT8+ coding... then I
can provide consultancy.
Well the fractal part is non so hard when processing the data. It just
means you construct a few of your "integrators" with different time lengths.
This is difficult in NT, and definitely not something you want to chart and show
on screen in realtime, as it clutters and we we don't know exactly which
of these timeframes work at any specific point in time. My feeling (to be
confirmed) is that they all work in specific mkt conditions, like MM risk aversion,
adverse selection and continuity of such. I.E. when the mkt is trending, longer
time frames are required and higher inventory levels will be accrued than then
we're in a ranging mkt. So it makes sense to have short time frames in the order
of the length of the trading window we have in mind (i.e. 30 seconds? 5 minutes etc)
and then also keep some higher time frame windows to general sense (i.e. hour,
day, week).
The 1-2-3Days frame seems very useful to enter in the mindstate of a longer term
swings a typical mm will work on (on the DAX at least).
I see a lot of analogy to the MTF moving average panels used often in retail except
here you'd apply it to the inventory levels.
I agree, not easy-peasy stuff but there's no easy stuff left in the mkt to profit from.
It's mostly tedious work and not exactly rocket science (if I may
I'll know who to ask in case I need NT as clearly you've done
a lot of work in NT.
To me, these are only tools. The real issue is finding ideas that work.
I am using python for research purposes as there is where the research
AI community lives and where most of the progress and breakthrough
happened in the last 10 years. It used to be R before and Matlab before that.
So, for example it's very easy to construct multiple timeframe "integrators" and check
their predicting power using python, some saved data ad not a lot of code.
Efficiency is not what I'd be after in research but simplicity and quick prototyping.
If the signal is there and proven to work in a quantistic way, then I'd have no issue
recoding it in C++ or whatever so that it's also fast.
BTW: regaring the rhytmic feed: How are the timestamps in the feed? Are they market
timestamps (i.e. true to market) even if they're delivered with some latency or are they
fake, i.e. the whatever time of arrival at the feed provider or NT ? That would be really
important for the type of analysis you're doing. For starting you'd be able to measure
your delay from the mkt. Then you'd be able to reconstruct the true order of events...
as opposed to some other fake arrival order.
Well, you cover a lot of ground there. I have just a few comments.
This "Simple Integrator" approach is not suitable for what I call "real or at least
greatly improved Deeper Inventory Analysis". That's when Profit-Matching takes
place, separating the Inventory into Closed (Profitable) transactions, leaving
Open (yet to be Profitable) positions on the part of Market Maker.
As for dealing with the "Fractal" nature of Market Maker's hypothetical timeframes,
which we hypothesize as being multiple..... My work is on the relatively short term,
e.g. max an hour or two; certainly not over many hours or days, as that would
require some realtime database which held all Time and Sales (I guess) and then
was able to do evaluations over an arbitrary timeframe of choice... WHEW !!!
[edit] Sure, you might say that Time and Sales history is available, but I'm
talking about that information Accurately Categorized by Bid/Ask price
relationships, which we're using in real time, to attribute the data to MM Buy
or MM Sell. Without accuracy there, just a history of Time and Sales won't
do the trick, in my view anyway. I've discussed elsewhere the difficulty of
getting Replay or Backtesting type results with such an Indicator which, again,
would require Accurate Bid/Ask pricing at the moment of each Trade report.
I believe the Rithmic data is "timestamped at source" and those timestamps can
be used, as opposed to latencies of arrival; however, I don't use the Rithmic Raw
Feed, so I don't really know that, just strongly suspect it; and NT8's feeder should
use those originating timestamps but, again, I can't be sure... Not sure it is
a Distinction with much of a Difference to it... ??
Yes, Python is likely to be quick for prototyping. I find that just the Capture phase
of data, however, requires such detailed design and high performance, that I have
to use C# multiple class definitions, and multi-threading, just to achieve
Accurate Capture. Then, of course, we move into the Analysis phase, etc........
I chuckle a bit when somebody says "Measured the Inside Bias, and there's
Nothing There"... Says Who? And How did you attempt to "measure" it ????
Believe me, there's lots in that data, for those who know how to "dig" for the
"signal"... And the "quality" of your data source is a huge factor as well; since
brokerages will give the cheapest one, often aggregated, knowing that really
it's "good enough" for most retail traders, of course; I get that...
A casual "stab" at "measuring" something, just won't cut it for the kind of things
I'm trying to evaluate.... I feel like saying "Don't try this at home, Kids" but with
all due respect, unless you've really worked at it (for weeks or months), then
don't come to me saying "There's Nothing There" LOL Just sayin'.....
That's not meant as a slight or a "dis" to anybody who Really Sincerely is trying to
figure this stuff out; but just to underscore that "It Ain't Trivial", which is the point you
were making as well...
So I've been using this little TradeFlowRisk indicator
constantly; which is unusual for me; but I think it's very
useful in gaining "situational awareness"
[edit] the scale of the breakdown of support is
quite impressive...
In my case, it's the Nasdaq market with the eMini contract NQ
and also the Micro contract MNQ.
Should you assume the accumulation and distribution trade
flow will be the same, since it's the same underlying index ?
In short, NO. Or, likely not, anyway.
So an interesting idea, if you trade a symbol which has its
corresponding Micro contract, is to do TWO charts, one for
your Primary symbol (e.g. NQ), and the second chart your Secondary
instrument (e.g. MNQ). Use the same expiration dates.
Run a copy of TradeFlowRisk on each one, but... importantly...
Be sure you have TWO distinct parameter files, one for each; and
use a Retention Interval which is significant; like 15 minutes or more.
Then observe whether there is a degree of Correlation between the
two in their trading patterns.
NOTE: Since Retail players tend to BUY a Rising Price; and SELL a
falling price; that trivial observation should lead to correlated
patterns. But perhaps not always.
Later today I'll edit to post a couple of comparison charts on NQ versus
MNQ and we'll see what happens.
[edit] Here, the Indicator overlaps the Price chart; and uses the left
axis. This is MANUALLY scaled so that ZERO (Net Inventory zero)
is in the middle, and Inventory is also manually scaled so that it
looks reasonable. With today's HARD Nasdaq crash, it's not surprising
that Retail Sentiment was strongly negative early in the session,
with Net Inventory showing negative (Retail Selling).
Hi hyperscaler,
I've given the concept some thought and my conclusion is that one of the two instruments will follow the other, precisely the big contract will lead the small regardless of the trades. To an extreme, in DAX where there're 3 such contracts (multiplier of 25, 5 and 1), the small contract often does not trade in low-volume periods. While the other two contracts trade in an almost identical way.
My explanation is that all are "retail" to the market maker and the volume that comes from institutionals that appears as the MM trades is on the other side and well hidden in plain sight
So, I'm actually using the total volume from the 3 contracts as a single volume multiplied by their multiplier (i.e. my risk analysis uses 3 contracts and comulative weighted volume). Seems to be more accurate as one technique to hide a big order is to split it over 2 contracts and execute in paralles wherever the volume lies.
BTW: How is the indicator working today in a one directional mkt?
Well, that's an interesting approach and I've considered "merging" data from the
eMini and Micro contracts as well.
Clearly today Nasdaq "Retail Sentiment" or expectation was a crash, since the Open from overnight trading was down so much !!! You see that clearly, but that ain't really
a big revelation.
Naturally, I have to say, as always, that External Factors, such as FEAR at the moment,
will certainly Overwhelm any possibility of predicting from Technicals... That's just true
for nearly all "sensitive" technicals; they are easily overwhelmed by these External
Fear factors...
IN NORMAL CONDITIONS, the definition of "which contract (eMini vs Micro) is Driving
the market" would revolve around which one dips into RISK, and then causes a
Trend Reversal, or something like that.
I do not agree that "Retail Sellers being 'in control'" pushes a market down; rather,
my view is Market Maker does that. But in an Index driven by Stocks/Equities, like
the Nasdaq 100, a general Selling Sentiment being "in control" may drive Prices down
until Accumulation makes it worthwhile for Market Maker to Forcibly raise Price,
to sell off excess Long Inventory, even though the concepts of Short and Long
don't make much sense, but historically it is clear to traders what we mean here.
MY CURRENT WORK is measuring "The Book" or "Depth of Market" "pressures" which
are associated with Trending. Nano, Micro, Short Term trending; with a Goal of
being able to do "Trend Following" based not simply on Price, but on the "Book Pressure".
At timescales which interest me, I think I've successfully done that; although it is
very difficult to do....
Having said that, a feeling for Net Inventory which this Indicator provides, is helpful
to some extent in making Trading decisions -- I hope
[EDIT] WITH REGARD TO THE ATTACHMENT, this is my custom charting showing
more detail than NinjaTrader charts can show, where the PURPLE line means
"Long Inventory" or excess Retail Selling over 30 SECONDS negative; and the RED
DOT is a Big Lot single SELL trade (too late!) in NQ greater than or equal to 8 lots.
And, remarkably; this combination of conditions is precisely where the Buy should
Trigger; with a nice Rise in Price... This is just using SIMPLE Inventory Analysis,
but just ignore all the other "stuff" LOL
I'm familiar with the concept. You're on the right track. Not sure if you're familiar with flying helicopters, but the idea is the same. You need to apply the right force (pressure) to counteract whatever the dynamics are at that point so as the chopper (mkt) goes in the direction you want.
Unfortunately, with choppers the physics are finite and well defined (i.e. deterministic). With the mkt, well, not as much so
I said: "MY CURRENT WORK is measuring "The Book" or "Depth of Market" "pressures" which
are associated with Trending. Nano, Micro, Short Term trending; with a Goal of
being able to do "Trend Following" based not simply on Price, but on the "Book Pressure"."
My reasoning here is a bit more subtle; and it has to do with Market Maker's
"pushing a market" and "overpowering" any contrary Retail trades.
Consider we push the Market Up; we want to maintain larger quote size at the BID,
in case ANY Retail player want to Sell to us. After all, we will BUY whatever it is; and
simply "muscle" the Price higher, thus performing the "magic" of making what we
just Bought, have more value, since we pushed the Price higher.
This is how Market Maker acquires futures contracts at NO RISK, since what is Bought,
(or Sold) in a "push move" to the trailing Price levels, is simply "up valued" by the
nature of the move.
Probably, that doesn't make much sense, because it has to do with thinking for a while
about "risk" versus "no risk at all" purchases by Market Maker.
So, in general, there is a "wave" of Size which PUSHES against the Price as the Price
is moved upwards; just in case Retail players want to give us (us Market Makers) contracts "for free" which we'll just push upwards in value instantaneously !!!
(Correspondingly, there is a withdrawal of size in the Price direction, since MM
does not yet want to transact on that side of the market nearly as much.)
"It's good to be the King!" which in this case means "It's good to be the Market Maker!"
LOL
