This page has been archived and commenting is disabled.
Guest Post: Deconstructing Algos, Part 1
Submitted by The World Complex
Deconstructing algos, part 1
The third part of the series on information theoretic
methods of analysis for dynamic systems is taking longer than
anticipated. Crunching the numbers is killing me. So I'll take a break
from it and look a little farther forward--how we can use the methods I
have been describing so far to forensically examine the algorithms used
in various high-frequency trading events of the recent past.
As seen on Nanex and Zero Hedge, there has recently been a lot of strange, algorithmically driven behaviour in the pricing of natural gas and individual stock prices on very short time frames. In an earlier article I pointed out that the apparent simple chaos we observe in the natural gas price appeared to be an emergent property of at least two duelling algorithms.
In this series of articles we will begin analysis of the algorithms
involved. Today's discussion will mostly focus on framing the issues
that must be addressed in order to study unknown algorithms on the basis
of their time-varying outputs. Future articles will present results
from the various analyses.
We begin by looking at the activity in the natural gas price on June 8, 2011:
Let us also consider the pricing action in CNTY on June 21, 2011:
In
both of these examples (many more such examples exist) there are three
time series of interest to us--the bid price, the ask price, and the
prices of trades. Additional information which may also be of use are
such things as volume, size of bids, size of asks, and so on. In
principal both the bid and ask prices form continuous series which are
prone to instantaneous changes. The actual trades form a discontinuous
time series with obsrevations at irregular intervals.
both of these examples (many more such examples exist) there are three
time series of interest to us--the bid price, the ask price, and the
prices of trades. Additional information which may also be of use are
such things as volume, size of bids, size of asks, and so on. In
principal both the bid and ask prices form continuous series which are
prone to instantaneous changes. The actual trades form a discontinuous
time series with obsrevations at irregular intervals.
We
don't have access to the code involved in these
algorithms--nevertheless, we can learn something about the computational
processes involved, within certain limitations. Unfortunately, just as
is the case in studying time series recorded in rocks, we have to make
some assumptions, and the validity of our assumptions goes a long way
towards predicting the success of our endeavours.
don't have access to the code involved in these
algorithms--nevertheless, we can learn something about the computational
processes involved, within certain limitations. Unfortunately, just as
is the case in studying time series recorded in rocks, we have to make
some assumptions, and the validity of our assumptions goes a long way
towards predicting the success of our endeavours.
Our first assumption is that the bid price and the ask price are
being set by competing interests. This assumption is extremely
important. It is possible that the bid and the ask are both being set by
a single entity, or by two closely related entities who are using them
to manipulate the natural gas price. We will go though in some detail
the reasoning behind our assumption that there are competing interests
involved below.
Secondly, we are approaching this problem assuming that prices are set
and changed discontinuously in time rather than continuously in time.
Subtleties of this assumption are discussed in the introduction of Bosi and Ragot (2010).
The methodologies we will explore are as follows:
Cross-correlation of the bid and ask series over selected windows. We
choose limited time intervals rather than the entire record because we
expect that each series will sometimes lead and sometimes follow. Peaks
here will show whether one of the series leads or trails the other
consistently or whether each one leads intermittently, which would
support the idea that these are distinct dueling algorithms. It seems
likely that the bid price will lead as both are declining, and the ask
will lead as both are climbing. We should test this hypothesis.
One goal of this analysis will be to see if we can detect trigger
points, where one stops following and begins leading. We will locate the
times and see if the trigger can be identified, which is only likely if
the trigger is some change in either price series, the price of a
trade, the volume of a trade. Unfortunately, many other triggers are
possible, and it may not be possible to identify them if they are, for
instance, a random number generator seeded by, say, the
thousandths-of-a-second digit at the instant of some distant event like
the first pitch of a Yankee's game or when the secretary in the front
office misspells 'the'.
Phase space reconstruction--the relevant time series (bid prices, ask
prices, trade prices) each represent one-dimensional data sets. If the
algorithms used can be visualized in higher-dimensional phase space, we
may be able to reconstruct the overall architecture.
The advantage of this approach is that in principle the dynamics of the
system will be contained no matter which output of the model we use. We
only have measurements of the bid price, but have no idea what other
outputs are generated by the same algorithm, even if these unknown
outputs are critical to the decision-making module of the algo. The
reconstructed phase space
The difficulties here are that 1) the function may change from leader to
follower so quickly that the resulting trajectory through phase space
is too short to interpret; 2) there may be multiple players on both the
bid and ask, meaning the reconstructed trajectory through phase space is
an amalgamation of two or more different functions, the instant of
joining of which may be impossible to determine; and 3) it may prove
impossible to properly define windows for the data, again creating an
amalgamation in phases space of two or more different functions.
Epsilon machine reconstruction--We will need to try to identify the
actual "work" done by these programs. How do they decide on a price? How
do they "decide" to drop or raise their offer? Do they change? How are
we to recognize when an algorithm changes its behaviour when all we have
to deal with is the output? Can we recognize when the structure of the
computation involved in the decision-making part of the algorithm
changes, given our extremely limited knowledge of that structure?
These questions may be addressed using the ε-machine reconstruction approach suggested by Crutchfield (1994).
The objective of this approach is to use an open-ended modeling scheme
to describe the computational structure objectively, so that different
practitioners working on the same data will come up with similar
(hopefully identical) constructs. By encouraging an heirarchical
architecture of undefined complexity, the method allows investigators to
identify changes in behaviour of the the system.
This particular approach is built around discrete computation, so is
amenable to data which are discrete rather than continuous in time. We
assume that the discrete outputs (the time series, or stream of values)
is the result of a computational process which is knowable. The data
have to be organized, and (this is the key) repeated states are identified. It is possible that these states will be identified from the reconstructed phase space portraits above;
alternatively they may be be defined by particular observations. These
states may be identified as key strings of data, or may be recognized in
complex functions by reconstructing the state space in a higher
dimension. The ordering of the states is significant, as the state that
appears first before another particular state is referred to as the predictive state, and the following state is the successor state.
The ε-machine is constructed by
identifying all the predictive and successor states and calculating the
probabilities of all of their observed relationships. If more than one ε-machine is inferred, the sequence of these first-order ε-machines can be used to build a higher-order ε-machine. Given sufficient data, you may construct ε-machines of arbitrary order.
Information theory--as seen in recent articles, information theory may be used to characterize the complexity of the ε-machine reconstruction and the probability density.
The yet-to-be completed third part of that series concerns methods of
using information theory to find the optimum window length for creating a
probability density plot of the reconstructed phase space. The
subsequent parts of this series will concern itself with the analyses
described above on the nat gas and CNTY algos, as well as others as they
are found.
Given the limitations of time and computing resources, I can't guarantee
a timeline. I regret that my speed of analysis is six or seven orders
of magnitude slower than the incidents in real time.
»
- 10015 reads
- Printer-friendly version
- Send to friend
- advertisements -
There's an assclown on a network channel who continues to pump natural gas. He only appears on TV because he reads from a teleprompter. All advice and disclaimers are disclosed at the ending of each episode. Fancy that.
This is beautifully written - having dyslexia I olny wsih I cuould wirte these good (or is it well)?
Well done!
Books
i only made it half way through. i'm old. this shit makes my brain bleed. and i'm really not that old. THIS is a big part of our problem. JIT! JIT! Faster! Faster Pussycat! GO! not by human hands.
yep brain blood on the street here too.
Just ran into good ol' Bill Still.
He predicts a major economic downturn this year.
Still claims to have met with a congressman in Washington about sound money this week.
Still on the next crisis, "Austrian Economists will beat down the doors of congress for Gold money".
Bill Still Video Report Full
out
I think the essence of the article is that the author doesn't really know what is taking place, but they might... soonish... if someone gives them the source code. If I wrote an algorithm that randomly selected statistical terms and assembled them in such a manner as to create complete, albeit, incoherent sentences, it would look much like this article.
HA! sleep tight and, thank you.
This sums it up...
http://www.youtube.com/watch?v=fJqWHDuOpc4
having sight is a curse not a blessing.. no one understands the constant pain of watching stupidity crush the World in Total.. do you know why people with sight stay stoned or like to be drunk / stoned? because its the only way to shut off the noise, its the only way to drown out the fucking non-stop, never fucking-ending stupid!
or thats my opinion.
Who the fuck junked you?
The algorithms aren't the problem. It's handling the data that takes the time.
Guns don't kill people...
Sure you can...
on Fri, 06/24/2011 - 22:23
#1400397
I think the essence of the article is that the author doesn't really know what is taking place, but they might... soonish... if someone gives them the source code.
*************************************************************
What the FUCK! does the source code have to do with anything?
"C ++,+" is the source code..
The Problem the Member of "Fight Club" is having is understanding the triggers.. He can not or more likely "Does Not! want to grasp the tug of war".
http://www.youtube.com/watch?v=qz8i1AL3uls
At Least Two! (2)!!
So, if the writer were to look at a single transaction (impossible) he would see a buy order and then he would see that stock passed around (sub penny) between multiple parties (he knows the parties, interesting to see the big names squeezed as the Old Brand Names re-assert themselves based on would be retail flow) before being sold.. this works as well for institutional parties in larger block movement..
1. Lagged out of the money.
2. This one he will hate.. Time Keeping? What time was it really? says who? the house? the houses? "O" that's much better!
3. back to it takes at least two.. or more to have a sub penny tug-o-war.
People who can program are fantastic, give me 5 programmer's who are super secret (not wanting to share info before "D" day..) and I can structure something they don't understand.. and that's what you have here.. one piece of a puzzle.. a sky blue piece.. (which works in a LOT! of other puzzles) trying to decide what the landscape will be. He has information as a Piece.. but not an over all because he is not structuring a game plan to implement. and for the record the math freaks are all about being accurate to provide a result (because that's their job) but that is not a plan of action in itself.
as for you bullshit response to someone who is a Member of Fight Club and who is very much sharing more in this one article that you could in total if you never stopped posting starting today!
Basically shut the fuck up, sheep. and let the grown up speak with verbally vomiting all over his efforts.
Kudos to our poster for his efforts! I hope they pay him! before he starts sharing to, too much!! LULZ!!!
Agree.
Modern authors use the terms and methods in an absolutely irresposible manner.
At first I thought to deconstruct the article, just to show the deepest discrepancies between the definitions of the terms and their use, but ... why should I, if Mr. Mark Twain from Connecticut already did it:
"... He put the paper on his lap, and while he polished his spectacles with his handkerchief he said, "Are you the new editor?"
I said I was.
"Have you ever edited an agricultural paper before?"
"No," I said; "this is my first attempt."
"Very likely. Have you had any experience in agriculture practically?"
"No; I believe I have not."
"Some instinct told me so," said the old gentleman, putting on his spectacles, and looking over them at me with asperity, while he folded his paper into a convenient shape. "I wish to read you what must have made me have that instinct. It was this editorial. Listen, and see if it was you that wrote it:
"'Turnips should never be pulled, it injures them. It is much better to send a boy up and let him shake the tree.'
"Now, what do you think of that? for I really suppose you wrote it?"
"Think of it? Why, I think it is good. I think it is sense. I have no doubt that every year millions and millions of bushels of turnips are spoiled in this township alone by being pulled in a half-ripe condition, when, if they had sent a boy up to shake the tree--"
"Shake your grandmother! Turnips don't grow on trees!"
"Oh, they don't, don't they? Well, who said they did? The language was intended to be figurative, wholly figurative. Anybody that knows anything will know that I meant that the boy should shake the vine."
http://www.readbookonline.net/readOnLine/1524/
"The How I Edited An Agricultural Paper".
This was my long response. The short one is:
"Shake your grandmother! Turnips don't grow on trees!"
Yeah, same here - tl;dr.
Once you've seen Lotka-Volterra, you've seen it all.
What we have is not so much an atto-fox problem as it is an atto-bunny problem.
sharing this with my ZH brethren -- stock market is ready to implode! Based on my Elliott Wave setup.
http://jeffreygtc.blogspot.com/2011/06/friday-june-24th.html
thank You for sharing your work with the idea of nothing more than helping others.. God Bless You and Yours! or if God is not your thing my best to you and yours.. with a Hearty Thank You Kind Sir for thinking selflessly of others!
wheres ken griffin when you need him
Just trying to generate volatility. Marketing 101 people
If "algos" really disrupt the market, it is new to me. I've been watching the market intently for 15 years and haven't noticed any change. Maybe it has happened too slowly. But, the same things seem to work to make money over time. Those are the classic rules.
1. Trade with the trend. Whatever time frame you are in...
2. Buy late and sell early..
3. Practice good money management. ie, don't add to losers, keep your losses small, and let your profits run. (as long as you can stand it!)
4. Don't try to pick bottoms. (you'll only get shit on your fingers)
If you can do these things consistently, you are guaranteed to make money. In any market, over any long period of time.....
Algos?? What the blank are they???
gh
No.
The 2010 flash crash was caused by a runaway or deliberately programmed algo.
NOT Waddell & Reed! SEC knows!
Watch this, some of it is in the dutch language, but all interviews in english. worth sitting through...
http://www.youtube.com/watch?v=x8SMjyXdc3I
Excellent documentary.
Mostly in English.
Nanex principal Scott Hunsader is interviewed.
Ask a buy side fund manager if they think the increasing use of a technological arms race to pick up pennies is a good idea ... and you will not get too many takers. Algos and HFT do not create liquidity or depth, they just create more 'volume'.
Having said that, I don't disagree with your maxims above per se. They are basic trend following/momentum thoughts, which work in a market which is not rigged. Good luck...
Fascinating stuff...
Is there a correlation to the concentration of the trade volume done algorithmically with the efficacy of the phase-space reconstruction? (Where does the 'noise' from non-algo-driven trades show up? In the price of trades series?)
That's a lot of words to say you don't know anything.
Was this post generated by an algo?
Yes, this one:
// From volume 2 Knuth
// rand() = 1.83 seconds Mersenne = 1.01 seconds GameRand = 0.24 seconds
//...........................................................................
// Mersenne numbers
//p0 = 2147483647, p1 = 2147483629, p2 = 2147483587, p3 = 2147483579,
//p4 = 2147483563, p5 = 2147483549, p6 = 2147483543, p7 = 2147483497,
//p8 = 2147483489, p9 = 2147483477, p10 = 2147483423, p11 = 2147483399,
//p12 = 2147483353, p13 = 2147483323, p14 = 2147483269, p15 = 2147483249,
//p16 = 2147483237, p17 = 2147483179, p18 = 2147483171, p19 = 2147483137,
//p20 = 2147483123, p21 = 2147483077, p22 = 2147483069, p23 = 2147483059,
//p24 = 2147483053, p25 = 2147483033, p26 = 2147483029, p27 = 2147482951,
#define MM 2147483647 // p0, Euclid number
#define AA 48271 // full-period multiplier, spectral criterium
#define QQ 44488 // (long) (MM/AA)
#define RR 3399 // (long) MM % AA, important RR < QQ
static long int X = 153;
static long int Y = 8479237;
//............................../ RNG_Mersenne \........................
long int RNG_Mersenne (void)
{ long int Z;
//.......................................................
//
// 1 key making LCG multiplatform fixing
// overflow problem multiplying two 32 bit is
// Schrage solved by the doing approximate factorization of m such that
// m = aq + r , q = (int)(m/a), r = m mod a
// xi = a(xi-1 mod q) - r*(int)(xi-1/q) if xi-1 >= 0
// xi = a(xi-1 mod q) - r*(int)(xi-1/q) + m otherwise
// m=16807 becomes q = 127773 and r = 2836
//........................................................
//
X = AA * (X % QQ) - RR * (long int) (X / QQ); // Schrage's approximate factorization
if (X < 0)
X = X + MM;
//......................................................................
//
// #define MMM 2147483399 // p11,
#define MMM 2147483629 // prime p1
#define AAA 40692 //
#define QQQ 52774 // (long) (MMM/AAA)
#define RRR 3791 // (long) MMM % AAA, important RRR < QQQ
Y = AAA * (Y % QQQ) - RRR * (long) (Y / QQQ);
if (Y < 0)
Y = Y + MMM;
Z = X - Y;
if (Z < 0)
Z = Z + MMM;
return (Z);
}
//............................./ DRNG_Mersenne \.................................
double DRNG_Mersenne (void)
{ return ( (double) RNG_Mersenne () * 4.656612875245796924105750827168e-10); // or / 2147483647.0
// return ( (double) RNG_Mersenne () / 2 147 483 647.0);
}
//________/End of this file \____________________/
Very good article. Thank you for your intuitive work. Looks to me like they find a low volume , illiquid time to hit the market and make a quick buk and stop. Probably the best strategy is to get out of their way and after they are done play it back to the accepted price (previous price before the manipulation and make a few buks). No sense in crying about it , figure out how to use this info to profit.
I'd agree with you but the rollback in trades above or below a threshold price may make this a losing strategy, particularly with the arbitrary nature of the cut-off.
6/22 just after midnight CST - 1 tick and 10 tick chart of comex silver. A similar natgas sinusodial pattern.
Im inclined to think that the gas pattern was just a fluke. The series used in the pattern is fairly simple and its not outside the realms of possibility that it might just be the product of a particular random stack of orders and bids on a particular day being knocked out sequentially. When you take into account that orders are usually only price dependant, couple it with the fact that there would have been thousands of other market participants at that moment, and the pattern looks like it could be natural.
(stack of offers)
4 nth orders
3 3rd orders
2 2nd orders
1 1st orders
T t1 t2 t3 tn
1 1st bids
2 2nd bids
3 3rd bids
4 nth bids
0 bids…
(stack of bids)
If you look at the nanex plot, you will see as each new phase increases and becomes larger than the previous, that’s when the offers and bids become thicker(the white dots)…because they were there to begin with, the market s doing what it always does and gravitates towards them.
But the main reason I don’t think it was intentional is because I cant work out how one entity could make any money by doing this. It doesn’t make any practical sense - they could have wiped out all the offers in one phase or wiped out the bids in another without the need for a prolonged whipsaw…how would they make money?
It's an unwarranted assumption to think that any single event "in itself" has to be motivated by a profitable outcome.
If you were to spend $50,000 today to learn information that would make you $500,000 next week, it's a no-brainer.
The prisoner's dilemma. The actual problems don't change much.
Yeah but you are jumping to conclusions there, conversations like that always converge to wild swings in the dark. For a start, the market is different everyday, so the problems do change, a lot - the market could have went bidless for no reason at all...so the $50000 spent to learn about something one day could have lost them $500000 on another, "its a no brainer".
I'm the one who's NOT jumping to conclusions. I think anyone who's trying to outwit the market is crazy--it's the search for the Holy Grail. It's not that you can't find the Holy Grail because you're not good enough or the data analysis was insufficient--it's because it DOESN'T EXIST.
So here's a guy telling us how he's going to backward-model an event. For what purpose?
To outwit the market.
Game over. Good luck, dude. I'm still working on squaring the circle. Everyone needs hobbies.
Sorry man, i misunderstood you there. I agree, its trying to reverse engineer randomness, he will end up with his head in the gas oven, never mind the gas market. I feel for people that waste their time and talents buy getting lock-jawed in trying to subordinate the market. I really do, because if they had of got their teeth into some solvable problems they might have done something useful by now.
To me, the above pattern just looks like the product of a market feedback loop generated by a fluke sequence of orders and market psychology, that had to break down at some point. Randomness has a habit of looking like it has a purpose sometimes, pity the fool that puts his youth into working it out…though that’s not to say he wont learn something along the way.
'just looks like the product of a market feedback loop generated by a fluke sequence of orders and market psychology'
Geeezzzzzzz, Wake Up. 90 days of perfect trading at GS, BAC, C and JPM is not a fluke. Who the hell do you think is doing this shit?
GS have hundreds of traders, they dont all win everyday, some of them have losing days and will be subject to the well-oiled constraints of a mid-office when they have. If you looked at the data you would see there are plenty of losers, it just so happened that in the span of 90 days the collective experience of all traders achieved an aggregate P&L that was positive...is it really that unbelievable that all those traders went 90 days in the mother of all ramp-up's without the majority being wrong? I dont think it is…i suppose you are just willing to believe anything you read.
Several things about the big banks and their "perfect" records:
1) They could be lying. It's been demonstrated they're prepared to lie about anything.
2) They have access to such enormous sums of fiat that they can engineer market behavior via brute force.
3) If they are handling a sufficiently great number of transactions for others, they could remain 100% neutral on every single equity, commodity, and bond, and STILL report a successful trading day based solely on arbitrage.
If there's some little guy who thinks he figured out how to be on the right side of Goldman's trades for more than a few days, he's most likely mistaken. Everyone gets lucky here and there, tho, and with a sufficient number of trials, you can eventually find any arbritrarily long string of successes.
Actually there perfect trading record is because of how the define trading. All they do is match buyer and seller and take their half a penny cut. Ever wonder why you can only enter orders in one penny increments but when your trade gets hit it might be at 89.958666665. Its because the houses are internalizing the order and stepping in between buyer and seller to make a sub penny, risk free, profit. Is not really trading its just stepping in between trades to steal money.
Maybe he's trying to reverse engineer to establish proof of manipulation so he can send the information to the proper authorities and be ignored?
How about de-constructing Chinas ghost banking system First?
Just use frame theory, dude.
http://www.math.nthu.edu.tw/~tjm/abstract/0006/tjm0006_1.pdf
Go for it JM, show him how its done...errr.
Noise makes attractor reconstruction tough to say the least. Frames, or some type of wavelet analysis offers the best shot.
But what will be the final conclusion dude - all he will be able to say (albeit in much more confusing language) is that at time t, the bid or offer was here or there...no. (lol, no)
Sir to you! The final conclusion is based on perpheries. Italy comes to mind. All 2 TRILLION of their ASS CLOWN) Holiday surplus! Any thoughts?
No yen, still trying to figure out greece dude(and wondering if I want to anymore). Italy is just another scene in the contagion act - and if contagion can be classed as fear then a 213bps jump says Italy has just contracted HIV. The whole crisis needs real leadership in my opinion if the euro is to stand a chance of surviving…but that’s grown-up shit, I’ll leave it to the adults and have come to accept that this crisis is well above my pay grade. I just stick to technical’s and headlines, that way I can eat and not disappear up my own asshole.
Static? Please respond over N.A.T.O. channels.? Garbled?
Nice response in some redirected way! +1(sarc off
My intial reaction is to say that maybe he is on to something here, and can extract something deterministic out of some series. Beyond this training period, he can make money. If he does, others will eventually catch on, the market will absorb the information content of what he is doing, and he'll have to start over.
However, Olsen and Associates (think they were the first?) has been around for over twenty years doing stuff like this. Either they are able to make money doing it, others with scale can make money using their advice, or there is a reputational "shock and awe" effect going on that keeps people pouring money down the drain.
He is defintely on to something but its not an easy task. Even if he is able to reconstruct some form of the underlying algo used in this one instance there is nothing to say that the algo doesn't change on a longer timeframes or is contantly dynamic in form.
I opened a sideways thread! Man that latest update helped.
I do my best to read everything worth reading. I need a day on this one.
Ad`obe is tricki!
You all are makeing a moutian out of a mole hill! Its a simple algo that dates back to the start of the ponzi mar... er uh stock market, its the Chewbacca algo structure! Very simple when x = x it does not equal x and attempts to fuck you in the ass! Oh and when you have access to who is buying and who is selling PLUS more money than god, you can make a few things happen, simple!
There's a pretty far out factor to consider relating to speed. When I used to work in neuro-science research I came across a very hard-to-comprehend but highly reproducible phenomenon, time is fuzzy at the nano-second level these algos are working at. Imagine a test subject with probes measuring the firing of some individual neurons. Show the subjects one of three types of photos and measure the neurons firing in response to the photos. The first type of photo is pleasant (fuzzy kitten, rainbow or such). The second type is startling (snake or tarantula for example) the third type is arousing (girl in bikini or the like). Different neurons respond to the different types of photos. The really bizarre thing is that the correct neurons will start to fire BEFORE the photo is shown, even when neither the subject and researcher know what type of photo will be shown. This effect is at the nano-second level, it's real, it incomprehesible, and it's so common I've seen it occur in several neural training research presentations where no one even bothers to comment on it. The best explanation I've heard suggests that time behaves similar to a quantum wave so there is a fuzziness that allows leakage of the future into the past. If you want to boggle your mind let Mr. Google lead you down this rabbit hole.
So if these algos are the self training types that change their own programming to improve their responses, could they be creating a feedback response because they are detecting and reacting to their own bid before they made the bid? If so Goldman or whoever runs the program will be just as confused about it as anyone else. It not likely but you may be able to check if the response is somehow connected to itself. I don't pretend to fully understand your computations but it occured to me that a responder reacting to itself may have detectable properties.
And in response to the question running through many of your minds, no I'm not smoking anything. Check it out on Google, this is a strange world we live in.
Jesus, i have heard of people giving Goldman Sachs way too much credit, but this takes the piss. Now they are frontrunning reality itself? phew. Your last part was right...it really is a strange world we live in, mainly due to the lunatics that inhabit it.
It's been known (although perhaps not understood) for at least 70 years that tomorrow affects today. This really shouldn't be that big a surprise.
GS pigmen=TimeBandits
hahahahaha
They're stealing that too?
LISTEN up close people. OUR IT departments are being HACKED every hour of every day.
I got smacked by some tool from 00z0yeyzoo something lasat night! Be careful, and clear your threads. Yen
Including Microsoft Office 2010 and the diagnostic Microsoft Office 2007 instrument is designed to Office 2010 detect and Office 2007 fix the problem, if there is chaos. Conclusion if Microsoft outlook 2010 you upgrade to Microsoft Office 2007? It office 2010 professional depends on how you work. If you are a Microsoft Office fashion, and hoped Office 2010 Microsoft that the new file Outlook 2010 playback templates, then Office 2007 should MS office 2007 be you. Outlook outshines its predecessor microsoft office 2010 trial, if you need it every day download Office 2007 Lean management Microsoft outlook meetings and office 2010 trial tasks. Also, if you have little-used Microsoft Office 2010 download functions of Office 2003 or earlier Office 2007 key, but also get along well Office 2010 key, then there is no need office 2010 home to spend hundreds of pounds Office 2007 download of new software.
I need to lay off the "Clark Gable" approach.
I suppose everyone read those charts? Can you say Wedge? Like the ones I gave to my now BIGGER younger brother!
Don't over analyze.
Look to who started and faded the trades.
Look to the big picture.
Who won? Who lost?
Are trading desks/exchanges battling for millisecond supremeacy or working in cohoots?
In macro moves are micro decisions, and vice-versa.
Start with outliers and work inwards.
Why do you always put people down? Matter of factly! I have a lot of respect for you.
YEN 9 dot9 (DOT) periiod?
June 16th, 2011 - Effort led by 12 investment banks promotes increased FIX use in the wake of Dodd-Frank reforms full story
HFN opens two new NJ Metro routes
June 14th, 2011 - Hudson Fiber Network opens fast paths to Secaucus full story
Good day, gentlemen. This is a pre-recorded briefing made prior to your departure and which, for security reasons of the highest importance, has been known on board during the mission only by your H-A-L 9000 computer. Now that you are in Jupiter space and the entire crew is revived, it can be told to you. Eighteen months ago, the first evidence of intelligent life off the Earth was discovered. It was buried forty feet below the lunar surface, near the crater Tycho. Except for a single, very powerful radio emission aimed at Jupiter, the four million-year-old black monolith has remained completely inert, its origin and purpose still a total mystery.
vive l'tard! (lurker)
dup
I have been following this series of articles and find it very interesting but I do not get the Nat gas algo. Would someone mind explaining it to me in simple terms? My understanding after reading that article is that the algo keeps lowering its bid and raising its offer little by little and so we think it is 2 algos 1 upping each other? But how could this cause a market disruption? It is not like they are raising their bids and lowering their offers.
Additional factors are:
-the role of the market makers
-the float (unless naked bids are somehow in play)
Currency & Power
Relative to the universe, Caesar is no bigger than you, as an individual, and Caesar is contained within a multiplexing filter of filters. You know, misery loves company. Monetization goes vertical when the virus goes vertical, when there is no resistance to the resistance to change. When leverage becomes inelastic the threshold is reached, and time to capitulation is easily calculated, triggering spring release, because all of the derivative variables are identified, once relativity is filtered out. The economy demonstrates that the difference between fusion and fission is relativity, which may be controlled by a quantum controller, IF YOU PROVIDE A SUITABLE PATHWAY FOR BACKLASH. Physics is physics is physics.
Physics controls chemical reactivity, which controls biology, which is why they say you cannot productively fight biology, but don’t bother telling Caesar’s minions, who are determined to control breeding. Who is breeding, who is not breeding, and why? Government has short-circuited the relationship between individual and planet, replacing real demand with artificial borders, to the end of monetization in a positive feedback cycle, hiding the losses in an accounting sink, mirrored as profits, with the expected derivatives.
Quality begins with unique individuals sharing a common virtual communication hub. Remember the tile game; if you satisfy a market, add a dimension and branch out. Incorporate diversity through purely municipal interest law, which will trump federal law if you eliminate costs in the process. Write off individual officer living expenses, pay up to the non-taxable limit, and provide intangible benefits. The faster Caesar prints money, issues credit, and increases margin volatility, the easier it is to eliminate the need for currency, homes, autos, and government make-work jobs, which will all be in such great surplus that their value tends to 0. Ultimately, no nation/state can defend a ponzi property economy without community participation in the destruction of individual identity, a character/dimension of which is a purely municipal interest. If you give Caesar access to your intranet, you have no one to blame but yourself.
Practical developers operate in the first two integral circles beyond Caesar’s derivative multiplexer, in dimensions beyond the looking glass. The intelligent kids/aliens will be at the looking glass on the derivative side, where the dinosaurs, over the centuries, have, by evolution, come to congregate. Provide them with the tools necessary to maintain their unique identity, beyond Caesar’s knowledge, in a community where privilege is a function of productive participation in problem solving, not currency, and your problems will be solved, productively. You don’t have to give Caesar your kids’ test scores.
If Caesar puts up a mirror, you put up a mirror. For every relative action, there is a relative reaction. Maintain Caesar’s gravity accordingly, depending on the power required for your slingshot. Whether an event is a black swan or a white swan, and its power, is relative to your position, which is relative to the looking glass. On the automatic side, simple symptoms are addressed first, by increasingly greedy operators, which is why the number of random events increases and become more incidental at an intersection, which you may reliably anticipate, with adaptive tools, over time. Property is a derivative of derivatives; the black hole is mesmerizing.
Move immediately from internal competition to cooperation. Caesar knows only what you tell him, how you self-identify. The point of credit is identification, in a positive feedback loop with make-work jobs. Adjust it to meet power, which is all about leverage and is a function of communication. The looking glass is the pivot. To get current, you need circulation. Set up your resistance to the resistors accordingly. The difference between voltage and current is flow, containment, relative to resistance. Adjust Caesar’s currency accordingly.
Did your old man ever come home at night and throw away your bike because it was in the driveway one too many times? That is where this economy is headed. Salvage what you can before the churches are bulldozed and the false prophets are thrown in the fire. No priest should be asking you for a marriage license. Marriage is an inalienable right. The reptiles and the sheep might want to check with the developers before they start f***ing with the kids again. Salvage what you can, and don’t look back.
Real power, in the one true circuit, does not corrupt. It is the illusionary parallels that self-corrupt into a black hole, by attempting to drain the main with no return, resulting in gamma. The difference between gas and diesel combustion is relativity. The end instruction in the evolutionary cycle is not mom and mama’s boys. Don’t bother calling the firefighters. Let nature take its course.
Caesar is a bully, nothing more than a mama’s boy with a gun and a badge. What you have is much, much more powerful, and you don’t need no stinking badge. All you need is an education system. It’s called responsible parenting, which Caesar knows nothing about, by design. Dimes and nickels can only hold up a multi-trillion dollar enterprise for so long. It’s about priorities, not proprieties. Decisions have consequences, especially omission. They can lie to themselves, but an honest man always gets to the truth first. In a world of corruption, he is easy to spot.
10 guys on the line, millions of dollars worth of equipment, running crap lumber, and all trained to blame each other for the outcome. The apple doesn’t fall far from the tree.
you might take a look at that quantum statistics tool again ...
You are one weird dude.
I bet he works in a mill.
...beautifully so
keep em comin kev
MAN legalizes MAN to UNITE with MAN to join the ONE BIG HAPPY PONZI.
meanwhile in 4 days time, in same state, the wells begin to become poisoned.
http://www.oilandgaseurasia.com/news/p/0/news/11792
FRACK IS WACK
"It is possible that these states will be identified from the reconstructed phase space portraits above"
The images shown are time series, not phase space. Phase space would require a Fourier or LaPlace transform resulting in a frequency map, or a k-space transform. The above time series would actually look normal in k-space if they were over a larger timeframe so one possible conclusion is that the algorithms take assumptions from the frequency domain and use deconvolutions to come up with bid/ask theses. However some information is lost in the process of going from frequency to time domain, particularly the order of the bid/ask increase/decrease. So they create crazy looking time series out of normal looking frequency histograms, partly because the assumptions taken from the histograms should be over a longer time period. Basically these guys are running the program too fast for them to work. These algos are failed attempts to make the tail wag the dog, and most likely they lost money.
The idea is to reconstruct a phase space from the time series, using methods that you might find in Packard et al. (1980).
http://astrophysics.fic.uni.lodz.pl/100yrs/pdf/06/132.pdf
I find too many problems using fourier analysis in the problems that I normally work with (which are admittedly not economic problems). Abarbanel's book describes how fourier analysis begins to run into problems when the systems are nonlinear.
I'm not certain that the programs are not working. We don't necessarily understand their objectives. Presumably they want to make money. But are they profiting by price, or have they set themselves up to profit from volatility, which they have created in spades?
I'd like to thank you all for the overwhelming response
I don't know where to post this so I'll try here, which is hopefully near the end. I find I have to reply to too many different comments to slot them in the appropriate place(s).
There was a post about me wanting the source code. It made me realize that I never said what it was I wanted to do.
What I am trying to do is put together a general approach to the forensic study of algorithmic trading. As in other forensic sciences, it would be nice to have a letter of confession from the perpetrator (read "source code")--but we cannot count on finding such a thing. What we can count on is the power of our own observations, our knowledge in applying some mathematical approaches, and our logical and/or intuitive approach to tying them together into what may evolve into a fruitful means of inquiry.
I do not represent that I know what I am doing. I do not know if what I am proposing will work. It is likely (perhaps most likely) that the methods of analysis proposed above will result in no useful conclusions, or worse--misleading conclusions.
I normally approach a scientific problem with a goal of testing a hypothesis. This problem is more like exploration--it is explored without clear hypotheses because I don't know what I may find. What would I have to see to convince me that the algorithms are manipulative rather than competitive? I don't know. I am merely hopeful that I will see something. In my experience exploration always teaches you something, even if it is only not to go there.
I am a geologist. I am accustomed to looking for gold and only finding poison ivy and black flies.
One poster commented that I would like to find evidence of manipulation so that I too could be ignored by the authorities. That seems to be close to the truth, and the outcome (if I did find such evidence) seems a little too likely to be funny. But even if I am to be ignored by them my thinking is that it is better to know than not to know. Through knowledge there may be a chance at some kind of action--perhaps by people like us, if not the authorities.
Many comments have been constructive, offering other lines of potential inquiry. I thank you all for them. Some comments have not been--but that tends to happen whenever there is an element of anonymity. I see the importance of not taking it personally.
For the purpose of your rehabilitation, I can tell You and other respectfull readers of ZeroHedge, that the "problems" described here were succesfully solved by 70% in 1900-mid-19xx years, and the solution is fully described in a very rear book on Geophysics. There are almost no copies in USA, there is one copy left in LOC, there are 5-6 copies in Europe's libraries. By some mystical purposes this rear book and methods are totally unknown to modern mathematicians and economists.
I dont have this book, but once have seen it. I almost never used it's formulas due to their extremal complexity for programming, ..... till today.
Troll.
As someone who likes to trade and make money, I'd like to know how algos affect the market in the aggregate. I don't care about sub penny scalping. If they want to do that all day it's fine with me.
I still see technical patterns in all the charts. That means markets are still being moved primarily by human greed and fear. If algos were really dominating, I would expect to see markets basically flat lining or moving randomly. After all, the algos should be emotionless and cancel each other out -- long side vs short side.
What's wrong with this logic?
Its your arrogance (for calling me "troll"). Its your arrogance badly affecting your trading strategy, approach and methods.
It is a common problem with americans, and american traders. Without an artifficial debt-based pump of money into stock markets they will fail to compete with computers' algo.
Dude, way ahead of you.
SciPy has all the tools you need to develop this. I implemented it in 6 lines Haskell last year and it would appear the strategy it can exploit is about to get arbed out. With so much of the action happening in dark pools, the returns from this kind of approach diminish rapidly. No open python or haskell libs or functions to deal with Nanex data as yet, so you have to do it from scratch.
Go ahead, do the work and see what you find. It's not as complex as all that. There are some other uses for it (hint: crypto). Cases in higher dimensional phase spaces become both computationally intensive, and sometimes surprisingly elegant. I can tell you that if you do take it to market, you will be able to say you are the smartest hobo on the railcar afterwards.
Would love to tell you how the rest of it is going, but why give it away?