[ 3 / biz / cgl / ck / diy / fa / ic / jp / lit / sci / vr / vt ] [ index / top / reports ] [ become a patron ] [ status ]
2023-11: Warosu is now out of extended maintenance.

/biz/ - Business & Finance


View post   

File: 34 KB, 497x280, if_eth0-day.png [View same] [iqdb] [saucenao] [google]
12426695 No.12426695 [Reply] [Original]

You can also do TA on server traffic-usage stats. There are resistance and support areas and trends. It somehow works!

What does that mean?

>> No.12426709

>>12426695
It means server usage is not a random walk like Brownian motion would be but usage patterns exist. What a surprise.

>> No.12426715
File: 33 KB, 497x280, if_eth0-day1.png [View same] [iqdb] [saucenao] [google]
12426715

I have more

>> No.12426718

>>12426695
I can count the elliot waves

>> No.12426722
File: 31 KB, 497x280, if_eth0-day2.png [View same] [iqdb] [saucenao] [google]
12426722

this one broke resistance

>> No.12426867
File: 479 KB, 2000x1000, u770707rl.jpg [View same] [iqdb] [saucenao] [google]
12426867

Yeah, it is a funny thing with TA, but none, of course, one of you semieducated wannabees could pin down scientifically. This actually is, why I mostly tend to react by cringing when having read a thread of the given topic where some of you here at /biz/ were making fun about the subject.

Never ever came up the concepts of deductive-nomological-model, the various problems of induction or questions around correlation of variables in financial markets for discussion.

It's pathetic to watch you try being superior.

>> No.12427407

>>12426867
Cool down bro, what I posted here is imho in favor of ta, although I use ironic biz-meme-speach. It was an important insight for me to know ta can be applied to more stuff than finance-charts. It seems to be an universal tool.

>> No.12427603
File: 1.02 MB, 3105x2033, Sigourney-Weaver.jpg [View same] [iqdb] [saucenao] [google]
12427603

>>12427407

Again, very cringy. Saying sth along the lines of "It was an important insight for me to know ta can be applied to more stuff than finance-chart" either means you did not understand that TA is

a) not a well defined method of any kind;
b) not by any means sth that could be called a theory.

Please also think about the following: If (!) TA would be a method and/or theory applicable to financial markets only, by what obstacle wouldn't it be applicable to another field? TA is as it stands now is applicable to time series represented by index data points in a two-dimensional coordinate system. Ok. Why exactly is it NOT applicable to three-dimensional representations of time series representations modelling the growth of say bananas over time in length AND weight? What makes it impossible to apply it to this field? The answer is: Nothing. How would that be a moment of insight for you? It goes by easy: TA is from the standpoint of theory of sciences bogus when applied to financial markets, because it cannot be made clear a) that is has any predictive value and b) it is not in any way a mathematically feasible theory.

>> No.12427641
File: 100 KB, 500x334, 1547238226064.jpg [View same] [iqdb] [saucenao] [google]
12427641

>>12427603
Nice post. I get that TA is not 100%, I don't pretend to even understand it. I just use ichimoku with mcad sometimes to short. But I think ta works when alot of people do ta, then share those graphs and shit and more people place positions accordingly, increasing the chance of that prediction coming true. But please anon tell me more you seem like you know your shit

>> No.12427680
File: 205 KB, 650x845, SIGOURNEY-WEAVER-2.jpg [View same] [iqdb] [saucenao] [google]
12427680

>>12427641
> But I think ta works when alot of people do ta

Yes, at a point in time I thought about this argument, too. Seemed to me like a self-fulfilling-prophecy-type thing. There can, however, be made a quite convincing empirical argument against this:

Retail traders do not move markets. Period.

I worked in the primary markets division (DCM) of a quite significant bank. Big volume is not moved around in the market on the basis of TA. Not even "not alone on the basis of TA". So there it goes: Even if some thousands of those poor retail guys think along the line of some chart formation (aka DOW-theory) or simultaneously decide to focus on something obscure like the ichimokus or fibonaccis, it wouldn't even come close to a storm in the water glass. By "empirical" I therefore meant: In my time in the capital markets I never saw people responsible for larger volume trades base their decision on TA. Same goes for clients (as far as I can tell) or other market participants from the institutional side of things.

>> No.12427727

>>12427603
Thanks for sharing this. Quality post.

It seems you already have put some thought in this. Unfortunately I have not put much thought in this. I just work day by day on some server stats, trade a small btc stack with not much success and read some finance books in my leisure time. I recently read Benoit Mandelbrots work "Misbehaviour of Markets", which gave some insights regarding resistance/support areas and trends - and the applicapability of fractals stuff besides finance markets. After reading this I saw ta-elements on server stats and thats why I shared this. I'm sorry for my incompetence - I have otherwise nothing good to contribute.

>> No.12427921
File: 43 KB, 630x630, 5ec4gp2gpqu11.jpg [View same] [iqdb] [saucenao] [google]
12427921

>>12427727

You by no means seem incompetent - you're thinking about complex stuff here. Maybe this could be inspiring: Instead of having the view that there are only combinations of the following "methods" - let us be liberal here - to be applied in the analysis and ultimately forecast of markets being

fundamental analysis
technical analysis

let's instead look at the confusion of TA and quantitative analysis. You have to understand how properties or characteristics of data coming directly of the physical manifestation of a certain financial market can be harvested for insight in the first place. Tick data from exchanges gives you information about open, high, low, close prices index by timestamps plus volume data. If your mandate is to go and make prognostics, then THIS is what you can use to deduce said properties/characteristics. Go and derive concepts suited, i.e. statistically relevant properties of parts of the time series in certain ways for the forecasting (not only) of prices, but also volume, volatility and many more properties. You have to see how different TA is: TA (DOW theory + the various indicators) does not do that. By simply drawing a horizontal line in a chart...how, statistically, mathematically, how do you justify that this WILL BE a relevant support or resistance in price action IN THE FUTURE? You merely claim it to be...

>> No.12427949

>>12427921
stfu dork. Your fancy punctuation and vocabulary doesn't even make sense and looks forced. Go larp somewhere else.

>> No.12427951
File: 100 KB, 1154x431, Screen Shot 2019-01-13 at 9.25.16 AM.jpg [View same] [iqdb] [saucenao] [google]
12427951

easiest swing trade in history here

>> No.12427972
File: 41 KB, 379x298, 1543897800101.png [View same] [iqdb] [saucenao] [google]
12427972

>>12427949

I seem to have enraged a zoomer. Curiously I do not feel sorry.

>> No.12428090

>>12427949
Even if it's larp, it's probably the best series of posts up on /biz/ right now, which isn't a big accomplishment

>> No.12428144

>>12426695
>>12426715
>>12426722
Can you post a longer chart? It's impossible to see on this whether it really moves like a "TA market"

>> No.12428178

>>12428144

Bait? That's what I was trying to bring to attention: There is no such thing as a "TA market". How could a wider section convince you TA was applicable to this?

>> No.12428227

>>12427921
Your points regarding TA seems to be correct for me. I'm a software developer and until some month ago I tried building trading algorithms. After thousands and thousands of backtesting (divided train + test data) using tools like neural networks, I came to the conclusion that it's really hard to create a trading algorithm based on TA/neural networks/etc which is profitable. Thinking afterwards this was expected, but it still taught me lots of lessons: to be realistic in this business and how to proper test your assumptions.
You are absolutely correct, you currently can not scientifically test TA. Fortunately I read some works like "Market Wizards", to look for hints how successful traders do profitable trading. It seems to me, that trading is more like art, instead of something scientific. That somehow made sense, it explained, why my algorithm was not able to be profitable. And that's one of the reasons why I'm still interested in TA and charts. I don't know where all this is leading, I don't know much, but after seeing lot of people talking about "profitable trading algorithms", I know lot of other people are more ignorant than myself.

>> No.12428286
File: 35 KB, 497x280, if_eth0-day3.png [View same] [iqdb] [saucenao] [google]
12428286

>>12428144
unfortunately I cannot change the size without some developing work. but here is the current chart of the first. if people are interested, I can regularly post more of this or work on bigger charts.

>> No.12428406
File: 39 KB, 497x280, if_eth0-day4.png [View same] [iqdb] [saucenao] [google]
12428406

Here is another chart which buffles me. Some independend activitys seem to form a resistance/support line right above 300Mbit/s.

>> No.12428416
File: 180 KB, 696x924, CcDXPZ-UsAAI2su.jpg large.jpg [View same] [iqdb] [saucenao] [google]
12428416

>>12428227

Ok, there is a plethora of things I'd like to say to various aspects you mentioned. I'll try:

"I came to the conclusion that it's really hard to create a trading algorithm based on TA/neural networks/etc which is profitable" - This is my experience, too. This goes for a greater extend for algos making use of classical TA indicators. The neural networks you were speaking of (or support vector machines) part is not the problem, because this is proper maths brought in to do statistically relevant discrimination in the data. With this I am trying to say that even the best HF/HFT's do not use anything different from this. I believe the main problem is that retail guys like you neither have the proper tools (high quality data, software, even hardware), nor the knowledge (could be you're a capacity in ONE mathematical theory, but what's needed is a way more expertise), nor the manpower to come up with what could be called proper indicators. By this I mean statistically proven to be relevant properties or characteristics in the time series to be used to build an algo upon.

"You are absolutely correct, you currently can not scientifically test TA" - Yes, I believe this, because there exists no such thing as a TA-theory that could be formalized and tested by scientific means to be true or false or statistically making "good" forecasting of certain properties of let's say prices for example.

"Fortunately I read some works like "Market Wizards" - I would not recommend this for anything else than entertainment. It's traders porn. You will not, I repeat, you will not learn anything applicable for building a trading strategy.

[comment too long, I will have to split]

>> No.12428424

>>12428286
I'm really interested. The first thing is that no one uses line charts, it would be cool to have the data in a usable format.
What would be really fun is having someone try to actively trade on it and see if they can make a profit.
>>12427921
>>12428227
You're clearly samefagging. Why the fuck ...?

>> No.12428451
File: 143 KB, 915x1113, Capture.png [View same] [iqdb] [saucenao] [google]
12428451

>>12428227

" It seems to me, that trading is more like art, instead of something scientific. That somehow made sense, it explained, why my algorithm was not able to be profitable." - I do not believe this to be correct. The whole "trading is an art"-narrative aka discretionary-trading-is-the-way-to-go-thing, market feeling, making years of experience necessary is just another meme in the industry. Do not take that hit! Schwager's stories do not explain in any way why your efforts in algorithmic trading failed. There simply is no connection.

"And that's one of the reasons why I'm still interested in TA and charts" - I can only try, again, to encourage you to get away from TA and look into quantitative trading literature.

>> No.12428484

>>12428424

What are you talking about? Posts no. 6, 8, 10, 12, 15, 18, 22 and 24 are mine. I do not believe I somewhere made the impression replying to any of my own posts and thereby claiming to be another person, if that's what you're implying.

>> No.12428514

>>12428406
Is there anything different about the days it goes over the line?

>> No.12428530

>>12428416
>You will not, I repeat, you will not learn anything applicable for building a trading strategy.
I read it after I had a little crisis after month of working on my algorithm and failed hard. Probably you are right, and I just used it to ease myself or remain sane.

>>12428424
>You're clearly samefagging. Why the fuck ...?
lol

>>12428424
>What would be really fun is having someone try to actively trade on it and see if they can make a profit.
It will take some time, since I'm currently busy with other stuff, but one day I will create a new thread. Here I created a tripcode, you can find the post on some archive website. It will probably be in february or so. I will also do some analysis on the data.

>>12428451
>look into quantitative trading literature.
Thank you! I have not read any of this. Seems interesting!

>> No.12428563
File: 101 KB, 1277x618, Dow_Jones_Time_Frame_Confluence.png [View same] [iqdb] [saucenao] [google]
12428563

>>12427921
>Go and derive concepts suited, i.e. statistically relevant properties of parts of the time series in certain ways for the forecasting (not only) of prices, but also volume, volatility and many more properties. You have to see how different TA is: TA (DOW theory + the various indicators) does not do that. By simply drawing a horizontal line in a chart...how, statistically, mathematically, how do you justify that this WILL BE a relevant support or resistance in price action IN THE FUTURE? You merely claim it to be...

Technical Analysis which is carried out on a single Time Frame (or time series) is NOT done correctly.

Charts aren't constructed of Single Time Frame. And so, if an assumption is made that Trends occur in markets, it must be assumed that they occur on ALL Time Frames, and so, Higher Time Frame Trends impact Lower Time Frame Movements.

Markets are NOT random, they Trend and Range. And this takes place on Multiple Time Frames.

A proper testing of an aspect of TA, such as Moving Averages, must incorporate analysis of its impact on ALL Time Frames.

Academics, frankly, don't know how to trade with TA (and I'm not claiming to be a fucking master, but I realize that there's more to it than just "drawing lines"). And these are the people who come up with the literature/opinions.

>> No.12428603

>>12428484
You are a regard trying to sound smart,which is why you don't use plain language.

Retail is far greater than institutional money in many asset classes.

What you call "Quantitative Analysis" is just a more advanced version of TA. Meme lines are very old.

Not all asset classes have computerized tradin or HFT.

Quantitative modeling has the exact same problem as TA. Market conditions change. The past can Not perfectly predict the future.

This isn't for the retard, its for the lurkers. It doesn't have the mental capacity to see its own low iq version of autism.

>> No.12428620

>>12428514
at 0 o'clock and 11 o'clock two independend publishings occur, which cause the spikes. But the chart seems to form a flag right before 11 o'clock like it knows a spike is about to occur. How does this happen?

>> No.12428688

>>12428620
Could be your imagination. Could be some artefact of how the system works. You'd know better than I would.

>> No.12428715
File: 44 KB, 497x280, if_eth0-day5.png [View same] [iqdb] [saucenao] [google]
12428715

here's the worst TA eeever. Our first chart retests the 350Mbit/s level.

>> No.12428717
File: 702 KB, 1118x1385, 1118full-sigourney-weaver.jpg [View same] [iqdb] [saucenao] [google]
12428717

>>12428603

I still have to digest post no. 28, but what you said in >>12428603 is nearly all wrong:

"You are a regard trying to sound smart,which is why you don't use plain language." - Well, I have a PHD in model theory and I was under the assumption that I broke things down nicely using as few technicalities as I could. I honestly didn't intend to sound smart, I do not know what to say else about this accusation.

"Retail is far greater than institutional money in many asset classes." - At first I thought this must be false on all possible ways of interpretation, but there is CFDs and binary options. You c-o-u-l-d see this as "markets" and then the retail side clearly outnumbers the institutionals, but this is, of course, because no institutional would be dumb enough to do business in there. Speaking proper capital markets here...well then institutionals always outnumber retail guys.

"What you call "Quantitative Analysis" is just a more advanced version of TA. Meme lines are very old." - This clearly is wrong. I tried to argue for this above. TA is by no means a method or theory. Algorithmic trading however seen as quantitative finance put into practice can at least be put under proper statistically evaluation, because it can be formalized and is therefore (every algo on its own) a proper theory.

[again, have to split]

>> No.12428761

"Not all asset classes have computerized tradin or HFT." - A conjunction by itself, so this is tricky. The metals' exchange in the uk is still open outcry, I believe. But I cannot recall any other market not being computerized in the proper sense by now. As for HFT's, well, there in the FX, equities, fixed income, futures markets.

"Quantitative modeling has the exact same problem as TA. Market conditions change. The past can Not perfectly predict the future." - Supposedly at random changing conditions are essentially what markets are, yes. Therefore any method or forecasting has to deal with this. What are you actually trying to say here? Prediction is at best statistically better then random, yes. Again: Have you anything worth to tell in addition?

>> No.12428945

>>12428715
Yes, well done, you've shown how anything with a Cyclical movement can be highlighted with Technical methods. It's almost as if TA was specifically designed to monitor changes in Market Cycles.

>> No.12429013

>>12428563

In logics (and I believe in mathematics, too) there is a awkward thing that goes a little like this: If someone really craps total nonsense out of the largest of his openings in the his head, then he will not experience colleagues lecturing him for the utter nonsense he said or so, but instead he will experience quietness. This is because what he said is clearly not true, but - even worse - it is not false. It is such crap, that it doesn't qualify as a thing called truth bearer: It can neither be true or false. It is my view that this is true for all you said in post no. 28, except these two things:

"Charts aren't constructed of Single Time Frame" - This is simply false, because charts are constructed out of atomic data aka tick data: You decide about the time frame for the actual chart and then start calculating out of the OHLC-data what you need to make your points for the graph.

"Markets are NOT random, they Trend and Range." - To the economists (and involved mathematicians) best knowledge this is a hell of a claim and most of them would not agree, but instead disagree. Also: I agree, that there are trends and ranges, BUT the appearance of trends and ranges is - to our best knowledge - random. Therefore trends and ranges do not help you to argue for non-randomness.

>> No.12429055

>>12429013
Yes, nobody knows when a Trend will end, and when there's going to be a Range breakout. To help with this, you wait for something called Confirmation.

A Candle must Close above/below a Trend Line, as it Breaks it.

A Candle must Close above/below Resistance/Support, as it Breaks it.

And even then it's not a "guarantee". There are NO guarantees in financial markets. And there is no form of analysis that can predict future prices (maybe hfts can, but only for like a Tick or two). TA helps put things into context for a trader.

>> No.12429094

>>12429013
>In logics (and I believe in mathematics, too) there is a awkward thing that goes a little like this: If someone really craps total nonsense out of the largest of his openings in the his head, then he will not experience colleagues lecturing him for the utter nonsense he said or so, but instead he will experience quietness.
In logic, there is a things that goes: if someone uses more words than they need to, they're probably bullshitting.

>> No.12429194

>>12426695
"When your mind becomes obsessed with anything, you will filter everything else out and find that thing everywhere”.

>> No.12429498
File: 952 KB, 500x685, 1545436585771.gif [View same] [iqdb] [saucenao] [google]
12429498

>>12429194 sounds like an explanation for >>12429055

>> No.12429735
File: 61 KB, 850x1024, B5B50F61-4FD7-4AFC-88FC-D1539FF84246.jpg [View same] [iqdb] [saucenao] [google]
12429735

>>12428227
HFTs don’t make money by trying to predict the market. The name of the game is quick, risk-free trades where they are in and out of a position in milliseconds. Usually this involves manipulating the order book somehow, like frontrunning your buy order and selling it to you one tick higher.

Trying to predict the markets is a giant meme. Hedge funds try to do this, and guess what 90% of them eventually blow up. It’s all random.

>> No.12429902
File: 1.65 MB, 1964x2552, 1413242409003.jpg [View same] [iqdb] [saucenao] [google]
12429902

>>12429735

"HFTs don’t make money by trying to predict the market" - Kinda, sorta true. But this goes mostly for what is called latency arbitrage - and this is what you are aiming at in your post. Here they do not out of the blue predict market behavior in hours, days or even longer periods in the future. What they do is they legally (by being just "faster") front-run buy/sell-orders of other market participants and they get this information by very high speed data bought from venues. This, however, does by no means deplete the available strategies of HFTs. There are other approaches (like market making, which indeed is a proper HFT strategy) in which risk is considerably higher (see Trading Machines/Haim Bodek).

"Usually this involves manipulating the order book somehow, like frontrunning your buy order and selling it to you one tick higher." - Yeah, no, this is not right. They front-run you, yes. But they don't do this by manipulating (spoofing in) the order book - this would be considered highly illegal by today's standards. They get the information of an actual interest of buying/selling just so early, that they can monetize this.

"Trying to predict the markets is a giant meme" - Well this is VERY interesting, because I believe this must be true to a very high degree. I strongly believe most cases in which someone tells you he's a profitable trader it's simply a lie. But we do know of cases in which it is a fact that they can predict the future and they do it over extended periods of time (RenTech).

"Hedge funds try to do this, and guess what 90% of them eventually blow up." - True, but wouldn't say "blow up", more like "loose money" or aren't profitable.

"It’s all random." - Yeah, begging the question then how RenTech did it year by year.

>> No.12429997

>>12429902
>Yeah, begging the question then how RenTech did it year by year.
Either
>insider trading
>randomly getting it right in a long streak (it’s possible to flip heads consistently 1000 times in a row)
>they actually have some secret knowledge on how to beat the market
which is more likely? Pareto’s principle, they’re doing insider trading.

>> No.12430045

>>12429997

Nah, don't think so. Initially Simons did it with pairs trading, which was working nicely back then (70s - mid 90s if I'm correct). They are all quant: Mathematicians, physicists and so on, no wall street infected soul has ever entered the holy halls of RenTech - it is an actual policy there. Mathematics can help you beat the market. Lately I read some articles about support vector machines in the FX market hitting mid 60% correctness with their trades. I endorse your being skeptical, but maybe there is a way.