Tfw you realize trading chart patterns can actually predict the future

>tfw you realize trading chart patterns can actually predict the future

Holy shit Veeky Forums. I'm literally shaking right now....After performing a Monte Carlo Simulation of 1000 randomly generated stock data I'm shocked to realize that some chart patterns actually work and generate low risk/high returns on average.

WTF is this black magic...?!

I have so many questions upon this revelation.

Does this prove that the universe is actually deterministic if the future can be predicted?

Is quantum mechanics and multi world theories BTFO?

Is the universe a simulation?

Are there any successful traders here on Veeky Forums who trade using chart patterns? What do you think about the nature of the universe if the future can be predicted?

Also
>Efficient market faggots and economishits please LEAVE

Other urls found in this thread:

en.wikipedia.org/wiki/Stationary_process
joshknows.com
twitter.com/SFWRedditImages

Yes they are based on psychology.
They are not always accurate tho, there are other factors, like news for example.

No, it just means us humans are incredibly dumb and predictable.

dump more pictures

>randomly generated stock data
>using TA based on people's action

wat. How did you generate the stock data?

>randomly generated stock data
No, you can't randomly generate stock data, or your results will be scewed. The data is always based on mass psychology, panic selling etc.

Yes. I do, in TradingView since I subscribed a few days ago, and I somehow got very good reputation for predicting crypto trends in a very short time.

I also gain about .05 BTC trading in polo by these trends, but I have to be very careful and often I do not have enough time because wagecucking. Also, I'm trying rather to recover the 1 BTC I lost in Polo before reading trends.

I still cannot believe some people take my predictions seriously, and I cannot believe they are so accurate for someone who does not even understand what longing or shorting is, like me.

Actually I'm afraid you will find that markets are efficient, you can't beat the market. Anything but long term holding index funds is gambling, you'd be better off playing roulette.

What's your @ ?

You've been predicting breakouts in an extreme bull market, so they've been breaking upwards alot.

Try this in more bearish market conditions and it won't have the same kind of success.

which patterns work OP?

>Try this in more bearish market conditions and it won't have the same kind of success.

k.

Ok waggie, don't buy LTC and stay poor.

All the major ones to some degree. But you really need to get your time scales, trade periods, stop levels and take profits exactly optimized to your risk level otherwise it won't work. Proper risk management is where most fail IMHO. Like there may be cases where the breakout is going to happen but you get closed out by a short term crash and then don't have enough money to trade the breakout.

ITT: 3 smart people and a bunch of Harry Potter students

angladaroth

you got yourself new follower

>tfw too dumb to understand TA

Yes the Universe is deterministic and free will doesn't exist.

This is the real red pill OP you can't unlearn this.

[citation needed]

>tfw predicted pivx would go up and it went down
Man I am a retard, where can I learn this shit?
Someone help :(

Do you look at price in USD or BTC?

note these triangles were drawn before the downward trend, although I see a descending triangle in the BTC price but not the dollar price.

head and shoulders apparently works 80% of the time.

The triangles roughly predict a 'breakout', in which direction it will break out is your guess.

I see. I sorta knew that, but I guessed in a bull market it should have gone up. So would you have recommended to just wait and see for next time?

>shows one graph

Nice survivorship bias you got going on there m8

ITT we tech analysis woke now

Can we apply this to stocks people can buy with actual accounts not crypto pls?

whyever would you want to do that?

Yes, people who have studied behavioural finance and stock markets for years and dedicated all their lifes to research the financial world, but you user have solved the mistery of life. Now everybody will be able to be richt thanks to your teachings.

t. Economic graduate with a master in finance and will start working in the financial sector next year.

Can anyone teach me about this?

your point?

that OP discovered what PhD in finance / economics and nobel prizes in economic sciences didn't before, proof that only with past information you can predit future information.

Maybe you should understand what you understand the Post first.
>Can anyone teach me about this?

>Also
>>Efficient market faggots and economishits please LEAVE
btw maybe you should inform yourself first before talking there are 3 different types of Efficient Market Hypothesis
- Weak Form
- Semi-Strong From
- Stron Form

Also there are thousand of studys out there why Fundamental wins Technical Analysis. Why do you even think companies pay $3,000 for a 4 pages report of 1 single company with funamental analysis. Probably people here don't even know this, IBs Research Department sell their reports to fund managers for around $500-$3,000 the report which usually only analysis one company and is around 5-7 pages long.

Also in uni used to have classes where we used DY, EPS, DG, and other ratios to try to see if it's correlated with share price running montecarlo analysis and then regressing it. Guess what the result was.

Sigh... Sometimes patterns work, sometimes not.

I've had plenty of uck just using 61.8% fibonacci level, that doesn't mean that's a solid trading strategy.

I really don't know why I write this, since last week this board was shilling pajeetcoins hard.

You guys are a bunch of idiots.
See pic related.
You know what this is? It is white noise.

A random time domain signal with a million "head and shoulders".

You know what happens in frequency domain?

A random time domain signal with a million "head and shoulders".

If you could extract information from white noise, it would stop being white noise.

The end.

>inb4 many typos and other mistakes. I'm using my phone and don't really care about writing correctly in here.

Technically it's called stationarity. As said before, I have tested past DY, EPS growth, DG, etc. data to try to predict future share prices in a class, and it never was correlated.

We always used stationarity for our daata. Using a first order auto-regressive model AR(1) which has the following structure:
yt = ϕ0 + ϕ1yt􀀀1 + ϵt , t = 2, , n.

Without the "white noise" you mentioned the model above wouldn't work. So yes. I know what it is. thanks.

But to be fair that looks absolutely nothing like any price chart.

Markets aren't just a bunch of random prices, the chart is a story showing the tug of war between the buyers and the sellers.

Let's rename "head and shoulders" to:
"one weak failed rally, followed by a stronger failed rally followed by another weak failed rally"

Maybe, just maybe that hints that it's not going up any further. Statistically head and shoulders indicates a trend reversal 80% of the time.

>But to be fair that looks absolutely nothing like any price chart.

yeah if you could learn about stationarity and autoregressive models before commenting it would be nice. thanks.

Wow, your first MATLAB plot, I am guessing you are a 2nd year electrical engineer. If you can't extract your signal from your noise reliably then your model is doomed to fail. A lot of these cryptocurrencies that have any volume are nowhere nearly as noisey as you think they are. Just look at the volume on coin market cap of a coin that has 20k market value and 200 million and you will see that one is noisey and one isn't.

No, it is called white noise for a reason. It is a technically well-established and well-understood term, because it is encountered in every communication engineering application.

One characteristic of white noise is the absolute lack of correlation between each data point - the autocorrelation in both frequency and time domain is a perfect dirac.

Your autoregressive model assumes there is some hidden signal along the noise, a pattern that has some information, hence correlation between data points.

As a model it is fine, but it does not represent financial movements, because they are a stochastic process.

I just googled this chart dude.

Okay reduce the high frequency components and make it some band-limited white noise, or model it as band-limited AWGN and your time domain will -look- more like the chart you see.

You could even do something like a moving average and apply it in real life and you could "see" some trend up intil the point the trend would flip.

All those methods have in common that they do NOT deliver you additional information but are merely averaging tools.

>No, it is called white noise for a reason

oh boy.

en.wikipedia.org/wiki/Stationary_process

> parameters such as mean and variance, if they are present, also do not change over time.
>As an example, white noise is stationary.

desu i don't know why I'm wasting my time here. Go to a master in finance and learn about financial markets and stochastic processes.

Nah, I'll continue making consistent returns in the market.

Probably because you are a bitch to put all your life savings into your nice models and watch them melt the second your nice predictions turn out to be different from reality, but you would still like to boast your intelligence.

Go ahead, and post screen caps. Do it for the lolz.

Honestly, look at Bitcoins price, the SNR is pretty high although the sites I look at do some averaging (which acts like a filter). The point is, the SNR is high enough that you can get something reliable, even with just your eyes.

I don't invest. I know the theory behind it and all the financial instruments, how they are priced, hedged, etc. Even the most complicated derivates that are out there. But still don't like it.

I prefer analysing companies and people paying for my information - this summer I'll join a bank, so I'll have my consistent returns too.

Everyone has his trading strategy ;)

There can not be an S though, which is the whole point I'm making. For an S to be there, you would need to have something like

- A buys 300 BTC every monday regardless of price, then sells 50 BTC every friday, still regardless of price.

Some kind of reaccuring periodicity, which would quickly be known and exploited.

There is none, and moving averages or any kind of filtering will only tell you what happened, but they are unable to predict the future, because they only take past values into account. Unless you implement a working noncausal filter.

what platform? can't find you on telegram or twitter

A signal that carries information does not have to be periodic. Think about sending binary data over a channel, such as 0111111101001. There is obviously a carrier frequency, but the signal itself can still be reconstructed. Am I missing something?

As for predicting the future, I get what you are saying. I guess people have studied this kind of stuff in the past and found recurring patterns (like ascending triangles or whatever) and that may be the best we can do?

>Holy shit Veeky Forums. I'm literally shaking right now....After performing a Monte Carlo Simulation of 1000 randomly generated stock data I'm shocked to realize that some chart patterns actually work and generate low risk/high returns on average.


this is just overfitting or your randomly generated data isn't actually randomly generated

if your data is just white noise then by definition you're not able to predict it - what you quite likely have done though is fitted a model to noise

on the other hand you could use conintegration on two white noise time series - but this isn't, in itself, giving you anything useful in the real world


anyway - you should test using real data not random noise - however if you're finding patterns in random noise that ought to serve as a warning to you that your methodology is flawed... your pattens so far you know shouldn't be there how are you going to properly back test with real data if you're overfitting noise?

investment bank equity research is more about marketing and is often given away to clients for free

You went to grad school before ever working? MBA straight to management right? Have fun wageslaving for accountants

user, do you want to know more? just search bulkowski.

just a tip, TA works every time, until it just doesn't.

Did you also randomly vary the standard deviation and mean of your monte carlo simulations? No? Then you're in for a bad time when you actually try to use this.

the other poster is correct - what he posted was white noise, that is a well understood term - you've conflated that with stationarity - yes a white noise time series is stationary however not all stationary series are white noise. The other posters point referred specifically to white noise, if you've not encountered the term before then I'd suggest you've only done a very superficial study of time series.

What can I read about day trading?
I didn't sell my shit before going to sleep and woke up to -50% fell asleep for one hour longer and thanks God I woke up for +10%

Trading view, dear.

Any good resources to check out for a newbie like me to learn how read charts and trends?

???? Here in Europe having a Masters degree is pretty basic. I did master in finance, mba is different shit. And as stated in another comment won't be wageslaving for accountants, I'll join an equity research department of a well known investment bank in london this summer... so yeah.

learn how to read the thread first

>I'll join an equity research department of a well known investment bank

replaced by machines within a decade... as will a large portion of the buy side PM roles those guys desire to go into later...

ETFs and hedge fund replication strategies will take over that sector. The fund managers that survive will be mostly quantitative and a PhD in Stats or Machine Learning will be more useful than an MSc in Finance and a CFA.

>Are there any successful traders here on Veeky Forums who trade using chart patterns?

Yes.

Seriously, so you do not believe that one can build an accurate predictive model that performs well in both bull and bear markets? That's cute.

Wait, do people actually believe in the efficient markets Bullshit? I mean even for people comfortable with volatility with sufficient capital not to need to sell at an inopportune time?

lol jesus christ

it's radiation left over from the big bang.

there is order in chaos.

also, probably everything is based on the Fibbonaci Sequence/Golden Ratio within an infinite inward spiral.

>there is order in chaos.

Amen, brother.

...

...

>white noise

I think you've missed the point the guy originally mentioning white noise was making

OP has used randomly generated stock data not actual data

you're a moron

Yeah not really. Eventho algo trading and HFT is increasing popularity that only affects fund managers and AM companies. For Research, M&A, ECM, DCM, you need to analyse companies business models and its balance sheet. Every company is different and industries vary a lot (it's no the same a bank, a reit, a pharma company and a logistics company), you need that personal judgment that AI cannot do for the moment (and when it's able to do it, it will mean 95% of jobs in the ALL the sectors will be taken by them). Same for trader, market making required the need to try to get a touch or feeling how the financial market is and what it will want.

Buy side investment management firms, yes; but investment banks are okay.

And even if that would be the case, you still need an MSc in finance to program the algo correctly in the financial world. I've enconteres this a lot here. Engineering students thinking their the best and because they're strong with maths they will get all the jobs in finance. I don't know man, I don't go to hospitals and pretend I know better than doctors. Yes maths and stats is important, but we cover that in finance too. There is a moment where you need knowledge of products, the markets and other technical stuff. Just go to any bank and check their employees background...

>fibonacci

Kek. My gf interned in a small company that did fibonacci trading and in her own words: that's so much bullshit

The ones and zeros are already information, because they are well-defined states.

If you used the most simple transmission method, on/off-keying, it would result in a big lobe from the carrier frequency along with smaller side-lobes representing the "1" or "0" signal being recieved, which would bounce up and down with time.

If you plotted said transmission on a histogram (x-axis is frequency, y-axis is delay, color is magnitude) you could clearly see an emerging pattern standing out from the noise.

it would look like pic related, source: joshknows.com

...

...

...

I can see head and shoulders just like when I'm in the shower.

>technical analysis is complete bullshit
>OP fell for the meme like thousdands of others
>not realizing the returns on technical analysis have been studied and they're worse than buy and hold strategies

The proposition here seems to be,

because I know where its been, I know where it is going.

And so how do you really test that objectively?

OP here. I adjusted some variables by one tiny factor and the entire completely broke down meaning chart patterns don't work for shit.

Excuse me while I go kill myself.

I'm a wealthy moron who makes bank based on his predictive models.

Confirmation bias is a hell of a thing user

So a long-term generalization error of

I'm saying, go and find a hard bear market and make some money using your technicals and then write up an ebook audited by independent accountants and then cash in your genious on CNBC

But I don't care about any of that shit. I have no intention of ever disclosing the internals of my models to anyone other than family. I haven't even disclosed to you what it is this I am so accurately modeling and I won't be. As for bear markets, my models are trained on data going back to the 80's. They have already seen a few excellent bear markets. I trust they will get it right when the time comes.

Yeah at uni we used model with data from early 20th century. And it doesn't work. You can keep your """"family secret"""" to yourself. Nobody is interested in bullshit

The secret to applying machine learning to the markets is knowing which questions to ask.

or uttering useless platitudes because you are actually fucking clueless.

Don't share your "models" you fucking dolt. You wouldn't want them picked apart mercilessly by the five anons that are actual quants on this board...

He probably uses moving average crossovers and few technical indicators thrown in there to get a 100% accuracy rate after maximum overfitting.

Its very easy to outperform during a bear market. Just filter out any stocks, bonds or commodities that have risen and ride the momentum.

Since when can the future not be predicted you fucking dipdog.

If you throw a ball in a gravitational field you can predict where the ball will be with much higher accuracy than these charts? What does that tell you about the nature of the universe.

I believe in your abilities user. If you're so inclined............email me [email protected]

Not biting on the model details, but here are some relevant facts on one of my favorite models:

training accuracy: 99.3%
testing accuracy: 98.5%
15 month generalization accuracy: 98.1%

Nothing about that even resembles overfitting,

>relevant facts

One anons relevant facts are another anons untestable evidence

What we really want to know is, how high will LTC go on this segwit launch in 6 hours

ITT: People who know what they're talking about shit all over OPs dreams.

c'mon, let OP learn the old fashioned way.

>pic very related
>