/biz/ - Business & Finance

First for my waifu, RKG
May her beauty surpass her skill for shitposting

first for bulls won and fuck the haters

Second for vaneck Fallen Angel bonds

>>10278760

nth for CHINKS BTFO

also nth for ratsuchan

whats some good stocks under 50mil market cap to watch this week? Anybody think Harblongs LTBR gunna do something soon?

>>10278848
Steel and Aluminum fucks both sides. I have people close to me in manufacturing whose profit margins got fucked twice over months ago.

>>10278836
Fuck.
I could not into anime girls until you started shitposting.

check it out goyim
my first attempt at making an indicator for tradingview in pine editor

its basically just a rip off of what that one ciovacco guy does with the array of a bunch of smas to judge trend and momentum. gonna try and add some more things to it to make it more interesting. comments and criticism very fucking welcome

------------------------------------------------------
//@version=3
study('ciovacco ripoff', overlay=true)
sma1 = sma(ohlc4, 20)
sma2 = sma(ohlc4, 50)
sma3 = sma(ohlc4, 75)
sma4 = sma(ohlc4, 100)
sma5 = sma(ohlc4, 125)
sma6 = sma(ohlc4, 150)
sma7 = sma(ohlc4, 175)
sma8 = sma(ohlc4, 200)
plot(sma1, title='20', color=#00FDDD, linewidth=1, style=line)
plot(sma2, title='50', color=#00FD84, linewidth=1, style=line)
plot(sma3, title='75', color=#00FD2A, linewidth=1, style=line)
plot(sma4, title='100', color=#3BFD00, linewidth=1, style=line)
plot(sma5, title='125', color=#AEFD00, linewidth=1, style=line)
plot(sma6, title='150', color=#EEFD00, linewidth=1, style=line)
plot(sma7, title='175', color=#FDA600, linewidth=1, style=line)
plot(sma8, title='200', color=#FD2600, linewidth=1, style=line)
-------------------------------------------------------

>>10278836
Please stop arousing me on the Lord's day

>>10278848
Greeeeeeeeen

>>10278895
>>10278925
Its the editions fault not mine!

early for long netflix

I'm so excited for tomorrow. I took work off so I can sip and watch the market all day.

>>10279043
I'm fully prepared to be raped, but on the same hand, I am prepared to rape.

Threadly reminder that all tripfags should be required to do the their all time P/L.

hope this iraq stuff makes my 7-20 USO 14 calls 3 baggers.

>>10279043
what stocks will you be watching? better to take of tuesdays-thursdays imo.. mondays and fridays are kinda boring in the market.

>>10279054
If SPY doesn't crash into the ground, please rape me to death

>>10279088
very cute

>>10279130
give it time. we need to see how the first couple ERs go. if its like last time, where 75% of the record ER beats result in 5-10% drops, then i have no doub that were going down in the near-term.

watch the SMI (smart money index). its headed STR8 down. and it hasnt slowed down at all. smart money has been in "sell the rally" mode since end of jan. if last quarters ER beats werent enough to convince them, and we see the same trend for this quarters ER beats, then its pretty clear where were headed in the short to medium term.

if were truly fugged, then they have to keep it propped up until they get all their shekels out. that means suckering retail into buying their bags BEFORE pulling the plug

>>10279318
>smart money index
they should really change that misleading name considering it has nothing to do with large specs

Short Tesla.

>>10279318
I'll for sure check out the SMI, sounds legit. My puts expire on the 23rd, so I need to get out on the 17th-18th to avoid getting stuck with them.

>>10278916
Ok, sick graph.

>>10278992
>thinkingface.png
Ok, I'll take that.

>>10279043
>>10279054
>>10279130
Why? More trade war shit?
I don't recall any new information in the news

>>10279088
What Iraq stuff?

>>10279400
What would a better name for it be?

So I was in on ALT at .43, where’s the exit on this one? Is there long term potential to hit $1+ or was Friday the pump?

Thanks frens

just a man doing his best, how ruined am I?

>>10279056
It's diluted because I've deposited a lot since starting, but still beating the indices

>>10279462
>green on GE
well done anon

scoops, I read that whitepaper. interesting stuff.
things I like
>nonlinear over linear models
this is essentially reducing the rationality in the models by drawing meme curves instead of lines
>careful attention to avoiding forward looking bias
>attention to meta parameters
>frequent retains
though this is a double edged sword
>breaking up by sector
a decent proxy for controlling for macro

things I don't like
>length of tail for training data
>meta parameter choices
they essentially hand tuned these values to maximize their results. also these were both static. much better if they were adaptive. I particularly also don't like 3 years tail for quarterly positions. I think it should be a longer tail but then that's based only on instinct
>no proper incorporation of macro changes
though to be fair this is really hard, but I bet weighting in a couple of indexes could do it
>training set 1983 to 2010
there's a lot of past bias here, and I worry that it doesn't account for the modern explosion of algo trading

>>10279440
the nobody index

>>10279440
"index tracking people who sleep in"

High chance that AMZN gets their boxes from SON or RKG anybody have any other companies that might be supplying AMZN with boxes?

>>10279516
oops i meant PKG

>>10279497
th-thank you anon

>>10279432
Its earnings season, no matter what you hold or trade massive PR waves come out about everything. Whether your a pajeet, chink, or goy, shit will happen. In my case, Netflix will effect IQ.

>>10279504
>>10279508
Kek. What's your favorite technical indicator?

>>10279572
TA is a crutch
You only need to glance once at the price and you know

>>10279432
https://www.wsj.com/articles/iraq-struggles-to-contain-protests-over-government-dysfunction-1531675702

>>10279502
>scoops, I read that whitepaper. interesting stuff.
I still need to just repost the whole link bundle :/

>things I like
>nonlinear over linear models
>this is essentially reducing the rationality in the models by drawing meme curves instead of lines
Lots of shit just doesn't fit well with linear models
This includes a good explanation and isn't very long:
https://www.datacamp.com/community/tutorials/svm-classification-scikit-learn-python
Using something "too nonlinear" results in overfitting, and RBFs are generally a pretty good middle

>careful attention to avoiding forward looking bias
This is a must. Otherwise, it's training off the answer key
>attention to meta parameters
Just using a grid search of C and gamma is all this really means imo
>frequent retains
>though this is a double edged sword
Yeah, I'm not planning to do that. The recency of my data shouldn't matter for what I have planned.

>breaking up by sector
>a decent proxy for controlling for macro
Yeah, I liked this. But, I'd prefer it to be a parameter on its own instead of completely separating the models. However, backtesting should be used to see if combining the models and just using a parameter will degrade prediction accuracy

>things I don't like
>length of tail for training data
>meta parameter choices
>they essentially hand tuned these values to maximize their results.
Correct. But, backtesting is supposed to prevent this parameter tuning from resulting in overfitting
>also these were both static. much better if they were adaptive.
The parameters can't be adaptive -- that'd be like switching back and forth between a linear and polynomial model for whenever it's convenient and gives better results
>I particularly also don't like 3 years tail for quarterly positions. I think it should be a longer tail but then that's based only on instinct
I'm mainly just stealing their TA and FA features and planning to experimentally determine what timeframes give the best predictions.

>>10279590
>that boomer looks at the price

buy red, sell green. Ez money

>>10279502
>no proper incorporation of macro changes
I do plan to throw in some shit like the yield curve and momentum of TLT's prices
>though to be fair this is really hard, but I bet weighting in a couple of indexes could do it
I'm going to at least throw in some features that follow SPY, but DJIA is based off 30 hand picked companies and NASDAQ is like 95% the same so those probably won't go in.
>training set 1983 to 2010
>there's a lot of past bias here, and I worry that it doesn't account for the modern explosion of algo trading
I like having lots of history. That's more data. I want to at least include a couple recessions of data in there.
Also trading will only be an issue if the model does daytrading, which isn't the timeframe that I have in mind.

>>10279400
its stil a massive amount of wealth sneaking out the backdoor at the same time. and it hasnt done anything except go straight down since february, so i can only assume that the event thats being sold off en masse for, hasnt been realized yet.

>>10279424
Elon is so fucking bad at twitter.

"Generally the view that I've had on Twitter is if you're on Twitter, you're in like the meme—you're in meme war land. If you're on Twitter, you're in the arena. And so essentially if you attack me, it is therefore OK for me to attack back," he said.

First for Bitcoin vix indicator (not pretty)

>>10279710
i saw this meme posted here the other day. ngl it gets me shakin in my boots

Is there a way to see a list of stocks trading below NAV?

Elon is utilizing the Trump effect, attention will only make him more powerful

>>10279717
What?

>>10279710
you see, a lot of money has been made in the market so it only looks bigger than the reality. Its just a drop in the bucket these days, and large specs know the truth, and arent selling

>>10279556
>GE
I guess now might be the time to go long.
I shorted the shit out of them until the mid twelves.

>>10279732
nah, he doesn't have the knack for it. Trump does it in a very unique semi-playful way.

>>10279722
After reading up on what it actually tracks, I think it just shows that you can currently buy at 3:01pm and sell at 3:59pm and make a nice profit...

>>10278916
this type of idea getting very popular, you aware of Super Guppy?
not bad for first script

Finished making this one,
this is all the SPDR sector funds on screen + spread of all XL averaged.

idea is SPX and XL Sector Funds averaged are basically the same thing which we can see kinda true

>>10279793
https://www.tradingview.com/scripts/guppy/

Guppy!

>>10279722
on one hand it makes sense that the sell off in the SMI is actually larger, as were UP higher, and all the stocks are worth more per share (>>10279742 hivemind). but its literally only gone in one direction since volpocolypse. i think at this point, its the largest divergence ever in history of it being tracked. and its definitely the longest its gone in a divergence w out the markets crashing

>>10279762
>you can currently buy at 3:01pm and sell at 3:59pm and make a nice profit...
now think about who trades on bells and think about why that would be happening...

>>10279824
people are happy with the overall gains up till now. It's just very defensive play and not necessarily anything else. 2007-2008 spooked everyone and it's still on minds to avoid the next one. It's extremely predictable behavior.

>>10279793
only 1.5 hours left until Futures gap up and liquidate you

>>10279824
>who trades on bells
every trader
Small specs are thinking shorter term
aka bullrun golden imminent while weakhands got rekt
baaaaased

>>10279682
>Using something "too nonlinear" results in overfitting, and RBFs are generally a pretty good middle
>Just using a grid search of C and gamma is all this really means imo
You're thinking far too narrowly. markets are irrational and I'm confident that an ml strategy which better approximates irrationality would produce significantly better performance than what's essentially hyper rigid ta. I've had great success doing cc fraud detection and weather pattern prediction with adaptive, self mutating graphs
>The recency of my data shouldn't matter for what I have planned.
enjoy getting your pants pulled down by major events or even earnings reports. the whole point of ml systems is adaptive behavior based on new incoming data
> I'd prefer it to be a parameter on its own instead of completely separating the models
this will introduce some significant amount of regression to the general market mean.
>backtesting is supposed to prevent this parameter tuning from resulting in overfitting
back testing is how they did the tuning. they've used it to fit to the training data, introducing significant selection bias
>The parameters can't be adaptive -- that'd be like switching back and forth between a linear and polynomial model
no it's adjusting coefficients and weighting, it doesn't change the the model. they even state outright that not adapting the meta parameters is a weakness

>>10279887
Don't say such terrible things, Anon
>>10279916
Reminder the CHAD Nasdaq beats the VIRGIN dow

Also keep in mind the market is effected by technology increases. As we see the normal up and down cycles keep in mind that technology is improving. It was largely wasted on the focus for advertising but it seems to be increasingly moving towards other areas such as the physical world. This can be most seen with the emphasis on self-driving cars from Waymo and Apple.

The entire crux of the tech valuations and entire market right now is literally entirely hinged along self-driving cars. Which are a proxy for machine learning's ability to influence the real world.

Basically, try to get as good as a predictive ability on how well self-driving cars will work out in the next 12 months. If you are bullish on progress then you should have all of your money long in relevant tech companies. As such success will overpower all recession possibility.

dow looks as solid as it comes. summer consolidating still in play tho

>>10279699
>I do plan to throw in some shit like the yield curve and momentum of TLT's prices
>I'm going to at least throw in some features that follow SPY, but DJIA is based off 30 hand picked companies and NASDAQ is like 95% the same so those probably won't go in.
I had rkg's derived penis index in mind
>I like having lots of history. That's more data. I want to at least include a couple recessions of data in there.
>trading will only be an issue if the model does daytrading, which isn't the timeframe that I have in mind.
missing the point entirely and wrong. I said nothing about your model doing day treading. algo trading in general significantly impacts the way the market functions these days and historical focused data will train off outdated signals. including large corrections is important though

>>10279936
the DOW is more fierce

>>10279991
How well do ML algos work? Seems almost pointless to try ML unless it's just to refine already existing algo trading.

I don't really see the point unless you are doing market-wide and world-wide market super complex correlation type stuff. You would have to input so much data to get results it seems silly for a retail to try it, unless you input your own sentiments to get some help on planned trades.

/Bull_Dogs round up/
We have Cisco on discount
We have CAT on discount
We have PG on discount
We have BAC on discount

Pump Pump Pump.

You guys talk about technology affecting companies and industries, but really it's affecting the nature of trading itself

>>10280061
which is why I don't bother with any numerical related research. They are better than me.

>>10279793
>>10279807
Is trading on the /RKG/ + Guppy the ultimate way to success??

I CAN'T WAIT TILL 4 FOR FUTURES, I NEED FUTURES FUTURES

>>10280129
why, they are going to be green, dow +120

>>10280022
I haven't done any work with the stock market. You need to have good domain knowledge to craft quality algos, plus understanding of ml to pick the right strategy. It's essentially leveraged ta (doing ta on ta) in this case, with the added benefit of being able to act faster on action than a human could. But flaws in your ta or meta ta get magnified quickly

>>10279991
>algo trading in general significantly impacts the way the market functions these days and historical focused data will train off outdated signals. including large corrections is important though
i agree. whenever i sit down and start studying historical data, it always strikes me how different the market behaved on a day to day back then, compared to now. algos definitely changed the way things move.

dunno how this would effect ML tho. id assume, the more data, the better

waymo jaguar

waymo plans for 1 million taxi trips by summer 2019.

20,000 planned electric vehicles from jaguar, built to have self-driving support.
82,000 ordered from Fiat Chrysler

>>10279424
This. Musk going to get btfo Monday by his leddit "investors" over rnc contributions plus calling dude a pedo on twatter

>>10280221
yeah down early in the day, recovery by afternoon seems most likely.

why are you all buying into an already overbought market?

where my /shortsellers/ at?

Why is Nikkei doing so well?

>>10280195
>The more data the better
Not necessarily. Overfitting is a primary concern when making models, and it's using too much data, making your algo susceptible to noise or random blips. You have to carefully discard a % of the input data to get best results

>>10280269
Use handcrafted abstract syntax trees to create bottlenecks?

>>10280255
its not...

>>10280255
T-those can't be real

>bought tvix at $38
>it's earnings week
Time for tvix to drop down to 30$ this week baby!

>>10280317
You can see distorted pixels around the boobs, don't be tempted by 3DP.

>>10280292
I guess you could use asts if you're using a semantic model. But almost always it's quantitative, so fitness functions and parameter tuning. Even the graph traversal work I've done used quant decision functions and wasn't pure ast style semantic choices
>>10280297
>>10280317
Bitches don't know about my hcup jav

>>10280252
Loaded up on sqqq at close Friday. Hoping for some pic related

>>10280349
saki yanase is real, stupid animenigger

>>10279887
T_T
pretty much everyone thinks we about to hit the ranging supply zone 2800-2840 that will fuel the rocket moon to 3000

>>10280095
gonna add some buy / sell signals to that script to see how effective it really is
need that ishraes boxx guy to identify more patterns @_@

>>10280439
Post butthole

>>10280439
Kill yourself

>>10280439
Kill yourself

>>10280454
>>10280464
>>10280472

30 more minutes
I can feel the green day already
woooooo

>>10280501
Hows your butthole feeling by the way?

>>10280501
>analize
Subtle

>wait for futures
>see futures are all green
>be bored again till markets open
>watch china and asia go green
>wait for USA open
>tesla down 5% and market green

this loop is boring

>>10280524
Like it about to be gapped open by gigantic throbbing red hot glicining ES_F

>>10280582
that was on purpose and totally no byproduct of mspaint not have spell checker... yes

>>10280677
Dumb bitch whore, your life will end in 20 minutes

YES

thenotoriousmmaToday I was invited to the World Cup final as a guest of Russian President Vladimir Putin.
This man is one of the greatest leaders of our time and I was honored to attend such a landmark event alongside him.
Today was an honor for me Mr. Putin. Thank you and congratulations on an amazing World Cup.

>>10280705
never seen more line up of incel stocks
all these earning will be flaccid

so what're your predictions for BAC tomorrow guys

>>10280824

I plan to try to maintain around 0.06-0.08, you get a nice buzz going that you can come down from without getting a hangover. Not driving, of course.

>>10280818
>DPZ

Fuck you bitch, Netflix will increase 12%, BAC will moon 1-2%
But the next day is when you least expect it. GS and MS will moon 6%

>>10280872

this guy fucks

>>10280818
don't shit on netflix yet. They probably kill earnings regardless of how overvalued they might be.

THATS IT
PUMP IT
WOOOOOOOOOOOOOOOOOO

>>10280824
At least .17 for me
>>10280777
Nflx could seriously improve if they started licensing UFC. They are really lacking on sports content. But they don't have money to pay and I doubt subscribers would want to adopt a tiered model and it wouldn't offset costs anyway. So in conclusion they are like iq in that dicks are tiny and can't satisfy anyone and shares will tank soon

>>10280922
NFLX strategy of no sports is more intelligent. With sports you own jack shit and pay huge premiums. Creating their own product pipeline is better. Also the whole no commercials thing doesn't work for sports.

>>10279441

Will hit 2$ eventually. That will be the exit pump. Likely on some sort of bogus PR.

>>10279088
I love this pic

>>10280936
Agree they can't afford sports, but it most certainly can work without commercials, especially UFC/ boxing like I referenced. content creation is a huge money pit, and the shit they're coming out with sucks and is mostly ripoffs/ remakes. They'd be better off improving their association searches to show a wider variety of things to users to keep them engaged, instead of 10 bars with the same 30 items

>>10281086
UFC is ppv and high ads even there. They would be taking on something expensive and have no way to monetize it.

Second largest oil producer has protestors breaking into oil companies and camping out with tents on supply routes to refineries and futures are Down/flat. What in the flying fuck?

Fuck

>>10280920
>>10280879
me blood... surprisingly warm.....

>>10281152
Holding my puts
This market will bleed soon enough

>>10281152
>>10281151
and when asia opens up
oooooooooh

Hopefully ALT owners come out with some good news tomorrow to get the ball rolling on this bad boy.

>>10280129
Nice wojak. Saved.

>>10280022
>How well do ML algos work? Seems almost pointless to try ML unless it's just to refine already existing algo trading.
They simulate human intuition. It's like a robot watching the ticker for millennia, then outputting its best guesses.
I do just plan to build off already existing shit.
>I don't really see the point unless you are doing market-wide and world-wide market super complex correlation type stuff.
Not necessarily, but the Medallion fund is doing that and achieved returns of like 60% per year for decades.
>You would have to input so much data to get results it seems silly for a retail to try it, unless you input your own sentiments to get some help on planned trades.
lol no, you don't want to train it off your own sentiments, the point is to remove the human element. Read the research paper that I posted.

>>10280061
That's true. Instead of dialing up your broker, we have RH and stop losses, which resulted in the 1987 crash. But, I don't see your point. I'm trying to take advantage of this new nature of trading.

>>10280164
>I haven't done any work with the stock market. You need to have good domain knowledge to craft quality algos, plus understanding of ml to pick the right strategy.
This is true. But, you can still get some interesting results without even giving it absolutely everything. Although "absolutely everything" would obviously put you over the medallion fund, which has been getting 60% ROI for decades
>It's essentially leveraged ta (doing ta on ta) in this case, with the added benefit of being able to act faster on action than a human could. But flaws in your ta or meta ta get magnified quickly
>the added benefit of being able to act faster on action than a human could
No, you're mixing up ML with HFT.
I'm not even considering HFT. I'm not even going to try to snipe huge market orders or statistical arbitrage. That's been done so much, I'd have no shot.
I want to target shit like that FB Cambridge Analytica dip, and value investing.
I'm not going to just use TA, I'm going to use FA. I'm going for long timeframes here.

>>10280195
>i agree. whenever i sit down and start studying historical data, it always strikes me how different the market behaved on a day to day back then, compared to now. algos definitely changed the way things move.
The crash of 1987 is a classic example of this; it was magnified by stop losses, which were introduced soon before.
That flash crash a few years ago was a great example of what HFTs can do.
It has fiercely fucked with and magnified short term irrationality.
This is why I plan to use longer term timeframes -- the market still returns to rationality on a multiple quarters or years timeframe... OR at the very least exemplifies greater rationality on this timeframe.
You can exploit short term irrationality for better deals on long term timeframes.

Technology has changed the nature of trading, but not the fundamentals of finance!

>dunno how this would effect ML tho. id assume, the more data, the better
That's generally the rule of thumb, but the data does have to be clean too.
You can't feed it data that has the incorrect answers.
But, the reason that artificial neural networks are picking up in popularity is that they are kind of tailored for and handle using lots of data very well.
More data = more trials for the model to build its intuition off of. It's "intuitition" though, so you just need to make sure that it's high quality, clean, and correct. Any wrong answers will fuck it up.

Alright guys this is kinda of new and unconfirmed but there's rumors of a coup in China (Xi's posters are being taken down)

If this were true, how would this play market wise?

Red day or green day.

Short what, buy what.

who else bought puts on friday

>>10280269
>Not necessarily. Overfitting is a primary concern when making models, and it's using too much data, making your algo susceptible to noise or random blips. You have to carefully discard a % of the input data to get best results
Overfitting is primarily an issue with failing to use a proper hold out for the test set, or using an overly complex model, like a deep neural network when you have no formal education in this shit. (I've taken a few grad courses on it, but this is why I'm still going to stick with RBF SVMs.)

>>10280292
>Use handcrafted abstract syntax trees to create bottlenecks?
I have no idea where you're going with this, senpai

>>10280439
Idk what's with all the thirsty, spiteful anons you're getting today.
Maybe all the pajeets finally have a holiday so they can go harass all the girls in the world about showing bobs and vegana

personally, I like the CAC 40

>>10281398
You can constrain the ML algo at bottlenecks. Forcing it to convert the input and trim it down to fit into the AST.

>>10281391
sounds like BS. and if it was "official" it'd be like the Turkish coup where it is all staged to consolidate power

>>10281391
Xi seems to be fucking up so it would probably be good longterm.

>During the credit crisis, a number of European countries banned the bearish strategy in financial stocks because they thought it was contributing to market volatility. They also speculated that traders' collective use of the strategy helped drive the market down to extremely depressed levels at the height of the crisis. In 2011, South Korea banned short selling after it hit a record high among foreign investors.
I'm glad I was born in a Free Country.

>>10281144
SHALE
H
A
L
E

>>10280872
Kek

>>10279923
>I'm confident that an ml strategy which better approximates irrationality would produce significantly better performance than what's essentially hyper rigid ta.
You don't need to predict irrationality to benefit from it.
I prefer an approach that would exploit irrationality and rely on the market's eventual return to rationality -- AAPL couldn't possibly stay at $1 per share if a panic sale ever got it there, right?
I'm looking at FB's Cambridge Analytica scandal. There was no "data breach" and their bottom line wasn't impacted. I bought up a fuck ton of calls and made 2-4 grand in a fucking month.
The market is irrational in the short term, but approaches rationality in the long term.

The markets simply can't be irrational forever.

>enjoy getting your pants pulled down by major events or even earnings reports. the whole point of ml systems is adaptive behavior based on new incoming data
But that's the point. Trading the news is largely irrationality. Cambridge Analytica was a classic example of this. The market overreacts to the news.
I won't be harmed by short term drops, that's where the model should signal to buy.
Your view of ML is restricted to reinforcement learning.
ML models, particularly supervised ones, are NOT what you would call "adaptive".

If there's NEW data coming in, data that is NOT in the training set, the statistical model would have to EXTRAPOLATE, and WILL give a wrong answer.

Reinforcement learning has some capacity for thinking on its feet like that, but I'm going to stay the fuck away from it because that's not what I'm trying to do here.
I think you have goals and timeframes in mind that are much shorter term than what I'm aiming for.

WARNING

https://www.cnbc.com/2018/07/13/this-is-the-biggest-bubble-in-the-history-of-mankind-and-its-going-.html

ITS HAPPENING

>>10279923
>I'd prefer it to be a parameter on its own instead of completely separating the models
>this will introduce some significant amount of regression to the general market mean.
Which is why I'd also want to use backtesting to determine if my prediction accuracy scores will degrade as a result.

>back testing is how they did the tuning. they've used it to fit to the training data, introducing significant selection bias
I didn't catch that part, but yeah, you tune the parameters to the testing data instead, with a grid search, and then possibly test again on a hold out set.

>The parameters can't be adaptive -- that'd be like switching back and forth between a linear and polynomial model
>no it's adjusting coefficients and weighting, it doesn't change the the model. they even state outright that not adapting the meta parameters is a weakness
Ok, I see what you meant by adapting the parameters now.
The market's long-term behavior shouldn't change drastically from month to month.
I don't care as much for recency or "freshness" of the data because I want the intuition to be built up for Warren Buffett's investment style -- in theory, the model should be making less buy and hold recommendations as it approaches the end of an economic cycle, like 2000/2001.
In fact, that's how I'll feel that my model is working properly.
Notice how Warren Buffett has lately been struggling to find good investment opportunities and is actually sitting mostly on an enormous pile of treasuries. He says "there just aren't that many good deals out there"

>>10281663
>CNBC
>Ron Paul
umm, sweetie?

>>10281663
SELL

EVERYTHING

>>10281667
Shhh, anon is too young to know Ron Paul has been calling for the apocalypse since his PhD thesis. He'll be right one day, even if it's not in his lifetime

>>10281663
https://www.youtube.com/watch?v=5l4OEJ38aLY
ron/pol/ is always right

>>10281183
;_;
China GDP will be 4.7% & cause biggest crash every

>>10281398
am have whored around so much in TA world, have like 100+ strategies collected (algos and everything)
main reason am learning ML is because not have time to custom configure their optimal setting for every single stock / currency looking at. ML would be perfect for it!
(still mourning me old SPY strat config that accidentally deleted, this one was tested 107% profitable, so goood, but took me 3 weeks to configure) (pic related double itchimoku cloud strat)

Also yes kek these guys have been bad lately, think the "Kill yourself" poster is a bot

The thread lastnight was real treat, like 5 people described how they would kill me, was great

>>10281663
>cnbc
;_; that mean opposite likely to happen

>>10281222
>They simulate human intuition
NO! That's a fundamental misunderstanding of ml or psychology. Human intuition is neither logical nor rational nor quantitative. The fact that they both use pattern recognition does not make them equitable. Why do you think I keep talking about irrationality
>>10281236
>No, you're mixing up ML with HFT
No that's exactly what you're doing. I haven't brought up hft at all but you keep doing it. Stop with the fucking autism. If you don't understand how building a quant model is like doing ta then you don't understand at least one of those things
>>10281304
>More data = more trials for the model to build its intuition off of. It's "intuitition" though, so you just need to make sure that it's high quality, clean, and correct. Any wrong answers will fuck it up.
Ugh, this is such a bad explanation. You're halfway to right but the analogies are wrong and you don't have a deep grasp of how data training actually functions
>The reason that artificial neural networks are picking up in popularity is that they are kind of tailored for and handle using lots of data very well
What? No. Scale and throughout are not the advantages, and nns aren't "picking up" in popularity; they're commoditized. The advantage is being able to self reference and have individual neurons make small recursive changes. This allows for reaching decisions faster

If you hadn't face posted I'd swear you were a chink because you can't into English reading comprehension and you're stealing ideas without understanding the basic underlying concepts

>>10281391
There's no fucking way.
I'd love that, but there's no fucking way.

>>10281597
Kek
I love that.
Commies don't understand how the equities market is supposed to work and the inherent value that stocks are supposed to have.
Shorts should have literally zero impact on the efficient market price of securities.

>>10281663
>CNBC
>>10281667
Yup.
This is exactly why I mainly read WSJ anymore. Autistic progressive liberals would feed themselves whatever reality it is that they want to believe, and those contrarian cucks that dream of being the only woke person in a room (ZeroHedge) will continue to screech and don their tinfoil hats until the aliens really do bring back Elvis.

>>10281663

>It's the end of the world
>Oh wait it's Ron Paul.

Top Kek!!!

>>10281751
>chinese gdp numbers
They make that stuff up. Why would they say 4.7?

>>10281666
>>10281656
You really, really do not understand the underlying concepts of ml or how to follow lines of thought that aren't your own. Stay in school and try to learn some humility

>>10281754
Are you the quant on break? or CS/ML person?

>>10281790
>They make that stuff up.
glob darnit, this sounds so truthful
screw this homo market ;_;

>>10281815
post full of this faggot

>>10281663
I was excited until I saw it was Ron

>>10281815
markets don't blow up till the bullet is expanding inside the soft tissue. You can see the bullet traveling in a certain trajectory towards the heart and the market will continue upward. It's why bulls win out.

>>10281809
The second one
>>10281815
>>10281825
>screw this homo
>Ron Paul
/smg/ next movie night is Bruno

>>10281844
Cool. I've been thinking about making an AGI sometime this year. Going to probably do it using something related to AST and a new architecture.

>>10281834
This is a scary visual, nightmare senpai

>>10281825

>>10281751
>am have whored around so much in TA world, have like 100+ strategies collected (algos and everything)
>main reason am learning ML is because not have time to custom configure their optimal setting for every single stock / currency looking at. ML would be perfect for it!
People really don't understand just how low the accuracy is on so much TA shit.
They realize that they aren't sure things, but no really, most of those patterns only have like 55% accuracy.
Yet, the human mind has an innate capacity to seek out these patterns and assign meaning to pure white noise.
ML would be able to statistically hunt down these patterns and determine whether they are truly significant.

>(still mourning me old SPY strat config that accidentally deleted, this one was tested 107% profitable, so goood, but took me 3 weeks to configure) (pic related double itchimoku cloud strat)
That's rough.
I'm super paranoid about my laptop's hard drive getting corrupted, but I'm too lazy to back it up.
I'm pretty good about programming and data science practices for not losing my shit though.
>Also yes kek these guys have been bad lately, think the "Kill yourself" poster is a bot
>The thread lastnight was real treat, like 5 people described how they would kill me, was great
Have you ever received any noose pics? When I go trolling on /b/, I rack up a ton of those.
I like to post a pic of Ciara Horan / Eliza-chan and tell them to count. It's fucking hilarious how thirsty those fuckers are.

in honor it happening, heres a new indicator i just made.

i call it the JUSTening oscillator
the rainbow shit is just a bunch of emas on the VIX which is what im using to calculate the JUSTening oscillator. the real meat and potatoes is the "JUSTening oscillator" and the "niggerline"

the JUSTening oscillator, under normal circumstances stays above the niggerline (which is adjustable for anywhere on your screen. simply right click the indicator). when the JUSTening oscillator starts turning red, it indicates VIX spikes. if this increases in frequency, we can expect a short term trend change in VIX. a sustained break below the niggerline, and consistent red oscillator means its happening. when we see the JUSTening oscillator turn back around sharply while under the nigger line, you go long

------------
//@version=3
study('JUSTening oscillator', overlay=true)
nigger = input(defval=15, title='niggerline')
VIX = security('CBOE:VIX', period, ohlc4)
ma1 = ema(VIX, 100)
ma2 = ema(VIX, 110)
ma3 = ema(VIX, 125)
ma4 = ema(VIX, 140)
ma5 = ema(VIX, 155)
ma6 = ema(VIX, 170)
ma7 = ema(VIX, 185)
ma8 = ema(VIX, 200)
width = (ma8 - ma1)+VIX
price = VIX
wpd = (width - price)
mycond = price > width
plot(ma1, title='100', color=#00FDDD, linewidth=1, style=line)
plot(ma2, title='110', color=#00FD84, linewidth=1, style=line)
plot(ma3, title='125', color=#00FD2A, linewidth=1, style=line)
plot(ma4, title='140', color=#3BFD00, linewidth=1, style=line)
plot(ma5, title='155', color=#AEFD00, linewidth=1, style=line)
plot(ma6, title='170', color=#EEFD00, linewidth=1, style=line)
plot(ma7, title='185', color=#FDA600, linewidth=1, style=line)
plot(ma8, title='200', color=#FD2600, linewidth=1, style=line)
plot(price, title='priceline', color=#FFFFFF, linewidth=1, style=line, transp=50)
plot(wpd + nigger, title='JUSTening oscillator', color=(mycond ? #FD0000 : #08FD00), linewidth=3, style=line, transp=50)
hline(nigger, title='niggerline', color=#FFFFFF, linewidth=1, linestyle=dotted)

i've got money ready for tomorrow / this week.

if BAC tanks from earnings, i'll probably buy in on the back of buffet's confidence in them.

if BAC keeps doing well and NFLX keeps going down before earnings call, probably will buy in.

am i gambling? is NFLX the next AAPL or the next IQ? should i not be buying until i finish reading The Intelligent Investor?

>>10281880
and heres a more compact version thats not an overlay, w only the JUSTening oscillator

------------------------
//@version=3
study('JUSTening oscillator compact')
nigger = input(defval=15, title='niggerline')
VIX = security('CBOE:VIX', period, ohlc4)
ma1 = ema(VIX, 100)
ma2 = ema(VIX, 110)
ma3 = ema(VIX, 125)
ma4 = ema(VIX, 140)
ma5 = ema(VIX, 155)
ma6 = ema(VIX, 170)
ma7 = ema(VIX, 185)
ma8 = ema(VIX, 200)
width = (ma8 - ma1)+VIX
price = VIX
wpd = (width - price)
mycond = price > width
plot(wpd + nigger, title='JUSTening oscillator', color=(mycond ? #FD0000 : #08FD00), linewidth=3, style=line, transp=50)
hline(nigger, title='niggerline', color=#FFFFFF, linewidth=1, linestyle=dotted)

>>10281751
Its over for you
Enjoy this

>>10281891
You are making the correct decision

>>10281902
better version of the compact JUSTening oscillator

--------------------
//@version=3
study('JUSTening oscillator compact')
VIX = security('CBOE:VIX', period, ohlc4)
ma1 = ema(VIX, 100)
ma2 = ema(VIX, 110)
ma3 = ema(VIX, 125)
ma4 = ema(VIX, 140)
ma5 = ema(VIX, 155)
ma6 = ema(VIX, 170)
ma7 = ema(VIX, 185)
ma8 = ema(VIX, 200)
width = (ma8 - ma1)+VIX
price = VIX
wpd = (width - price)
mycond = price > width
plot(wpd, title='JUSTening oscillator', color=(mycond ? #FD0000 : #08FD00), linewidth=3, style=line, transp=50)
hline(0, title='niggerline', color=#FFFFFF, linewidth=1, linestyle=dotted)

>>10281877
REEEEEE IT'S A BUBBLE

>>10281754
>>They simulate human intuition
>NO! That's a fundamental misunderstanding of ml or psychology. Human intuition is neither logical nor rational nor quantitative. The fact that they both use pattern recognition does not make them equitable. Why do you think I keep talking about irrationality
They do simulate human intuition -- that's the innovation of ML.
Human intuition is flawed and imperfect, just like ML.
ML almost never has 100% accuracy in practice. Every algorithm ever written outside of AI works every fucking time. Perfectly.
ML does not.
Why would you put up with it? When there's no alternative.
ML is software finally bringing to the table that capacity for creative and non-algorithmic shit that's so hard to replicate -- human intuition.

>>No, you're mixing up ML with HFT
>No that's exactly what you're doing. I haven't brought up hft at all but you keep doing it. Stop with the fucking autism. If you don't understand how building a quant model is like doing ta then you don't understand at least one of those things
No, you're fucking retarded, it's exactly what you're describing.
Look here:
>>10280164
>with the added benefit of being able to act faster on action than a human could.
That's fucking HFT.
HFT = High Frequency Trading; reacting to the markets faster than humans
You described the fucking definition of HFT.
>with the added benefit of being able to act faster on action than a human could.
I'm not doing that.
I'm not doing HFT.
It's going to be a batch job that runs every day after market close and sets up limit orders for the following day.
Zero probing of L2 data ever involved.

does anyone have a concrete list of the signs of the 2008 bubble?

>>10281754
>>More data = more trials for the model to build its intuition off of. It's "intuitition" though, so you just need to make sure that it's high quality, clean, and correct. Any wrong answers will fuck it up.
>Ugh, this is such a bad explanation. You're halfway to right but the analogies are wrong and you don't have a deep grasp of how data training actually functions
You are 100% incorrect.
That's what epochs are when training NNs. It's a run of a grouping of the training set, typically about 1000 trials -- for every trial that returns with an incorrect prediction, backpropagation is performed.
>>The reason that artificial neural networks are picking up in popularity is that they are kind of tailored for and handle using lots of data very well
>What? No. Scale and throughout are not the advantages, and nns aren't "picking up" in popularity; they're commoditized. The advantage is being able to self reference and have individual neurons make small recursive changes. This allows for reaching decisions faster
>Scale and throughout
Right. More data.
>The advantage is being able to self reference
Recurrent neural networks aren't that popular.
They're cool, but deep convolutional networks like the ones for self-driving cars are what's picking up and the key innovation from shit like Alex-net that's pushing forward the technology behind self-driving cars.
>If you hadn't face posted I'd swear you were a chink because you can't into English reading comprehension and you're stealing ideas without understanding the basic underlying concepts
Idk, bro, your track record has been pretty shit so far this thread. Have you done much reading beyond Medium?

>>10281793
You're a moron that probably can't even code.
I've been doing this ML shit for like 3 years.

>>10281809
Neither.
He's a faggot LARPer that keeps confusing ML with HFT or constraining it to reinforcement learning

>>10282017
just watch bank ERs this season. if they keep doing what they have been doing its ogre

Dow will fall 2000 points tomorrow, you will remember the great crash of 7/16 for the rest of your life

>>10282054
You were asking for my opinion on this two weeks ago. I'm the tech executive and I have far more experience at this than you do, "I took a couple 500 level courses"-kouhai. Go look over your last couple posts when you stop being a buttmad child and realize how many times you're claiming the literal exact opposite of what's plainly written. Career pro tip: communication is the most important skill despite cs being the profession of choice for autists. Your inputs and outputs are both garbage

>>10282089
dont tease me lad

but i think ur right
i can feel my VIX calls vibrating from my brokerage account

>>10282055
Tempted to say fuck it and go all in SQQQ
>>10282089
You have me rock hard, baby

>>10282017
* treasury yield curve inversion (none of that other faggot shit that ZH keeps pointing at! Specifically the inversion of the yield for 2 year, 10 year bonds)
* banks missing earnings
* buying shit to sell it later -- greater fool theory taking over
* bank "liquidity" issues

Banks passed their stress tests, except for the investment ones, which rely on degenerate gambling.

>>10282055
>just watch bank ERs this season. if they keep doing what they have been doing its ogre
>if they keep doing what they have been doing
Which is what?
They haven't fucked up earnings projections, and they are passing their stress tests

>>10282105
should give me a few million to set up an underground fortress for AGI.

500 ALT shares here
Praying that I can actually win for once

>>10282126
>They haven't fucked up earnings projections, and they are passing their stress tests
sure hasnt made a lick of fucking difference for their stock price

im rooting for em, cuz i dont wanna see nikkei financially destroyed, but i also dont have any faith in them either

>>10282017
The housing market

>>10282111
trips say fuck everything

>>10282105
>I'm the tech executive
I know you are

>Go look over your last couple posts
I'm having fun, but I will

>and realize how many times you're claiming the literal exact opposite of what's plainly written
I'll start counting

>communication is the most important skill
I agree

>>10282173
HV on SPY is low and a wave is coming. Kek has confirmed a happening

>>10281857
That sounds like a lot of fun. I prefer working with semantics and grammars over raw numerics
>>10282135
If I were going to invest in an underground fortress it'd be to rent out to spergs from /k/
>>10282111
Desu

>>10282225
I try to stay original and hobbyist. Only done a bit of lex/parsing things. I should get into ML but of course I'd do only do it from a unique angle.

I don't think AGI is that hard not compared to something like P=NP

>>10282242
Wtf is AGI?

>>10282257
Something that can abstractly reason

>>10281663
>"This didn't need to happen."

>>10282299
Dude, I've only seen that done in factorial runtime, but that shit's great

made a new more compact 5min JUSTening oscillator. i think im getting the hang of the basics of this shit.should i just stop bothering w this babby-tier pine editor shit for tradingview and start learning python?

---------------------------
//@version=3
study('JUSTening oscillator compact')
VIX = security('CBOE:VIX', period, ohlc4)
ma1 = ema(VIX, 20)
ma8 = ema(VIX, 100)
width = (ma8 - ma1)+VIX
price = VIX
wpd = (width - price)
mycond = price > width
plot(wpd, title='JUSTening oscillator', color=(mycond ? #FD0000 : #08FD00), linewidth=1, style=histogram, histbase=0.0)
hline(0, title='niggerline', color=#FFFFFF, linewidth=1, linestyle=dotted)
--------------------------------------------

>>10282225
>tech exec bearish on tech
well good to know im not just being fucking retarded

Good morning /smg/.
Did you rest well during the week-end?
Is your lawn mowed? Did you remember to buy enough Monster Energy Drink™ for the upcoming week?

>>10282411
>Did you remember to buy enough Monster Energy Drink™ for the upcoming week?
i didnt even remember to eat desu senpai

>>10282411
rockstar

>>10282411
/SIP/ HAS MADE IT INTO /SMG/
WHERE MY SIPBOIS AT

Bubble indicators:

2 yr vs 10 year inversion
almost equal
PE
24. It was 45 in 2001/2008
MCAP vs gdp
All time highs of 1.4
Tobins Q
1.1, vs mean 0.7. inflated
Jobless claims
All time lows, good indicator
Monthly supply of Homes
normal

Seems we could see a correction because of inflated market cap vs gdp, but we don't have the precursors for a recession until we see home supply rising, unemployment rising, and earnings start to tumble.

>>10282319
is it just me or does gerome powell look a lot like ron paul?
do you think powell is some sort of metaphysical aberration/projection of ron paul from another timeline? what does it all mean?

>>10282460
>Seems we could see a correction because of inflated market cap vs gdp
thats all i really think is coming desu
itll seem pretty crazy in comparison to prior crashes, as all the stocks are much higher priced, but i think were just gonna get a big dip. the real reckoning is still a ways out

or maybe ER season will save us
idk, doubt it tho

>>10282089
Larp?

>>10281177
Yeah, hopefully 5 completes tonight and we open in A, or better a C.

>>10282242
Agreed, and that's a great way to keep things interesting.
>>10282384
The sqqq buy was at close on Friday and it's integer as a very short term position but yeah. I think there's a large tech bubble that'll pop in the relatively near future. Social media and data, combined with the fact that the talent market's average has been consistently going down for a while (particularly at the giants). Read up on the Facebook ipo; look at the outrageous speculation driving three prices of tsla, Amazon, Netflix; investigate the way Google is deeply on bed with the government; think about the implications of the ruling that Twitter is a public forum

nikkei

>>10282411
>>10282448
>>10282433
/S I P B O Y S 2 0 1 8/
Two zero ultras a day keeps the faggots away

>>10282460
JUST FUCKING DIE

>>10282509
The Nikki is closed you fap

>>10282551

It's opening soon

70 minutes till everyone sees whatever number China decided to put out for GDP

>>10282509
I just realized the nikkei is price weighted
Looks like it's shit, just like the DOW

>>10282563
holiday

>>10282026
>>10281754
triggering my autism
>>>More data = more trials for the model to build its intuition off of. It's "intuition" though, so you just need to make sure that it's high quality, clean, and correct. Any wrong answers will fuck it up.

An SVM does not simulate human intuition. An SVM finds the maximum margin hyperplane between two sets of data in an implicit feature space. Nearly all ML methods work by iteratively minimizing a cost function/functional. That's it. Don't get hurt because you trust the output of your model; always be skeptical. It is simply trying to separate the data as much as possible with the dimensions you have provided it.

Neural Networks are picking up in popularity because people discovered how to efficiently run them on GPU hardware. Everything else(adding reinforcement learning, transfer learning, deep learning, etc) that has followed is because of that and are relatively minor theoretical advancements. Recurrent neural networks are not popular because they are much more difficult to understand and train. CNNs are popular because they are good at visual tasks and can be easily trained. Deep Learning is popular because we have a lot of hardware and data to throw at problems.

>>10282566
checked
rolling for 6%
they prolly think were dumb enough to buy that

>>10282503
>the fact that the talent market's average has been consistently going down for a while (particularly at the giants)
you saying the tech field has been affirmative actioned to death? i tend to agree, but im an outsider
>look at the outrageous speculation driving three prices of tsla, Amazon, Netflix
ive been posting FUD about TSLA since their bond yield blowout. fucking Caa1. elons gonna lose a lotta peoples money
>investigate the way Google is deeply on bed with the government
way ahead of ya senpai
>think about the implications of the ruling that Twitter is a public forum
didnt hear this. whats that all about?

you're all a bunch of god dman niggers fuck shit ass

>>10281663

I actually do agree with the skeptics that the market is overinflated and that all the personal, corporate, and federal debt floating around has the potential to turn a recession into a depression. That being said, Ron Paul is chicken little and has been for years. And fuck the gold standard.

And while we're discussing ML, I've found random forests to be my favorite method. It isn't always the best method in the end but damned if it doesn't always do a good job when I don't know what else to do.

>>10282646
why tho?

>>10282384
>>tech exec bearish on tech
>well good to know im not just being fucking retarded
Oh dude, yeah. The only tech company that I'm holding is AAPL.
All those IPOs are fucked though.

Check this out:
>https://www.amazon.com/Disrupted-My-Misadventure-Start-Up-Bubble-ebook/dp/B013CATZIC#productDescription_secondary_view_div_1531520680856
>HubSpotters were true believers: They were making the world a better place ... by selling email spam.
>Dan's absentee boss sent cryptic emails about employees who had "graduated" (read: been fired).
>and where everybody is trying to hang on just long enough to reach an IPO and cash out.
>Disruptedis a gripping and definitive account of life in the (second) tech bubble.
WSJ said there hasn't been a better time to IPO since 2000 or 2001
(Some of the stuff I quoted was just me roasting the start-up community -- San Francisco will turn into a fucking cesspit like Detroit when the next recession hits)

>>10282500
>Yeah, hopefully 5 completes tonight and we open in A, or better a C.
>5 completes
>open in A
>better a C
What the fuck are you describing here?

>>10282460
>Monthly supply of Homes
>we don't have the precursors for a recession until we see home supply rising, unemployment rising, and earnings start to tumble.
I heard the housing market has peaked and that it's related to supply and demand, unlike the last peak, which was fueled by speculation

>>10282663
cuz im fuking pissed and im drunk

Why is everyone on /biz/ so bullish all of the sudden? Did I miss some U.S. news or something

>>10282699
>I heard the housing market has peaked
Possibly. This time we might see a collapse in demand, rather than an increase in supply. (if that makes any sense)
>>10282711
it's easier to make money in a bull market so no one wants it to end

>>10282699
I see a housing market crash before tech crash. Tech as in not social media shit though.

>>10282699
>What the fuck are you describing here?
Wave theory TA.

>>10282601
>An SVM does not simulate human intuition.
No, of course it doesn't actually simulate the neurology of the human fucking brain.
Machine learning "simulates human intuition" in that it similarly hunts for patterns through lots of trial and error.
That's how humans learn basic addition -- you do hundreds of fucking math problems until your brain just sees the patterns to it.
>An SVM finds the maximum margin hyperplane between two sets of data in an implicit feature space. Nearly all ML methods work by iteratively minimizing a cost function/functional. That's it.
Yeah, I know about the hyperplane and kernel trick, I had to do that shit by hand for a final.
>Don't get hurt because you trust the output of your model; always be skeptical. It is simply trying to separate the data as much as possible with the dimensions you have provided it.
Of course I won't trust the output.
ML is not 100% accurate. It's one of the few areas of computer science where you don't "know" what the output is going to be.

>>10282711
OVER CONFIDENT COMPLACENT FAGGOTS NEED TO BE LEARN REALITY OF THE MARKET NHJDAFSBHJFDSABHUJFDBSAHUJFBDSAHJFBDSHJAFBHJDSAFBHJSA5

>>10282646
I can agree with that, senpai

>>10282814
at least im not the only fucking retarded faggot holding puts rn

>>10282642
>you saying the tech field has been affirmative actioned to death?
Not to death yet but yes. The big tech meccas are disgusting hotbeds of this agenda, and the industry giants perpetuate it. The other half of the equation is that 95% of the market is basic lob crud apps and generating those is a highly fungible commodity now. Coding isn't science, it's a trade. The general rule of thumb in industry is 80% is good enough, so paying half the price for kids fresh out of school, pajeets overseas, or schmucks that did an 8 week bootcamp is usually what happens. Of course, the quality is dogshit but you'd be amazed how far a business can go on crap.
>whats that all about?
Federal judge ruled Trump can't block Twitter users because his tweets constitute a "public forum". That's a first amendment term of art so this has serious implications for how courts may come to view social media and the way companies apply content filtering algorithms. It's a slow brew but, especially combined with eu's article 13, it could spell serious trouble for tech giants
>>10282711
Because rkg and uvxy got btfo last week

>>10282657
>And while we're discussing ML, I've found random forests to be my favorite method. It isn't always the best method in the end but damned if it doesn't always do a good job when I don't know what else to do.
Random forests are the shit.
I've been considering starting with them instead of SVMs. They're fucking great. It's fucking plug & play; no tuning.

>>10282763
That would be incredibly disappointing.
I really want to see San Francisco turn into the next Detroit, and all those homeless fucks over there overrun the place and start eating the smug pieces of shit that let them in.

I heard the next housing market collapse won't be even close to as bad as 2008 though.

>>10282827
I don't even want to buy calls anymore. Theta decay and IV always fucks me.
Fucking puts on SPY though. I'm triggered.

>>10282703
im sorry
plz dont hit me again daddy

>>10282827
pretty sure a lot of us are near term bearish desu senpai

>>10282892
>Of course, the quality is dogshit but you'd be amazed how far a business can go on crap.
makes sense tho
most of these companies successes comes from brand recognition, and their sheer size making serious competition virtually impossible

very interesting about the TWTR thing tho. ill keep it in mind. plus their puts are a lot cheaper than some of the other big tech names

one thing ive been wondering is how the repeal of net neutrality will start to effect these big tech companies ability to make money. do you think itll be a big deal, or negligible?

Does /smg/ dab every time they make 1 dollar?

>>10278916
Does Trading View read options

>>10282956
>most of these companies successes comes from brand recognition, and their sheer size making serious competition virtually impossible
Yes, but not just the big boys. Small companies fall prey to this too because they're more cost sensitive and less likely on average to have someone with the experience to warn against this. I've begged unsuccessfully more than once to hire half the heads at 80% of budget for 3x the experience but incompetent leadership often cannot into math and thinks more bodies means faster work (a clear indicator of code as trade rather than science)
>net neutrality
I think this is overblown, especially for the giants. They control a staggering amount of market share (traffic) and demand for their product is relatively inelastic. Worst case scenario it'll be like $90 oil: people will bitch but pay the extra and continue as usual. I don't think nn will have a serious effect on much of anything, though it is curious to note that this political decision's bent runs counter to the Trump tweet one (looser vs stricter regulation)

I found a really cool etf, look at CLIX. It's long ecommerce and short retail, Im going to check out it's holdings

>>10282956
starting to feel better now

>>10283116
ah-bloo-bloo

>>10283077
>Does Trading View read options
how do you mean? like is there an options chain? no they do not have options to my knowledge. dunno tho as i only fuck around on the free version for charts on things my broker doesnt have

>>10283111
Thanks just bought 100k

>>10283201
thx just sold 200k

China shook the 9 ball and got 6.7% hitting estimates perfectly.

Consumer and industrial also exactly hit estimates.

>>10283224
Thanks just looked at it and did nothing with 300k

>>10282384
Do you know how to do this in think-script?

>>10283240
thx just murdered 400k koala bears

>>10283078
> Small companies fall prey to this too because they're more cost sensitive and less likely on average to have someone with the experience to warn against this
those that are aware of it are probably afraid to speak up about it too, for fear of losing their job

>net neutrality
>I think this is overblown, especially for the giants
i guess well see, but there was a MASSIVE disinfo campaign that was being pushed on all channels leading up to it. they had everyone convinced that it was the "end of teh innernette". if you spoke out against it, or pointed out that NN was an obama era piece of legislation, you were shouted down by an army of shills and useful idiots.

dunno if they would have spent all that money for an astroturfing campaign of that magnitude if it was unimportant. youre probably right tho, that the companies are gonna bounce the cost onto their customers, and that people will still buy it regardless.

i was just interested to see if we get a shock/hiccup this quarter for their ERs as a result of them having to change their business models somewhat. im really excited to see how this week pans out. i feel its pretty make or break rn

>>10283255
ive literally never written a line of code in my whole life until friday afternoon.
so no

How hard will TSLA drop monday morning?

>>10283291
is their ER before the bell?

>>10283232
Chinks claim that their economy behaves exactly the way everyone thought it would? I'm shocked
>>10283240
I looked at holdings and man I really like the thesis but not their longs
>>10283271
I'm familiar with the hullabaloo and agree with you that it's a giant nothing
>see if we get a shock/hiccup this quarter for their ERs as a result of them having to change their business models somewhat
Why would you expect this? No business models have changed as far as I know - no isps have released the scary new content pricing structures everyone was worried about - so there's no news to swing earnings

>>10283240
i see your rarearnold, and raise with a rarerarnold

>>10283232
theyre good guessers

>>10283334
dunno
im just some fucking dude. figured id ask since you seem to know the field

>>10283335
Call your rareArnold
>>10283323
Go read twatter about musk
>>10283291
>3dpd
REEEEEEEEEEE
>>10283357
No news yet, it ain't happening until it's happening, but it's a fair question

>>10283419
thats fucking niche

If ALT moons, what’s a realistic exit position?

>>10282586

Fucking hell I'm a retard. I live in Japan and I don't even know when it's bank holiday.
I really wish I read this sooner, I wanted to go to the pool before the markets open, pool was crowded as fuck, had to give up on swimming midway

chinese market down on good GDP numbers, wonder why.

>>10283480
Why?

>>10283419
>Go read twatter about musk
desu it sounds like another one of those PR stunts where he gets retail to short, writes puts, then puts the squeeze on em

been saying for a while all these FUD and subsequent, well timed PR moves resulting in short squeezes, are the only thing keeping that stock up rn

but christ. what an insufferable faggot that guy is. he acts like a fucking child

>>10283484
cuz those good numbers are fake and gay

>>10283469
This senpai

Been trying to figure if I should put in a limit sell.

My biggest problems in trading are bad fomo on entry and leaving too much on the table

>>10283335
I save your rare arnold and raise you my second rare arnold

Also please note: If the market doesn't crash tomorrow I'm going 100% JNUG. 100% JNUG until the crash. Nothing can stop me. J N U G

>>10283568
>that five year
>>10283490
Fucking chinks

>>10283469
>>10283566
PROTIP: decide your own goddamn exit point and dont post it here. especially not on a fucking penny stock

>>10283469
Anything positive

HWAT DO I MAKE MONEY ON AAAA

>>10283585
Edit: Calls on IAU, not JNUG

>>10283624
ALT
Buy now cash out in August

>>10283635
Faggot

>>10283631
does IAU even have the penny program? why not just trade GLD?

>>10283667
This is a better idea

>>10283631
Should have bought one year ago
t. your best fan

>>10283698
But it's only up .12% YTD

>>10283728
hold on, I gotta watch The ChartGuys indexes videos

>>10279441
>>10281201
>>10283469
>>10282146
>>10283566

why do any of you expect ALT to go up? is this new director supposed to turn shit around? his time at immunogen had a weird mooning period but came right back down. is that what you're praying for again?

Testing a tripcode I really have no need to use

Anyway, does anyone know if you can have money in multiple currencies in Degiro? Like, can I have EUR and JPY or will the site automatically convert money to the currency of the country I used to register?

>>10283798
It's at a 52-week low and the technicals look good for a break out.

>>10283823
It was at a 52 week low last week

>>10283823
>>10283851
but aren't a lot of stocks at all time lows? a lot of stocks are just dying.

I mean, get on the ALT boat now or miss it.

>>10283851>>10283823
and at last month

>>10283490
>desu it sounds like another one of those PR stunts where he gets retail to short, writes puts, then puts the squeeze on em

seems to be holding the JRE interview for near Q2 report.

>>10282225
GBR was very good to me last week, all in at 3.21 thursday limit sold at 3.62. have no idea why it went that high on friday morning (went back sub 3.30 in 4 min) nor do i care.

xiaomi btfo

China changes rules to fuck over Hong Kong listed stocks and stop mainland investors from investing in non mainland stocks

https://www.bloombergquint.com/china/2018/07/15/china-bars-foreign-companies-other-securities-from-stock-link

The HK exchange was given 1 hour warning before the massive switch was made.

(Bloomberg) -- China’s stock exchanges said they won’t allow mainland investors to buy shares with weighted-voting rights in Hong Kong, sending Xiaomi Corp. shares slumping.

The bourses will also bar trading in foreign companies and stapled securities via the Hong Kong stock link, according to a statement by the Shanghai Stock Exchange on Saturday, which said many investors don’t understand the risks associated with new products. Xiaomi tumbled as much as 9.6 percent before paring declines.

The move comes before the three classes of shares are included in the city’s Hang Seng Composite Index in the third quarter, which would otherwise have made them eligible for the stock connect.

>>10284047

This kills the IQ

>>10284093
>>10284047
https://www.youtube.com/watch?v=CmZAyMx7vjQ

>>10284093
isn't effected

>>10284099
Dank

>>10284099
you're not planes of zim and planes are you?

>>10284183
no

>>10281921
WHEN WILL BANG COME TO THE US STOCK MARKET AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA

>https://www.politico.com/story/2018/07/15/china-trade-war-trump-722242
I really want Trump to just go for Xi Jinping's fucking throat, even if it hurts a little bit.
Those fuckers have what's coming to them. They've deserved it for a full 8 years under the limp-wristed Oh-bummer administration.
Exciting times, frens.

>>10284025
im holding some. i think oils gonna pop up again. it seems to like it when oil is running

>>10284465
>>10284025
I'm still holding some GBR too but i'm not buying more until it dips down below $3 maybe even down to $2 it'll take patience.. but it's one of those ones that you cant just start buying when it starts going up you have to catch the falling knife because if you try to buy on the way up you will up holding bags since it moves so fast.

Well boys we doing tvix today?

>>10284579
Too expensive for my taste. When it gets bumpy I'll probably go all in on UVXY.

>>10284588

I've been watching UVXY for a long time but know very little about how to trade it. Can anyone give me a quick rundown or post some links for a UVXY noon?

Recommend me positive earnings penny stock, please.

>>10284588
>too expensive
were at yearly lows!

Anybody watching dry shippers? BDI is up 100% yoy and is still climbing, if BDI breaks 1,720 it should see 2,300 fast. I'm in NMM due to their financing isn't as dilutive as other shippers plus they have a 30% equity stake in Navios containers which is about to go public (if it hasn't already) does anybody have any shippers to watch? preferrably ones that aren't loaded with toxic financing

Just put in a sell order for S&P etf. Screw this the bubble is going to pop soon

Anyone got recommendations on what to buy with the 8 EUR I have left? "Absolute madman" kind of stocks are fine too obviously.

HMNY maybe?

>>10284652
I am sempai! Navios containers is already public trading OTC, I'm still trying to understand the relationship between NM, NMM and NMCI, it's a bizarre partnership desu
My Index of good dry shippers is
NM
SBLK
CMRE
GNK
SSW
My friend keep shilling me SB, but I don't like it for some reason

guise
check it out
i fixed the chaikin money flow. normally it has problems with gaps up and down. if theres too many gaps in price/direction it can get out of sync w itself. so i decided to add some inputs. now you can adjust how many periods you want to average out the variables with.

--------------------------------------------------
//@version=3
study("shekel flow")
nigger = input(defval=16, title='period length', minval=5, maxval=42)
fag = input(defval=2, title='gap adjust period', minval=1, maxval=10)
clo = sum(close, fag)
lo = sum(low, fag)
hi = sum(high, fag)
clov = clo / fag
lov = lo / fag
hiv = hi / fag
clow = clov - lov
hicx = hiv - clov
hlox = hiv - lov
clic = clow - hicx
mfm = clic / hlox
vol = sum(volume, fag)
volk = vol / fag
mfv = mfm * volk
mfmx = sum(mfm, nigger)
mfvx = sum(volk, nigger)
sf = mfmx / mfvx
sfx = sf * 1000000
cond = sfx > 0
plot(sfx, title="shekel flow", color=(cond ? #22FD00 : #FD0000), style=histogram, linewidth=2)
--------------------------------------------------

this stuffs not that hard desu senpai

give me more retarded meme indicator ideas to try and make senpaitachi

>>10284775
Oh my bad, it's not trading yet, but it should start 2019?

>>10284093
lol no

>>10284781
What is this shit

>>10285026
its the chad shekel flow.
the answer to the virgin, beta bitchboy chaikin money flow.
do you know which way your shekels are flowin?
chad does.

>>10284921
WHY DO I NOT HAVE A DAY TRADE I THOUGHT I WAS SO CLEVER FOR SELLING AT 35 AND WAS GOING TO REBUY BEFORE EARNINGS

NOW IM GOING TO REBUY AT 42 IF THAT

REEEEEEEEE

>start with 1000 dollars
>buy stock before ex div, sell on ex day for stock with ex div day after
>do this 1000 stocks
>Gain dividends for 1 million dollars, or up to 5000% per year

AJ FILES

Are the Investopedia courses worth buying? My faggot-sense is tingling and telling me "no".
199 USD is a whole fucking lot

>>10285147
stock drops by the div amount bud

>>10285238
thats why you have to pick the ones that go back up immediately

>>10285225
This thread will really warping my sexuality towards 2D women

>>10285272
namefag

>>10285272

>>10285272

Buy anime, sell manufactory

>>10285351
Nothing left to the imagination, how mundane.

>>10285372
nigga I'll fite you.
Buy HMNY

>>10285411
Telling someone to buy HMNY is akin to cursing them, but don't worry I got funbags.

>they still havent bought STZ

>>10285411
How many victims has HMNY claimed thus far? I feel like there should be a montage of all the bagholders with Adagio for Strings playing in the background.

Give me a quick rundown analysis on LON: GAW and FRA: G7W.
I'll pay you in 2D girls

>>10285510
The people who do warhammer? Ask again tmr night if you don't get a tldr by then. I gotta crash so I can get up and go work in a few hours.

FRESHLY BAKED
>>10285571
>>10285571
>>10285571
>>10285571
BREAD

>>10285565

Will do

>>10285478
>Adagio for Strings
fuck dat
needs something less melodramatic, and more hopeless, confusing, and with too many key changes
https://www.youtube.com/watch?v=vWY5bVPTyLA