thcsa

>There is a law against asking people who interact with the state for data that it already has.

I find that very interesting. Could you provide a link?

I once saw a pdf which listed all the third parties that PayPal shares its information with. It was more than 30 pages.

Could you try to articulate that in a simpler way? Without proper punctuation this is hard to follow. Thank you.

It's not the companies (and their earnings) who will pay back investors. Other bagholders will do that once they go public.

Wrong. I shoot the photo, and can afterwards (or whenever I stumble upon the photo in my gallery) adjust the depth of field with a slider. Pretty similar to what they did in the presentation. I can even change the subject of focus.

That's a smartphone.

The Vivo Nex can do this too btw. It's a cool feature though.

They presented the dynamic depth control with the words: "No other smartphone right now can do that".

Lie. I'm writing this comment on a Vivo Nex and I have been using this for months.

>High amount of intelligence has a negative corelation with creativity.

This has not been proven and I don't think this is the case. Here is a great post that highlights the faulty methodology of most studies that came to this conclusion.

https://news.ycombinator.com/item?id=5088314

Trading the right instruments is crucial. Pick highly leveraged instruments with low competition and liquidity. Trading big instruments like USDEUR is pointless unless 1) you have the right infrastructure or 2) you are a genius, since competition is incredibly high there. But there are small instruments out there where I (not a genius) easily make a 1500% annual return on $10k. However, give me $100m and I'll make less than 0.25% a year, because liquidity is low and orderbooks are thin.

Considering the trading system itself, I think what makes my system pretty neat is that I can backtest on Level 2 data. Level 2 data basically means information about the order book (and not just the top bid/ask). I collected L2 data for specific instruments for months. So if someone posted a limit buy order on instrument X with amount Y 0.54% below the top bid on October 15th, 21:01:13.746Z and canceled it 4287 milliseconds later - I have that information. Now the cool thing is that I can 'replay' everything, which basically allows me to know how the structure of the whole order book was at any time. This allows you to do pretty accurate backtesting, since you know the exact execution price of your market orders (market orders = orders that take liquidity from the book). Coding that was probably the hardest challenge, since performance matters for backtesting (for trading, it depends on the instrument and the strategy - for me, it doesn't matter a lot). You need to have a good design. Gladly, Kotlin makes concurrency a breeze with coroutines, so I basically have one routine for each instrument (which has to process order book messages sequentially for obvious reasons).

I am currently working on a machine learning approach to detect buy and sell pressure based on changes in the order book structure. Order placement (depending on the instrument) involves a lot of 'playing games', so people bait, people spoof, people try to push the orderbook down or up etc. I naively threw some pre-engineered features on a couple of LSTM layers, but results were not outstanding. But I think that this approach has some potential if executed properly.

EDIT: Obviously this is not my first coding project. My first 'project' was about autocompleting words. I used a trie structure for that, every node (letter) had a treemap of its child nodes, sorted by the occurences. So in the end I was reading a huge data set into my program, which had millions of hashmaps and other objects floating around, eating up 9 gigs of RAM :D. This was on desktop for obvious reasons, I never had plans to code a app. Fun times.

I started coding when I was 28. Knew nothing about it, didn't even know what a compiler was. I'm 30 now and self employed, built a profitable algorithmic trading system in Kotlin. There are so many free quality resources to learn programming, it's ridiculous. And all the software you need is free, OS, IDE etc... This is pretty much unique to the field of software engineering. Try to become a doctor by yourself, good luck.

Tips to start: The most important part is to just start coding as quickly as possible and learn as you go. A personal project can really help to maximize motivation and actual hours spent coding. As for the language, I don't think it matters what you pick and everybody suggests something different. I didn't like coding in python at all. Personally, I prefer languages with static typing, especially when you start out (so the IDE can help you). For me, that perfect beginner language was Java, since it's simple, structured and you can completely forget about memory management.

Just don't spend too much time thinking whether you should start or not. I wasted years with that mindset. Just do it.

EDIT: Sorry I didn't really answer your questions, I just don't know enough about web development to help you in that regard. Consider this a motivational comment that gets you started.

Austria also has a very good social safety net, free universities and subsidized childcare - and we have very few startups, because people just don't like to take risks here.

For ARM, always use Oracle VM. It's around 15x faster for my application.

Comparing Kahnemann's work with Newton's is absurd.