kensoh

Thanks OP, I really like this part of the whole article - 'It also uses an API interface that lets you use CatBoost from the command line or via API for Python or R'

Would some form of computer-aided tool be able to help the programmer analyze GPs? It sounds like there's some repetitive process when you say 'GPs with a lot of patience'. I was thinking along the line that if such a tool is possible, then it might be possible to let the development and evolution of GPs iterate faster than deep learning. I found this thread really interesting, because of the small size, such implementations can be easily done locally without expensive and huge hardware setup. I'm definitely amazed at what DL has achieved so far. But it looks to me like a brute force way to solve a problem by gathering huge amounts of data, crunching on huge amounts of hardware, to solve a very specific classification problem. Not saying that is not good, just saying it is hard for mass participation at a high production quality level due to lack of hardware and stuffs.

For the branch of trading technical analysis which assumes that all information is factored into price and price action, the variables will drastically be reduced probably to something like volume, low, high, open, close etc. That makes it ideal to develop algos. But as the fundamentalists would say, assuming that everything (demand from insider news etc) has been factored into price is an overly simplistic way of looking at markets.

I see.. Thanks for sharing. I assume you mean evaluating game board states, finding valuations objectively and figuring out the pattern of where to move that can lead to higher probability of winning? 20 years ago I made an Othello game that use a search tree (I think configurable from 3 to 9 levels deep), assign valuations to different board positions and let the computer assume that the player would make the best move to his advantage. It turns out can beat human players easily, but that is such a small search space and valuation weights are very clear. Thus I was amazed at the computer winning with such a large game board where even for professional players, it is arguable which state is a better state. EDIT - ok I just googled monte carlo tree search, I must have accidentally implemented that search at that time, base on common sense and what I can do with the programming langauge I had at that time (Visual Basic).

Thanks for sharing your experience. I'm happy that my previous exposure to trading algorithms at least helped me understand more what the experts here are talking about. I believe the output model is only as good as the data (at least for the deep learning branch of ML). If the dataset does not cover data-points which exist in a wider space but in the same domain of the problem, or which haven't yet have a precedent, then we really can't simply assume that it is the algo/model that needs tweaking when shit hits the fan.

I see.. That's true. Though credit still goes to the algo for choosing that particular weird move out of the entire search space (it's just 'weird' and something you will think is a move made by a total newbie to the game). I remembered for that whole week during lunchtime I would watch the broadcast live on YouTube. How devastated I was to see Lee Sedol losing match after match. It was a moment I would never forget, in my mind the computer had crossed an imaginary threshold and it won. I know ML/DL experts will say it is only for a very specific area. But what's stopping more mastery of enough 'specific' areas that the mastery will be broad enough to pass Turing tests?

I don't have ML or deep learning background (no Masters or PhD), adding comment from experience with backtesting trading systems. We will collect market data and design algorithms that seem to produce the kind of outcomes we want. Then test on some other data sets which the algorithms have never been applied on. Many iterations later, you can get a decent profitable algorithm. And if the 'holy grail' algo is run in market long enough, eventually there will be severe drawdown and going bust. The quality of the algo and I assume the deep learning model lies in the quality (breadth and depth) of the data, and how honest with himself the person choose to model it. There will be time and again new 'black swan' or edge events happening (remember LTCM), because using machine learning is like using the past to predict the future.

I guess as long as the users' expectations are correct it can be useful in some very specific areas. Referencing the AlphaGo game last year, I was a Go player for more than a decade. But yet AlphaGo's weird move inspires new insights that break the conventional structure / thinking-framework of a Go player. From that angle, I do think that even though DL is somewhat a blackbox, humans can pick up new insights because it explores areas which are normally ridiculous to a human with 'common sense' to explore.

I don't have ML or deep learning background, but I second what you said base on my experience with backtesting trading systems. We will collect market data and design algorithms that seem to produce the kind of outcomes we want. Then test on some other data sets which the algorithms have never been applied on. Many iterations later, you can get a decent profitable algorithm. And if the 'holy grail' algo is run in market long enough, eventually there will be severe drawdown and going bust. The quality of the algo and I assume the deep learning model lies in the quality (breadth and depth) of the data, and how honest with himself the person choose to model it. There will be time and again new 'black swan' or edge events happening (remember LTCM), because using machine learning is like using the past to predict the future.

I guess as long as the users' expectations are correct it can be useful in some very specific areas. Referencing the AlphaGo game last year, I was a Go player for more than a decade. But yet AlphaGo's weird move inspires new insights that break the conventional structure / thinking-framework of a Go player. From that angle, I do think that even though DL is somewhat a blackbox, humans can pick up new insights because it explores areas which are normally ridiculous to a human with 'common sense' to explore.

I believe there is also merit in spending a lot of time repeatedly doing something, before proceeding to automate that workflow. Because only through that, you gain a deep understanding of 'edge/exception' cases which you can directly code into your script later on to manage.

Cool! Maybe you'll enjoy something such as Sikuli. It basically lets you automate the actions to take on your desktop or laptop visually. It's based on OpenCV and has a simple interface to use.

Have fun! This one is a trend-following system for swing/position trading. It hasn't been backtested to factor new market behavior for the last 6-7 years, so it'll probably do badly now. My failure was not being able to convert that for daytrading timeframe.

PS: anyway, sorry for digressing away from the topic of the HN thread lol

The trading domain knowledge can be found from various online resources. I used to make a 200+ slide powerpoint on the technical analysis aspect of trading but it is probably irrelevant now as markets had changed much since the introduction of algorithmic and high-frequency trading.

I was writing that script for Amibroker platform (https://www.amibroker.com) but I believe the market leader is probably something like MetaTrader (https://www.metatrader4.com). Also, for developers or software engineers familiar with Python, Quantopian might be something you'll enjoy using (https://www.quantopian.com).

I was kinda like a backup freak for my PC, but once I got starting using cloud drives like google drive, it kinda takes care of the problems of hard-disk failures (which are definitely bound to happen every couple/few years).

I used to write trading systems for Amibroker. This one signaled the 2009 march bull market in equities and oil. But it is a swing/longterm trading system. I was a daytrader so it is not really useful for that.

https://github.com/tebelorg/Tump/blob/master/trading_z.js

In theory they shouldn't, though I believe many financial websites explicitly declare in their terms of use that automated access or web scraping data is not allowed.

Same sentiments. In the beginning, the user and developer of the PC is the same person. I just feel that there are so much untapped computing power waiting for people to tell the computers to go do intelligent things.

I automate as much as possible the tasks involved in coding web automation scripts - https://github.com/tebelorg/TagUI

Basically use a web automation tool to simulate the login process and then grab the data you want, save it to .csv file or something. For saved HTML pages, web automation tools generally can work with them. For saved PDF I'm not sure what is good.

I did the same lol (via scraping), run every midnight to multiple bank accounts, then sync to google spreadsheet. I was writing web automation scripts the hard way for a while, until I made a tool to simplify the process of writing web automation scripts - https://github.com/tebelorg/TagUI (frankly I am surprised why my banks never banned or send me a letter)

and this - https://github.com/ariya/phantomjs/pull/15066

I did the same too, unfollowed all friends and pages. Then from time to time when I miss a friend I go check out his page for updates.

This is re-posted from a Facebook post a year ago. Universal basic income might be stating the obvious for some of you, just wanted to start a conversation going to learn other people's thoughts on the topic.

As the voice of universal basic income gets louder, it gives me more courage to post something like that. Otherwise, in my part of the world, I would sound like a mad person to even talk about the idea.

Otherwise, in my part of the world, I would sound like a mad person to even talk about the idea.

Forgot to add that this is re-posted from a Facebook post a year ago. As the voice of universal basic income gets louder, it gives me more courage to post something like that.

Hi! This is the initial release of TA.Gui. It is an open-source tool that helps automate user intentions by reproducing their web interactions. Underlying is CasperJS / PhantomJS / SlimerJS. Workflow is record using Chrome extension, fine-tune intentions and conditions using simple English, then run (visible or invisible mode).

Supports datatables for batch automation and makes 1-liner API calls. The beta was released a month ago and it now supports running out of the box for Windows, macOS, Linux and Node.js. No setup needed, just unzip and run. HN thread for the beta release - https://news.ycombinator.com/item?id=13674669

UPDATE — since then, I have not used the flows or manually send thank you notes anymore. Better err on side of caution than risk being spammy.

Thanks gus_massa! I was still figuring it out. This is actually my first open-source project, and first time using GitHub.

I wasn't sure of the correct social etiquette there. Yep I made it clear that it was a one-time thank you email, and do think that it borderlines as spam.

Yeah that's true.. I really should have just stick to using the issues feature. Learning point for me, thanks minimaxir!

I wanted to create a free open-source web automation tool. I find that I could go faster in the iteration with more feedback. So, besides thanking them, I gave my personal email so that more feedback can reach me easily. Besides going through the issues page.. Maybe I'm rushing it.

Is it ok to view a person's profile individually and send an one-time thank you note? If that is ok, why can't that be automated? Since the intent and purpose and outcome is the same. Just that a computer executes that human intent.

How else to reach out to users without using a 3rd-party mailing list? I wouldn't trust a 3rd-party with users' details.

I would do it manually if I could, really. But it is not humanly possible without stopping work for a couple of days on the repo.