I have some interest in this thread since I married a Brazilian. My research indicates: 1) DTN IQFeed supports the B3 exchange....they call it BMF 2) Interactive Brokers does NOT support the B3 Exchange. I was very surprised at that. 3) Rithmic does support the B3 Exchange....they call it the BM&FBovespa: https://www.rithmic.com/connectivity.html So technically Dude, you could be up and running within days with a Multicharts subscription and a Rithmic account....So much for "my country is so far behind". That's nonsense. You just need a Brazilian broker who subscribes to Rithmic. Of course, you instead want to create your own platform that encompasses all of the functionality of Multicharts and Rithmic ? So, in that case, check back with us in about a year and a half million lines of code.
Thanks! I didn't know that. I do need historical data and a way to run custom simulations. Just that. I've used Metatrader/MQL5 and some custom dll to analyze, test and paper trade. And run ML strategies with Ninjatrader/C#. I used RStudio/R to run simulations on tick data (and plot nice charts, of course). I want to do something different now. So I'm improving my julia and c++ skills (but I don't intend to write half a million lines of codes... maybe a couple of thousands ). Next year I have to submit a course completion work. It's not rocket science: data analysis + nice graphs + simulations. I'd like to enter another masters in the sequence (maybe in applied math) and keep the "tech stack" to avoid relearning everything every time.
Interesting journal. I was always studying price action. One day it dawned on my monkey brain. And innerstanding of the underlying structure of price action will help a lot. Thus microstructure is introduced. You discover a lot of things, and the changes within our modern market are moving at such a rapid pace. We act as if this is all normal, that you can sell a spread in $TSLA from your ipad on your couch, when just 50 years ago Fischer Black was pondering if market making could be automated. Shortly after the DOT arrived and forever changed markets. The rise of the machines ever since. Man has always made his own market, but ever since the tech boom our markets have become digitized, decimalization extinction. This rapidity has increased volatility. Markets today move in magnitude larger than markets of the past. Not sure if thats a fact, but i do know ever since 2005 with the introduction of Reg NMS, this ultimately established HFT in the markets and the velocity of every metric (volume, volatility, liquidity, var etc) has risen. I'm going to read over this thread now, wondering if anyone discussed LOB's. Cheers!
My goal is to elaborate a paper and put it in arXiv.org (and to submit it as my course completion work in Stats too). About order book modeling: the approaches, what have been suggested and tested. Of course not everything. Maybe not economic/game-theoretic stuff, or econophysics, for instance. But I'll do it for fun, not for a academic degree (like a ton of papers I've read). But in the end I want to devise a trading system. I have a general idea of where to look. I have ~4 hours a day to dedicate myself to this. I'll start with measure theory (I'm obliged to that), c++ and reading some selected papers about this kind of modelling. While I still don't have anything concrete to show, I'll share my ideas about trading in general. PS: I have a weird sense of humor and sometimes people don't get it. But I swear everything I write would make perfect sense in Portuguese .
As SLE said just sick with - what happens to the order book when price changes? First level only. You are looking for stickiness. Not counting the purchased orderflow HFT firms, everyone else is basically looking for the biggest iceberg and making sure they aren't trading against it.
I've done a lot of work lately on analyzing the InsideBid/InsideAsk. Remember, this is Level 1 only. The jury is still out as to it's usefullness and ability to lead the price action.
I would rather suggest that you start out very basic and formulate the problem you want to solve first: Here's a market. Preferrably a liquid and volatile one. How can I make money trading it? How can I make the most money trading it? Where is the most money located? What's required of me or my system to profit day in/day out? Then start your problem solving... Maybe you'll need your advanced tools and theory. Maybe not...
Today I started to negotiate with a vendor the acquisition of market data. I'll have to start my own database, as there's no historical data available. "Almost nobody ask for this", they said. It feels like I'm buying rubbish. The book is noisy and probably useless 99% of the time. At first (depending on the price), I'll manage ~70 streams of data (one for each order book). Basically everything I need to know, including the id of the institution. There's no market fragmentation here, so it's all pretty simple. I'm resisting the urge to pick a single instrument (dollar future contracts, for instance), and include some "intelligence" to the book: I would concentrate my efforts on something more viable and cheaper. But let's keep in mind that this is a school project too.
"In 2020, however, $SPY's intraday range volatility increased from its 10-year average by 119%, surpassing the 2020 range volatility for $AMZN and $GOOGL" This is hard to believe but its amazing. 2020 has been one of the most volatile years we've seen yet.