It’s interesting. I thought machine learning would be further along in taking large data sets and organizing it in meaningful way for business metrics and decisions. It’s not even remotely close yet. Companies have more data than ever, but they are struggling to use the data to make their business better. Tableau helps with visualizing it, but you have to structure the data and create relationships first. I see most humans struggle badly with it. The ability to turn data into something useful is becoming super critical. Everyone should take some basic classes on relational databases.
I have no idea how you structure data into a coherent format without relationships. Without getting it into a usable format data is just noise. It’s like an orchestra with everyone playing randomly. There’s a reason why there’s sections, harmonies, etc for music to sound good.
I didn’t realize I can put machine learning on my resume. That’s awesome. Haha.
I was thinking of relationships in terms of having 6 tables with data and how do you join it all together then group it, so it can be useful. Most people just see lots of data and get overwhelmed by it.
True HFT just front runs other orders due to have better data access. If you are willing to pay $100.00/share and someone is willing to sell for $99.90, they step between you two and pocket the $0.10. It should be illegal for them to have access to that data.
Algo trading is interesting. The ones I’ve heard of are pretty simple. The watch twitter data to see if a company is suddenly getting a ton of positive or negative mentions and trade on it. One trades based on if Trump makes a positive or negative tweet about a company. There’s also a bunch of algo trading based on technical analysis. All they did was automate what the human was doing in terms of watching moving averages, RSI indicator, or whatever your technical measure of choice is. I don’t consider any of that machine learning though. Humans realize those patterns are there.
I want something that discovers patterns and correlations that we aren’t even aware of or thinking about. Tell me bug spray sales in Brazil can predict how the farming season in China will be. That’s a huge advantage that’ll take others forever to figure out. That’s the type of stuff ML should be capable of delivering.
My friend works (investment division program manager) one of the insurance companies, he used to mention this to me. There are plenty of such programs are running at wall street hedge funds, big institutions, insurance companies. They pay $300/hr for those technical consultants…and most of them are specific business/tasks. Those software, such as Bloomberg terminals (subscription is $24000/month !) or its derivatives, are expensive, and focused on big firms.
Having such knowledge is wonderful but you don’t really need it. Personally I’ve never had such advantage but I still laugh all the way to the bank with my stock portfolio. The trick, of course, is the looked-down-upon-but-so-very-effective buy and hold for the long term.
Even though I have not bought at the low of Feb/Mar for F10s, is already up 15% from my average purchase price. Many folds of the cost of dim sum with manch. Hail F10s. But only meagre sum invested didn’t follow wuqijun advice… refuse to admit is my fault, is his poor teaching. DCA purchasing of good stocks work well.