I was invited to talk to some City folks about using data analytics for trading. It made me pause for reflection, as I’ve been talking about analytics for trading for a long, long time. Twenty years ago in fact.
Back in the 1990s, we were taking about object oriented programming, neural networking for stock picking and program trading.
A decade later we had moved along to focus far more on low latency, high frequency trading using algorithms and black boxes.
Now we talk about open sourcing structures using cloud base machine learning and sentiment trading. In other words SSDD: Same Sh*t, Different Day.
The themes may all be the same, but the times have changed. Twenty years ago, we were taking Data Warehouses for enterprises. These were incredibly expensive systems that were designed to trawl through terabytes of data. Such machines are no longer needed as all data can be homogenized and cleansed into a cloud-based AI engine and leveraged madly.
Unfortunately, many financial institutions will miss this trick, as they don’t allow any data off premise. That mentality needs to change if they are to keep up, as we’re now taking petabytes of data. Most of the data ever created in this world has been created in the last year. And it’s growing every day.
For example, most traders will use Bloomberg or Thomson Reuters feeds to trade, but what about using my Facebook posts? Some do do this, for example Stocktwits tracks tweets for trading trends, but what about LinkedIn and WeChat?
Now we are talk trading using sentiments and emotional analytics. I heard this for the first time this week, and the idea is to catch trending news via what people are saying before the markets catch the trend. The example used was the issues with the Samsung smartphone when launched. By the time the market knew there was an issue, it had been debated online for days. What if you had caught this on the first day?
I liken it to colocation and proximity hosting. This was a trend at the end of the last decade for the big market makers to pay megabucks to have their servers placed as near as possible go the stock exchange systems. The game was that whoever’s servers were nearest would match trades first and the first to match a trade gets the deal. There are no prizes for second place.
Ten years later, the first to hear of a deal will get the business first, and so listening to the sentiment of the markets via their chosen media is becoming a criticality.
A great way to thin of it is when you post something on Facebook or twitter. Are you the first person posting this, or the tenth or the hundredth or one of millions? When the babies crashed the academics finest moment on BBC News, were you the first to see it, the tenth or the millionth?
You get the idea. One posting or ten probably makes little difference, but if you see 1,000 retweets or 10,000 Likes occur in the space of minutes, something is happening and if you can catch that something before it appears on twitter or Facebook trends charts before everyone else, you have an advantage.
This is why we are talking about masses of data, both internal and external, that needs analysis non-stop in real-time, and the players who do this well will be making ten times more than those who hear it through their market feeds last.
Today is a world of smart trading through smart systems using smart data. Keep up.