Big Data Debate Panel on Data in Finance
Last night we had the pleasure of hosting yet another Big Data Debate at the wonderful Level39 in Canary Wharf. As we were in the heart of the financial sector we thought it would be a great opportunity to cover Big Data in Finance. It was a great event, full of insight and contention.
Is the financial service as an industry incompetent with data? They benefit from a high level of data maturity, with more people focussing on it than any other sector. Yet it is still unable to predict any big financial crisis. Considering its highly evolved risk models and teams consisting of highly intelligent people, how can this industry still suffer from a constant cycle of massive boom and busts? What’s the future for Big Data in Finance?
John Burn-Murdoch (moderator) – Data Journalist for the Financial Times
Stephen Bonner – Partner in the Information Protection team at KPMG
Jon McLoone – Director of Business Development at Wolfram Research
David White – Founder and CEO of import.io
Jonathan Segal – Director at law firm Wragge Lawrence Graham & Co
Let the Debate Begin!
To what extent has data played a role in shaping the finance industry as it is today?
The panel agreed that data is essential in finance. This is evident in the fact that financial services have made considerable investment into technology and data.
“The whole of investment is all about data. That’s what finance is, it’s data”
– Stephen Bonner
The other interesting thing about data in finance, Jon mentioned, is that it is possible to get your hands on all the same market data that large banks have. and then take that data and start your own company if you think you can use it better.
Can we supplement financial data with data from the web for better insights?
This issue sparked a bit of contention among the panel. Jon started by pointing out that when everyone is using the same data you can only get ahead by either getting it faster or by using it in a smarter way. Extra data you pull from the web is only good if you can use it to find some new insight which you could then add to financial data. David mentioned that there has been a significant amount of investigation into whether or not this is possible or even helpful.
“The reality is, I don’t think anyone has cracked it yet. And if they do, they certainly won’t be sharing it with the Financial Times.”
– David White
In finance there is always a desire to find the tiniest of advantages and people will explore literally everything to try and get it. Outside data like social media can potentially be used to predict what the herd will do (which is useful), but it will never help you predict the underlying market factors. It is also much easier to manipulate than traditional financial data.
“Everyone is trying to rig the system to their advantage, so you’d get into a lot of game theory.”
– Stephen Bonner
Is human involvement being squeezed out of finance?
When things go wrong the first instinct is to try and make the model better and a lot of that is about removing the human bias in the model.
“The crisis happened because of bad underwriting, human error and game theory.”
– David White
He went on to say that maybe the move to having everything done by A.I. is a good thing. Data can’t lie or be wrong but the people reporting it can and that is where a lot of the traditional problems have come from.
Stephen disagreed arguing that data lacks a very fundamental skill: common sense. It’s important that the data make sense in context and machines don’t understand that. You need to be able to look at the outliers and say that doesn’t make sense, something must be wrong.
How can we use data to prevent another crisis?
It takes a real human to consider the edge cases and figure out what does and doesn’t make sense. One of the ways data can help with this, Jon mentioned, is by looking more closely at the individual’s data.
“If we get better at knowing who is a sound investment it can prevent us from doing things like making bad mortgages in the first place”
– Jon McLoone
David agreed saying that this system is actually much fairer way for consumer finance to function.
“Banks have the ability to monitor this micro data these days, it’s just that no one has tried it yet because it’s not data they are familiar with.”
– David White
Stephen vehemently disagreed, saying that the idea of algorithms being able to predict humanity was ridiculous. Jon countered arguing that current algorithms are based on too little data, but if we were able to use the micro-data the algorithms would get better. Jonathan chimed in to say that Wragge was already seeing lenders experimenting with using this kind of personal data to help them make decisions.
Does using this personal data violate our right to privacy?
David didn’t think so, saying that this personal information is already out there and most of it is public. Most people, he argued, would be willing to give it away if they thought they could benefit from lower premiums or gain some other form of financial gain. He cautioned however that using this personal data is much less reliable than traditional data because it is much easier to game the system and post the things you know the insurance company is looking for. Jon disagreed with this last point arguing that more data is the key.
“Systems get harder to game the more things they depend on. If you have to change everything about what you say about yourself online to get a small discount, no one is going to bother.”
– Jon McLoone
Stephen didn’t think so mentioning that if we start monitoring all this personal information about people they may be forced to start playing the system at least to some extent because otherwise they’d never get approved.
“I don’t think any of this is OK if it violates our fundamental human right to privacy”
– Stephen Bonner
Beautiful sunset view from Level39
To what extent is this influx of data a problem in security?
The panel generally agreed that banks are some of the most secure places for your data. They have always known that the data they hold is very valuable and they are practiced at protecting it – more so than a lot of other places we trade data with. The really scary thing David and Stephen mentioned is that the people who will try and steal this data are likely to be large crime syndicates and terrorist organizations with lots of resources.
What is the biggest data challenge you face today?
Jonathan: Disclosure is a big one in the legal industry at the moment.
Jon: Building an ontology that describes the world.
Stephen: Identifying what normal is and then spotting the abnormal.
David: Use of unstructured data in risk models.
I would just like to thank everyone who attended the event, I hope you all had as much fun as we did. We’d love to see you at more BDDs which you can sign up for on our meetup page. Another huge thank you also has to go out to our sponsors Wolfram Research, Teradata and Claritize without whom we wouldn’t be able to run these events. And finally, thank you to John, Stephen, David, Jon and Jonathan for agreeing to be on the panel and giving us all an entertaining debate!
Hope to see you at the next one!