The Bloomberg Terminal, a Wall Street Fixture, Faces Upstarts | Hacker News Reader

The Bloomberg Terminal, a Wall Street Fixture, Faces Upstarts

nytimes.com

48 pointswalterbell27 comments

Calling Bloomberg a tech company is generous. They are a data company, and the quality and breadth of their data is unmatched. Reproducing this completeness of their data requires would require agreements with a huge number of data providers (exchanges for publicly offered securities and over the counter exotic securities). You also need in-house collection and quality assurance (for example, some exchanges send bad ticks or bad close prices and these need manual correcting) for many classes of data that either not publicly provided (LLP ownership data, M&A league tables, leveraged loan data) or simply needs follow-on collection (such and IPO from S-1 filing through to pre-market pricing). While it's possible to chip away at their datasets individually, creating a product that provides the same thing would be nearly impossible for a small company.

I think you're right, though I definitely think they're a tech company as well. It's just that they don't really have to compete on user-facing technology.

We have many, many people that work on data. We also have ballpark 4,000 programmers. If that doesn't make a "tech" company, then what is your definition?

As a trader and dev who's used Bloomberg for over a decade:

The only reason Bloomberg chat is used is because all the traders have a terminal. There's no reason that couldn't be replicated. Heck, they could just get Slack. It has the main requirement, which is the log is stored (compliance). You also have group chats and "ring the bell" functions.

The thing Bloomberg gives you that's useful is a way to look at all sorts of data and pull it out in many formats. Your average trader is an amateur at coding, but he can still figure out how to pull some data onto his excel spreadsheet. Once you graduate to a somewhat more robust system, there's a bunch of APIs in all the popular languages that you can use to pull the data into your own database (are you supposed to do that? I don't know, but I'll bet you everyone does it). The data part is a bit harder to dislodge, because BBG provides a common way to refer to every instrument, and you get used to the field codes. You also understand what that data means, ie details like when the data is published, whether the numbers are adjusted for splits, etc. That would be an annoying and tedious thing to replicate with some other provider, and could easily cost you more than 20K in salaries to recreate.

Bloomberg is indeed a major overhead. If you get one for every trader/salesperson, that's quite an overhead on top of that person's salary. Most people don't event use it enough to justify having their own, but I guess they'd feel bad if they didn't have one, status being what it is on a trading floor. When things are going well, you can say "oh it's the commission on a single trade" (which is true), but when things are bad you might think twice. In order to pull all the money out of the customers, Bloomberg prevents you from using RDP into the machine.

Replacing the chat is the idea of Symphony, based on the idea that if major banks all buy in and force their traders to use it then it has a chance at adoption, which leaves the possibility of other start-ups chipping away at the individual data issues one by one.

I've heard mixed feedback on whether it is realistic that Symphony gets to sufficient scale.

Not just banks. All their clients across the world too. Bloomberg is the "latin" of finance.

Absolutely it's about the status. No-one in the industry wants to be the person saying "Um I don't have Bloomberg Chat. Can you email it to me instead?"

Regarding just replacing IBchat with another chat program, Reuters Messenger has been on most trader stations for over a decade now but has failed to gain traction.

In my experience, this is not about the chat software (Bloomberg's not that great at it) but more about the best customer service ever. And that's really difficult to copy, in particular by startups which usually prefer to forget about customer service at all...

The best customer service ever?

If you ask them something mundane like "How do I show dividends for GE?" they will know.

If you ask anything even slightly technical like "How do I get the API to adjust the divendends?" you will hit a wall.

The customer service people seem to be there only for the terminal users.

Whenever I have issues with their API I just use their API support web page, not the terminal. In my experience, the terminal support team doesn't know anything about their APIs. If you are as lucky as I am, you also get to directly email or call the person responsible for handling all about your company at Bloomberg. I understand other people might have a different experience but in my view they are the best custome service I ever used. On the opposite side I can think of their main competitor, Reuters, and their awful customer service, in particular the API support team. I always need to schedule a meeting if I want something done about my support emails. I would add their APIs quality match their customer support...

As far as the APIs, they're there specifically so you can pull data into your own tools. I wrote a few Access-based tools (it was the only thing on offer in a big pension fund) using those APIs; it was invaluable for things like tracking unsettled trades.

Bloomberg runs on 25 million lines of Fortran...

https://etrading.wordpress.com/2006/06/01/25-million-lines-o...

The NY Times is right to point out that it's the range of offerings in Bloomberg that make it so hard to dislodge: news, dealer quotes, market data, reference data, historical data, email & chat, charting etc. There are many firms tackling parts of that puzzle, but nobody attempting to do it all. Bloomberg is a classic platform business; with such strong demand from one constituency - traders renting terminals - that it can afford to treat other platform clients with disdain. When I worked for a dealer contributing quotes to BBG I found them incredibly difficult to work with.

"June 1, 2006"

The BBG UI certainly hasn't changed since then. They bought in John Lakos (https://www.linkedin.com/pub/john-lakos/4/117/888), the "Large Scale C++" author to help them make the jump to C++ in 2001 and linkedin says he's still there. Given they have a winning business model, and the forces of inertia being what they are, I'm betting nothing's changed...

The UI looks archaic but I wouldn't presume that says anything about how it works under the hood. There are other forces at work on the UI. See e.g. https://uxmag.com/articles/the-impossible-bloomberg-makeover

The UI and the technology it is built on have changed drastically since a decade ago, but not all changes are cosmetic. Changes to the core UX have to happen gradually and evolve with users because, as you know, this is an actual real product used 24/7 and you can't just change everything out from most users on a Friday when they come back into work the following Monday.

I'm not going to say much about that linked article -- just that it is almost complete nonsense and wish it would stop being brought up whenever the software is mentioned.

Yes, I think we're saying the same thing. And I didn't realize that was a frequently cited article.

In my last job, I used the Bloomberg terminal on a daily basis. I loved it -- the breadth and reliability of their data is unmatched.

As I see it, the two main weaknesses to the business model an upstart could use are:

(a) DOS-like UI/UX: though many of their clients take pride in their skills at navigating the arcane maze of commands to access data (and once you know how to access the data you need it becomes lightning fast), it's a serious hurdle to overcome for more casual users. The terminals sat relatively idle in my office when I first joined because no one went through the trouble to access them.

They have excellent customer service on call who basically does the work for you at no further cost if you need it, but that's a very expensive band-aid, and it's time-consuming. Many of the senior directors at my company preferred to use Thomson's browser-based data from their desktops to get information more quickly.

(b) Privacy/secrecy: In May 2013, a few banks, including Goldman, complained Bloomberg reporters was spying on their traders' activity on the terminals:

http://www.nytimes.com/2013/05/13/business/media/bloomberg-a...

Secrecy is paramount to these traders. I'm not surprised they're backing a new chat feature.

> it's a serious hurdle to overcome for more casual users.

Bloomberg has no casual users at $21k/terminal. And they probably like it that way.

I'm sure there's a way to make the interface less arcane, but it IS effecive for regular users, and I suspect that the main complexity problem is in the data domain, rather than in Bloomberg's specific UX.

It says a Bloomberg Terminal costs 21k dollars per year. I was expecting it to be more expensive, since an on-premises license for Hipchat in a large corp (5k users) costs 3.5 times as much. Is it 21k $ for how many users?

I think it means 21k for each panel.

21k$ per user and per computer.

per user or per computer (Bloomberg Anywhere or Bloomberg Open). Real-time data is extra since the exchanges charge more, but 15' delayed to pretty much everything is included, which goes a very long way.

That's in the ballpark for basic service where I've used it. However, there are add-ons like exchange connectivity that will require extra money. If you want to see real-time orderbooks in exchanges around the world, it's a good chunk more. I'm sure there are other add-ons as well.

21k for 1 user for 1 year. Your average "bulge-bracket" IB is paying for thousands of users. You do the math...

That's cute. Good that there are alternatives, but Bloomberg isn't going anywhere, it forms part of the foundation for half the hedge funds and investment banks in the world. It's not just a source of data, in many cases, it is the definitive yardstick to which all other data is measured, to the point where, if Bloomberg has actual, verifiable errors, people will often knowingly mangle their data only to match Bloomberg.

Here is the application : https://www.money.net/#see

It seems written in Java.

Reply on news.ycombinator.com