Think of it as Consumer Reports for portfolio managers. In little more than a year, Citigroup has acquired more than 100 customers for its Best Execution Consulting Services (BECS), which offers handsome graphic portrayals to show an individual asset manager how the services from his broker compare with what his competitors get from that same broker or from the universe of brokers.

“We tried to get the look and feel of the desktop and bring it to a Web-based offering,” – Stavros Kokkoris

Citigroup has been in pre-trade analytics for several years, explained Stavros Kokkoris, vice president of CitiTech Services. It had developed pre-trade analytics with Delphi development tools before deciding to move off the 10-year old technology to Microsoft .NET.

Over the last year, the group built both pre-trade and post-trade analytics on .NET. It operates as an ASP from existing server farms in the US and Europe with new data centers under construction in Japan and Australia.

Citigroup’s approach has broad appeal because it is not UK-centric or US-centric, said Stavros Siokos, managing director, head of alternative execution sales, at Citigroup in London.

“It doesn’t depend on using any single order management system. So long as you have Web access you can use BECS, and it is fully customizable,” he said. With BECS, users can use any period of trading and any framework to measure execution.

“BECS generates reports that compare all client trade execution data to a universe of trade executions we have globally,” said Kokkoris. Users can see how their trade execution compares against the universe and against a standard baseline benchmark such as VWAP, HLOC T-5, or T+5.

The user interface, developed by usability experts, makes it easy to see where a firm stands. The client data is displayed in a white line; the universe in yellow.

Kokkoris points to one set of results that show the client’s trades are four basis points above the benchmark while the rest of the universe are three basis points above the volume weighted average price (VWAP). It also displays the client’s commissions compared to everyone else’s.

Like many data providers, Citigroup’s BECS provides its service by collecting data from clients. Clients can see their own statistics while the aggregate data that forms the universe is available only as the summary.

The system is broker neutral, and information is confidential; no one at Citigroup can see the information from individual firms, and an independent auditor certifies the secrecy of data.

Users can drill down by broker to see which brokers offer the best execution and best pricing; they can also see if their broker is offering better terms to the universe. They can also drill down by country.

Brokerage firms find they can also make good use of the information. BECS breaks down trades by trader so a head trader can analyze the work done on his desk and see how well his people are performing.

Unlike competitors, BECS operates in real-time. Its architecture was designed for speed, and even a large query comes back in a second or two. Nothing else executes at the speed of BECS, added Kokkoris. The entire data layer runs on Microsoft SQL Server; CitiTech Services used mid-tier caching to speed response times. Because the entire system is Microsoft-based, users can copy and paste data from Excel and Word. “We tried to get the look and feel of the desktop and bring it to a Web-based offering,” Kokkoris said. “You can bring up a basket of stocks and perform real-time operations to get a sense of the market impact of trading that basket. Citi has had a model to do this for more than 10 years, and with BECS we rebuilt the platform from scratch.” For a hypothetical basket for 100 issues, the analytics will show the liquidity and transaction costs to trade in 12-minute intervals. It offers a breakdown by liquidity and market cap or the trading trajectory on a single stock. Scatter charts show dispersion and make it easy to identify outliers and heat maps provide visualization for impact cost analysis. While the average basket is about 60 stocks, Citi has stress tested it with 6000.

In addition to its powerful analytics, BECS contains a crossing engine so a client trading a basket can receive a summary report of trades that could be done internally.

“You can see how well a broker is executing for you. Every asset manager can use it,” said Kokkoris.

Citigroup's Best Execution Consulting Services provides real-time reports on trade execution, showing performance in multiple views, including performance against benchmarks, by country, and in total distribution.

Of course, the bank uses the tool on its own trading.

“We can see only the trades that we do ourselves,” said Siokos, “and we use it to evaluate all our traders against a benchmark so we can service our clients much better.”

Citi has seen rapid growth in the number of post-trade analytics licenses it is selling. The more users, the more data, and the greater the accuracy of the universe statistics it reports back to its clients.

Because BECS can look at individual issues, it also can be deployed as part of a firm’s compliance operations. With its historical regulatory testing for trade operations, it provides a database of all trades.

For Kokkoris, who was formerly at Macgregor, the trade order management system vendor, BECS is the third platform he has built in .NET, so he has learned his way around the development environment. His experience with Microsoft development goes back to 1986.

The Microsoft development environment has become easier to use and more powerful. Using a single vendor, like Microsoft, in the technology stack minimizes the development and maintenance costs.

The client is just 200-300 kb, a tiny footprint on the desktop machine, he said. Microsoft lets the back-end server do a lot of the heavy lifting, and our architects have taken full advantage of that. Because it is offered over the Internet, Citi can take new customers live in less than an hour.

For more on trading cost analysis see the commentary by Larry Tabb, founder of The Tabb Group, at http://www.tabbgroup.com/thought/Transaction_cost_analysis_Mar2005.pdf.

How Does It Work?
It’s Academic

Robert Almgren, director of the Mathematical Finance Program at the University of Toronto, consulted with Citigroup as they developed the models for BECS.

Wall Street is full of technology secrets, but BECS is not one of them, said Stavros Siokos, managing director, head of alternative execution sales, at Citigroup in London. In fact, the developers worked closely with Robert Almgren, director of the Mathematical Finance Program at the University of Toronto, in developing models to use in evaluating trading.

“People don’t like black box approaches where they don’t know what is happening,” Siokos explained. “All the models up to now have been based on theoretical models for pre-trade; we had these huge trading volumes but we never went into using this data to create new models.”

Almgren served a consulting role with Citigroup in developing the models for BECS.

“We came up with a model on how much it will cost to trade between 9 to 10 VWAP so you can compare existing trading data with what you expect, and then identify your expectations for future trades,” Almgren said.

Since Almgren is an academic, Siokos noted, he wanted everything published.

“Our papers have come under editorial review and we have published all the formulas. It’s all transparent, which makes our users feel comfortable,” Siokos said.

On his Web site, Almgren notes that a paper he did with Neil Chriss of SAC Capital Management in a 2000 issue of Journal of Risk, on optimal trading trajectories has been influential in the finance industry. He quotes Robert Kissel and Morton Glantz in Optional Trading Strategies who wrote: “Neil Chriss and Robert Almgren pioneered much of the early research in the field... Without their leading edge work, transaction cost analysis and trading cost estimation could not be at the state it is today. The Almgren and Chriss efficient trading frontier will truly revolutionize financial decision-making for years to come.”

In addition to papers on finance, Almgren’s site includes a link to a paper he wrote with Agnes Tourin, associate professor of mathematics at Canada’s McMaster University, on the mathematics of optimal glider flying. The problem is to determine the optimal strategy for flying speed and whether to stop in lift, with uncertainty about future atmosphere conditions, he noted.