What’s Next For Brazil?


By Christian J. Zimmer and Hellinton Hatsuo Takada, Itaú Asset Management.
The electronic trading landscape in Brazil is highly connected to BMFBovespa. Fortunately, BMFBovespa uses FIX as their communication protocol. However, the adoption of FIX does not imply the possibility of electronic execution of an order of any security and any size. At BMFBovespa, mostly equities, a few futures and some option contracts have significant liquidity for electronic trading. Obviously, when executing large orders electronically, it is necessary to have liquidity to avoid price distortions and, at BMFBovespa, possible unscheduled auctions.
On the other hand, BMFBovespa recently introduced the PUMA trading environment which reduced the execution time significantly. This system was developed by BMFBovespa together with the CME and started at the end of 2011 as the new platform for derivatives. During the first half of 2012 the migration of stocks and stock derivatives was concluded too. In the future, the fixed-income platforms Bovespa FIX and Sisbex (from the Central Bank) are supposed to migrate to the unified PUMA platform.
Additionally, the BMFBovespa’s market data feed is now faster with UMDF/Fast FIX. In the marketplace, we can state with confidence that there is now a mature technology setup for exchange-traded assets.
But there is more – especially, when it comes to the question of liquidity and the integration of other markets and functionalities. Firstly, let’s take a broader view of the trading environment. When it comes to government bonds, two platforms are the most relevant in Brazil: CETIP-Trader and Bloomberg. Additionally, there is the BovespaFIX environment with focus on corporate bonds, which is under reformulation and a new service is to be launched in mid 2013.
Many broker-dealers display their prices via individual Bloomberg screens or they can be aggregated into a Bloomberg function. The same happens for many options: broker-dealers offer screens where the clients can also follow the morning and afternoon auctions.
Finally, even in the equity space trading of big blocks is still a manual process where the brokers look for counterparties. Information leakage is potentially very high.
The Darkstone project
In order to achieve better results in big blocks for equities, Itaú Asset Management (IAM) started at the end of 2012, the Darkstone project. IAM is one of the largest asset managers with a focus on Latin America with approximately USD300 Billion AUM. Big asset managers outside of Brazil typically access dark pools and other alternative venues for executing big positions. The lack of alternatives to Bovespa for stock trading has made it common practice to find counterparties for blocks via phone/chat with the broker-dealer. After agreeing on the price they cross the parties and inform the exchange of the trade. The trade is thus a Bovespa direct cross, not a broker internalisation. Unfortunately, this manual process opens the door for operational errors – following about 20 chat windows and negotiating terms with all of them is not easy when the market becomes nervous. Therefore, in 2013 we started to connect our brokers and to negotiate big blocks with them based on the IOI-platform from Raptor Trading Systems.
By now, three alternatives, all based on the FIX Protocol, are possible for arranging the trades:

  1. Acceptor approach: This alternative represents the typical IOI flow where the broker basically sends an IOI and receives a NOS as a response. In order to better guide the brokers, at the beginning of the day, IAM sends a list of stocks we intend to trade. Then, during the day, at any time, the brokers can send actionable IOIs. If our traders have posted a corresponding order into Darkstone, immediately a NOS is created and sent to the broker. A fill execution report is expected from the broker. The advantage of IAM is the reduction of footprint in the market.
  2. Initiator approach: Following a performance based ranking per stock, IAM sends IOIs to the first broker. The IOI has a 5-10 minutes lifetime with price replaced as the markets move. If the sell-side trader finds a counterparty, he sends a referring IOI and the flow continues as in case 1. If the sell-side trader does not find a counterparty, the IOI is canceled and sent to the next broker in the ranking. As this alternative creates a certain footprint, our traders observe the participating brokers’ market behavior. Any misbehavior has the consequence of excluding the broker from our list.
  3. Broadcasting approach: Like in a classical auction, we invite our brokers via RFQ to give us a price for a defined position. If no satisfying price is received, IAM’s traders may step back and not trade at all. Once again, the market footprint is a critical concern.

As Brazilian laws do not allow for internalisation, this is not a dark pool setup and all the trades are sent to the exchange and registered there. The prices and quantities we promote are constantly controlled within the Raptor system to avoid unscheduled auctions going to the market.
Of course, making direct crosses is already very popular in Brazil. But it is based on chatting, and making it FIX-based allows for a relevant increase in efficiency. Further, we expect in the near horizon the maintenance of the total trading volume from direct crosses and a fair performance-based evaluation of our brokers.
Internal messaging
It is not by occasion that the Darkstone project is FIX-based. In 2012 at Itaú Asset Management, we reformulated our electronic trading environment using FIX. FIX allows us to work in a modular form.
Now, our OMS is focused on OM-tasks. At the trading level, each trader can use the EMS he likes, as long as it talks FIX. The traders send the orders to our FIX-based risk module from there they go to our router that manages all the connections. As the risk and the routing modules are also from Raptor, the integration with the Darkstone project was seamless. Being able to individually treat every piece of our ET-architecture allows us to switch to newer technologies without affecting other parts of the workflow. But, in order to make the internal FIX-communication a success, some points have to be observed:

  1. At the OMS-level, neither the execution broker nor the (mother) accounts are known. This information comes back from the EMS. So, the OMS must allow for possible changes of these points when it receives response messages from the EMS. For example, tag 1 and 453 must be changeable.
  2. In order to correctly check trading risk limits, the EMS must send the individual accounts to the risk module in repeating group 78. And the risk module must process the repeating group correctly. The importance of the rep group 78 goes beyond pre-allocation in this case. For more details on risk controls and FPL guidelines, please refer to http://www.fixprotocol.org/ working_groups/riskwg/documents.

FIX for post-trade is another key area we will be focused on in the second half of 2013: (see http://www.fixprotocol.org/ for links to the most recent guidelines). Today, FIX-allocation is still very rare in Brazil, even in the equity space. Allocations are still confirmed via the exchange of Excel files. This is a slow and failure prone process.
If the buy-side OMS has an integrated post-trade module, emitting FIX-allocations is straight forward. But even if not, it is not too much of an effort to get the executed trades and follow the process proposed by FPL.
The problems of using FIX-allocations in Brazil are mainly due to the sell-side, which does not seem to be prepared. Nowadays, the brokers extract their clients’ trades from the exchange’s Sinacor database. A few brokers created web-services where the buy-side can access their confirmations. Instead, a FIX-session should be made available.
We hope to see FIX-allocations become reality in the Brazilian equity space by the end of 2014.
As a final point, here is a short list of potential FIX usages that should receive further attention in the next years:

  1. Government bonds trading
  2. Individual sell-side feeds for illiquid and OTC products, not only for market data, but also for trading.
  3. Stock-lending/borrowing platform