As the Japanese Market Matures, Risk Checks Take Center Stage

Even still, this competition has created a challenging environment where brokers would like to see the exchanges, namely the Japan Exchange Group (JPX), take on more of the responsibilities ─ perhaps by offering a hosted risk check solution at the exchange ─ and/or providing more guidance as to which risk checks a broker must provide.

The panel was moderated by consultant Wayne Schmidt. The panelists were:

● Francis Huang, assistant vice president of equity IT, Credit Suisse
● Sean Lawrence, CEO, ABN Amro Clearing Tokyo
● Miguel Ortega, head of market data, Deutsche Securities
● Matthew Hassan, quantitative prime services, Nomura Securities

Ortega: We're talking about [infrastructure] technology that we need to refresh consistently and constantly, and we also have to refresh the risk checks and the technology that is sending and receiving the orders from the client. You have to maximize the performance of that and it's not like it used to be where you'd buy a better server and your latency would automatically get better. Now you have to really look at the code and look for new technologies that will allow you to perform these risk checks while making the system faster at the same time.

In the end, it's all about protecting your customers and protecting yourself. The risk checks have to be done as fast as possible and that will continue to be an arms race and it will continue to be until such a point where speed of light becomes the limitation. We're not there yet and until we get there, we're going to keep fighting. Now we will look to shave off nanoseconds rather than milliseconds, which we have been doing for the last few years.

The risk checks have to be done as fast as possible and that will continue to be an arms race and it will continue to be until such a point where speed of light becomes the limitation. -- Miguel Ortega

Hassan: Actually it's quite interesting these days, the brokers here in Japan are in a race where they're competing against risk checks ─ who's got the fastest risk checks right now? In other exchange venues, if we go to a hosted risk check type of solution it would level the playing field for all of the brokers and we're not competing on this risk check thing.

I'm interested to see how this shakes out because there are a lot of different interpretations going around the street right now about what constitutes a proper risk check.

Schmidt: Do you see there being a formal pronouncement where people would go through and say you must have "these things" in order to constitute a proper risk check?

Hassan: Are we going there? I don't have a feel for that yet. We're still in this phase where everybody is competing against each other, trying to push the envelope a little bit, checking and rechecking to see what the competition is doing, what is the marketplace dictating right now, what is compliance comfortable with right now. That's the most important factor right now: We want to protect ourselves, we want to protect clients and we want to be compliant, but with that said this is a competitive marketplace that we can't ignore as well. So it's getting that balance.

Ortega: It is a competitive advantage and it is a selling point, but it's also ─ frankly ─ a marketing point: Do I have good risk checks that are going to protect you? It's something that sways clients one way or the other.

Lawrence: In that debate, there are three levels of risk that we need to contemplate. First is the mathematical definition of what should all of us look for ─ is it max price deviation, max orders per second, and the definition of the attributes that the brokers should look out for when sending orders.

For that ─ and this is something I'm trying to lobby with JPX (Japan Exchange Group) and other exchanges ─ I think they should come out and define that. They should say, "If you're a broker, you need to have these basic mathematical checks to perform." That's No. 1.

The second one is ─ and this is a technology question ─ is there architecture of risk, is there a device, is it in software, is it on the broker side, is it on the exchange side, is it in between? This is an area where regulators should have a definition. Maybe they say, "There has to be a device, it doesn't matter if it's hardware; it has to be in the domain of the broker and not their investor; the exchanges should have them as well, etc."

The third component is pure technology: How do you perform it? Latency? Hardware/software?

When we talk about risk, it's got multiple layers. The first two can be defined and prescribed, and it's the last one that would be the job of all the technologists in the room to solve.

Ortega: On that point, I can see that the regulators are looking at this and they are auditing the risk checks. They do ask for that.

Schmidt: You talked about a hosted solution before that can assess your position and assess whether you can make executions. Do we see that down the line being something that's another service?

Lawrence: My opinion is ─ and this relates to co-lo and those things ─ is that we have an inefficient market for goods and services for IT in Japan, and that's driven by certain structural elements, such as how NTT [the Japanese telecom giant] owns all the lines even those there's supposedly competition on the top, and there's a certain amount of inertia. But when we come to that broad question on the risk layer, I'm not a fan of saying that there has to be a device, and the device has certain types of technology.

I think the right way to do it is to set certain standards and a definition for a type of risk control, and it has to have an actual component where the broker controls it and not the investor. Beyond that, it should be up to each broker and each technology firm to determine what the best technology is to suit the purpose.

So if someone comes along with something like an FPGA card and it's great and super fast, someone should be able to build their risk checks way back up in the algorithm of the trading calculation; as long as I can demonstrate that risk calculation is performed way back up the stack, then I meet the requirement and tick the box.

I'm interested in the efficiency of the market for goods and services for IT in Japan, which I think needs to be worked on.

Ortega: What we do as brokers is we're creating dedicated hardware with a dedicated FPGA card or processor. It's scalable and it's an appliance.

Now, whether you do your checks high in the stacks or as they're going out to the market, I think the point that Sean makes is valid: It doesn't matter where you do it, but there have to be standards and we all have to be in the same boat. If you bypass some of the risk checks you may be faster than the others, but is that safe for the market?

Huang: I'd like to make the point that the exchange is doing some of that already with the short-sell, uptick rule check. Before, it was up to the broker to do that, but now they are doing it. I think eventually we will see [some of these risk checks] slowly moving over to the exchange.

[This is Part3 of a three-part series. Part 1 can be found here; Part 2 can be found here.]