Randy, how long have you been trading foreign exchange and what do you really like about the job?
I’ve been in foreign exchange for four years. One of the things about making the switch from the exchange traded environment to FX is the amount of continuous data to which you have access and are able to analyze. In the typical exchange traded environment, technical trading is a harder endeavour due to the uncertainty and emotion surrounding both the open and close, it is a daily cycle. FX is a more continuous market given that the cycle resets weekly. This, in addition to the edges that we have developed, have assisted my transition to foreign exchange.
Can you tell us something about how your business model is set up and operates?
We execute on an agency basis (24 hours a day), where clients will supply us with encrypted trading algorithms for hosting on our desk. We then consult with the client to ensure that the strategies are viable from a technical and execution standpoint. In addition, those clients can take advantage of our proprietary sentiment and range trading models to improve their own base ideas. The majority of hosting clients that we onboard are looking for our edge to improve or supplement their own returns. Most price derived strategies give you a clear understanding of past price action, our sentiment models can give you a clue on possible future price action. The combination of the two can improve most price derived models almost immediately. For clients that are looking for assistance on their models, our consulting, usually includes the addition of our edges overlaid on their own. For clients that house models on their end and simply hit our liquidity, we can, under certain circumstances, share raw aggregated sentiment data that they can use and process with their own strategies.
For clients looking only for liquidity, we supply them with the appropriate API connection needed.
On the agency side, apart from TradeStation, which other modelling platforms do you support and are you seeing an increase in the range of platforms clients are asking you to support?
One of the issues recently has been that most of the institutional platforms in FX are primarily geared around the OMS concept. The limitation here is that if your strategy is not relying on execution superiority or catching the banks off-guard, what can you use? Two of the up and comers in the modelling platform arena are 4th Story and Quanthouse.
For very quick development of price based strategies that concentrate on only one instrument at a time, Tradestation is the easiest option. So, for us, it is a starting point to test a hunch. Once we have it, we usually move onto different venues to test and develop our ideas.
When you provide agency clients with quality feedback on their strategies what in/out of sample testing process do you use to determine whether or not they might be viable?
There are two types of entities that we will deal with. One will quantitatively determine the current market conditions based upon much testing and analysis without much fundamental insight. They primarily use price and volatility to paint their picture of the market. The second type of entity is somewhat fundamental in nature and will make some assumptions about market conditions, then will build their strategy around those assumptions.
For quantitative entities their view of the past and present will be spot on but does their view extend far enough back? How does their strategy perform if the conditions in which they have tested then change? For instance, do their strategies rely upon the low volatility of the last three to four years? If so, what happens when we extend this view back further? Have they crafted a low volatility strategy that trades EUR/USD? Why trade the Euro when you can trade EUR/CHF through the legs to get your size off? For these entities, we want their strategy to use more time than they are using to test and we want it to perform over multiple pairs without falling off of a cliff. If their strategy is tailored for EUR/USD and can hold its own on USD/JPY … then you know that there is some validity to its parameters. If it can them move to an un-related cross rate and simply not lose anything… you may have something unique. We always want to stress the strategy in conditions that it was not built for.
For fundamental entities we want to challenge their market assumptions with frank talk about what is working right now. Are they trading a strategy that is relying on some sort of market consensus to perform? Is that consensus influencing your fundamental assumption, such as being a USD bull or bear? What happens when the market challenges your assumptions, or more commonly, what happens when the consensus is wrong. We want to play devil’s advocate with this type of trader.
For both of these entities, in addition to consulting, we can share our competitive advantages. Our aggregated sentiment data is a forward looking indicator in a world that usually observes the past in order to speculate on the future. Insight into levels that are breaking points for the majority can greatly assist them in avoiding loss. After all, if 90% of traders are on the same side of the market, you more than likely will not want to replicate their position.
Have you noticed any particular trends as regards client strategies in terms of either time frame or modelling concept?
The main trend has been a large scale die-off in the typical trend following strategies. That in itself may be a signal but in addition, low volatility and range strategies on G7 cross rates have and continue to do well.
Do you evaluate client strategies for their impact on your technology infrastructure? (For example, do you evaluate their likely traffic volume and reject strategies with an exceptionally high ratio of order traffic to trades?)
We absolute do. One of the good things about the migration from exchange traded environments to OTC FX is that the new players are coming to FX fully armed for the difficulties and speed of the exchange. This has caused many liquidity providers to step up their technological development in order to keep pace. Unfortunately, one of the side effects for the liquidity providers that cannot keep pace is that the trend to increase volume/market share at all costs is coming to an end as they come up against these technologically superior takers.
And again, for takers that flood the system with high ratios of unexecuted orders, whether the liquidity provider sees the orders or if its an anonymous venue and sees only the executed flow, its one and the same … the liquidity providers are left holding the bag on those transactions.
We have used this as an opportunity. Takers who act have found it difficult to scale their strategies. It typically takes only hours to days to see the aggressive trading activity. We have, in turn, transformed these takers into makers themselves. The exchange based markets have honed the skills of these aggressive takers and we have used them and their skills to create prices in order to add liquidity to the system. With our captive order flow we have given them ripe opportunity to compete for deal flow along with other traditional makers, but, on a passive not aggressive basis.
This trend is in its infancy but our captive flow has given us an advantage in attracting these players to our platform. In this case, I don’t care how many orders they send as long as it narrows our spreads.
Do you feel that clients fully appreciate the impact of concepts such as network and application latency on the performance of their strategies?
At this point, there are very few clients who do not have a good understanding of these issues. But again, are you trading based on execution superiority? It’s an uphill battle if your only trading edge is speed. Counterparties who rely strictly on this have found that makers are not as motivated for volume at all costs as they were even one year ago.
There are a lot of things that can be done if you are being plagued by latency or liquidity issues but the majority of them have to do with breaking pairs into component legs and using the liquidity available in the majors to assist execution. This is easily done with G7 currencies, which is why we concentrate on them.
Are you continuing to add connections to aggregation services like Currenex, or do you feel you already have access to sufficient liquidity?
The majority of our liquidity comes directly from our providers. The last thing we want to do is to trade with a price maker in two different venues. We’re not aggressively trying to add the latest aggregation service to our liquidity pool. It’s preferable to deal directly with price makers so as to avoid hitting replicated liquidity.
Do you feel that the technological challenge of connecting to such aggregation services is in general becoming easier?
It is. As volumes through aggregation services have increased, the number of mediums to which they have built out to has grown. The aggregation services have pre-built adapters to the majority of liquidity providers already. It makes using them very convenient for most takers.
Does your hosting model technology allow you to dissect the performance of client trading models by strategy type and time frame?
That is the primary reason why we would host a strategy instead of the client simply accessing our liquidity directly through an API. The access to feedback and analysis as well as our trading edges are the reason to use our hosting service. For the rest that do not need this, we provide the full suite of API connections. When deals are done through our API there is no way to dissect performance as you only get a portion of most deals. For clients on the hosting side, we need to see the open and close in order to track the performance of our overlaid models.
Is your SSI primarily based upon aggregate flows or does it also incorporate technical and/or quantitative modelling?
The raw data is based on aggregate trade flows across the majors. We then create synthetic data for cross rates. The data is provided in raw format or we can work with the end users to analyse and manipulate the data to create the indicators they require.
As regards your proprietary trading, are liquidity considerations noticeably affecting your ability to capture alpha in the non-majors?
Our edge is in G7 currencies. The majority of deals are in these pairs. Liquidity has not been a huge consideration given their relative liquidity compared to most EM’s.
How much of the performance of your proprietary trading would you attribute to the efficiency of your technology platform and how much to the intellectual property of your models? And does this ratio vary much depending upon how high frequency a particular strategy is?
Our advantage does not lie in execution, so for us, 90% of our returns can be attributed to our models with a relatively small amount attributed to execution. The models are fully scalable by utilizing our edges in sentiment and range trading. Given their light footprint on the market, spreads are a negligible concern.
Do you expect the future development of online trading technology and e-tools to have as much impact on your current job as they have since you first started trading?
Modelling and testing capabilities of the current suite of algo. products will improve as end users demand it. The majority of these tools do not have fully built out facilities for testing and modelling, at the moment. The increasing demand is starting to give products like 4th story and Quanthouse an edge over other products which serve only as an OMS or has limited analytical abilities. I hope to see this trend continue.