FX Algos: Changing Technology And Trading

With John Radle, Global Head of Trading, Campbell & Company
Currently we trade more than 90% of our FX flow electronically, that is a combination of bank supplied algorithms and point and click trading via an aggregator that we assembled leveraging our bank relationships and in conjunction with Portware our EMS provider. We put together the liquidity pool to give us the ability to trade point and click if that’s the best method of execution for the order. Or, we have the ability to use a variety of bank supplied algorithms that we use to target specific execution benchmarks, including time weighted average price and arrival price. Most recently we’ve been trying to take the bank algo approach a step further by partnering with Pragma. Pragma is working with us to build on their extensive experience with equity algos to develop custom FX execution algorithms. We assisted them in establishing a liquidity pool that their algos sit on top of accessing various bank and ECN streams.
Our trading is strictly execution based. Our quantitative group on the research side creates the systematic alpha models, which then produce orders that are delivered to the trading desk. The orders route into our OMS and then the traders look at each one and select an appropriate execution algorithm to match the desired benchmark.
We’ve also built some proprietary TCA tools. We have a platform that allows us to watch the orders in real time, compare the benchmark price to our average price and effectively monitor slippage in real time. As we’ve gone increasingly electronic and we have multiple execution algorithms trading simultaneously with different banks and at Pragma, the traders have a view as to how those algorithms are performing versus the stated benchmark that they’re targeting. That way if we start to deviate too much from a specific benchmark, the trader can intervene or investigate why that trade isn’t working as planned and adjust accordingly.
Extending FX into TCA
We see this as being part of our real time TCA, as we can see what percentage of the order is complete and how much is left. We can see our start time on the order, the end time, what the benchmark price is being calculated at and what our average price is at that moment. We can also see the difference in terms of dollars so it’s a very clear view. What we have found is, as we became increasingly heavy users of algorithmic trading, we could not just rely on an end of day, after the fact, transaction cost analysis report.
Moving that process to real time has assisted us in our goal to provide best execution for our investors. This allows us to effectively monitor trades as they are happening, as liquidity is evolving in the market. It tells us whether spreads are tightening or widening and then it’s very apparent to us as the trade goes on, what the slippage is, and it gives us a great view into what’s actually happening at the order level. If we have 10 or 20 algos trading at one time we can monitor that whole group in real time and it gives the traders another tool to ascertain whether the algo is performing as we expected. The whole real-time TCA concept is something that the brokers really don’t offer. We had to build the systems ourselves.
The liquidity pool that we created allows us to do point and click trading on that pool. It allows us the opportunity for better execution for our investors because we’re able to stream a variety of competing banks into that pool and we’re also able to include some ECNs in there too. We feel that this provides better price discovery by having the banks and ECNs compete.
The future of RFQ voice trading
There will always be a need for voice trading FX. It just depends on what trading style you use. Obviously, there’s more of a need for that type of execution when you are trying to move larger blocks quickly. If you want a big position put on immediately and then a few hours later you want to take the entire thing off then that type of trading would suit. We have become more granular in our trading over the years, and now we take our orders and break them out to trade in different timeframes instead of doing those very large, potentially market impacting, trades. This allows us to take a large order and break it up into smaller pieces. Then we can execute those smaller pieces on much tighter spreads in much smaller amounts which allows us to effectively match our benchmarks more closely.
Generating the benchmarks
Our research team has done a lot of work around our execution benchmarks; when they deliver an order to us it has a pre-determined benchmark, a start time and an end time. Then we’re just measuring that benchmark over that timeframe. We did have benchmarking challenges when we were trading larger sizes more quickly because trading windows were less defined. But as we moved our trading style to smaller sizes over defined windows, we were able to target our benchmarks within those specific trading windows. That also helps us to understand if we’re trading at optimal times, because we’re able to compare our benchmark slippage across the trading day and see if in certain periods we’re seeing higher slippage or lower slippage. This has helped us fine tune our FX trading.
The impact on trading, and the traders
We’ve seen a meaningful improvement in our slippage and our executions by having the traders really hands-on, in terms of monitoring slippage and making recommendations to our research group, such as modifying the times that we’re trading some of the currency pairs.
We leverage a lot of knowledge to try to get better in other asset classes. Initially we saw good improvement in slippage in our futures trading and we applied that same methodology to FX. We’ve definitely seen the trading job evolve. When everything was done over the phone, it was a much more manual process. Now that we’ve got this proprietary platform that allows us to monitor slippage in real time, it gets more granular for the traders to understand how an algorithm is trading. There is also greater leeway in the type of parameters traders can tweak when setting up an order, because of their past experiences doing similar trades. By looking at their impact they can make a decision based on what’s going on in the marketplace. If the trader determines that the algorithm should be a little bit more passive or a little bit more aggressive, they can make appropriate adjustments. If the trader needs to stretch or change the window, they can within certain parameters because, at the end of the day, everything is about getting the best execution for our investors.
Our real focus is on experienced traders that understand the market and have old school knowledge but also a deep understanding of electronic trading, slippage management, how the algorithms work, and how the parameters within the algorithms work. We are effectively pairing an experienced trader with those electronic trading tools, optimising those tools and monitoring the results in real time to get the best possible execution.
Always advancing
Trading overall is going electronic; all traders have to evolve and adjust. If you’re an old school trader whose only skillset is the ability to pick up a phone to get a trade done in a block, you’re unlikely to be successful long-term in tomorrow’s world. Clearly there’s a need for people who have that ability, but even on a desk where they do a lot of block trading, you also need people that are able to select the right algorithm, set the parameters and be more sophisticated in the way they handle an order.
With the evolution of these trading tools, it’s difficult to see a world where one size fits all or one solution is the best for all trading. There could be times during a holiday period where maybe it isn’t appropriate to use an algorithm. We give our traders flexibility on the execution front to look at the marketplace, to look at the tools they have, at the order itself and make the best decision as to how to handle that order and get the best execution.
This is why our proprietary platform is so important, because you don’t ever want to get into a “set it and forget it” mentality. You want to do your best to select the right execution strategy, appropriately set the parameters, and then monitor it so that if it’s not trading in an optimal fashion you can adjust and take action as needed.
Once people start to trade electronically and they see the efficiencies that it can bring to a trading desk, the next step is to push deeper and utilise even more tools that are available. People are exploring the algorithmic tools that are out there. That is leading to more algorithmic usage, which is what the surveys have shown and what we’ve heard from our brokerage counterparts.
We’ve always pushed for innovation on the electronic trading front and feel it is a necessary tool in order to provide best execution for the end investor. With the markets and trading tools constantly evolving you’ve always got to be looking at what’s new and where else you can advance so that you’re always at that cutting edge of what’s available to help deliver the best results for your clients.