Last week, Exegy added intraday signals to its AI-powered iceberg order detection offering, Liquidity Lamp. The enhancement provides quantitative traders with visibility on the volume of iceberg orders throughout the day using summary files delivered every ten minutes.
Traders Magazine spoke with Andy Lee, Director of Quantitative Research at Exegy, to learn more.
What is the background of Liquidity Lamp?
Liquidity Lamp first came to market in 2020, as a real-time product for customers who consume data via the Exegy Ticker Plant.
Liquidity Lamp is originally a product that detects when a reserve order is found on US equity markets. Once we find it, we can track it in the order book and also the price book. But a lot of customers came to us and said, “Is there a way that a mid-frequency, or even a low-frequency product can be derived from that HFT signal?” That’s when we came out with our initial end-of-day product.
The end-of-day product basically provides a summary by tracking the reserve orders. It allows us to note that a particular order was traded against at this exchange for this symbol at that time of the day, and we can roll that up and tell customers really easily what symbols, how much volume, and how much notional value was traded on any given day.
That product really galvanized Liquidity Lamp, because it was the buy-side firms that were looking for alpha and looking for unique, differentiated data sets that were uncorrelated to anything they were already working with. So that’s the storyline of how Liquidity Lamp became a success.
What is the essence of the new offering?
This new product bridges the gap between the real-time product and the end-of-day product. We had been thinking, “Is there a way we can deliver the customer the same end-of-day summary file that’s easily consumable, but every 10 minutes instead of end-of-day so they can action something when it’s most tactically opportunistic rather than having to wait until the next day?” This is the answer.
Who are the core users of Liquidity Lamp?
It’s quantitative, systematic funds that are looking for statistically significant data points and using Liquidity Lamp in AI and in their models. Other users would be stat arbs, mid- to low-frequency intra-day traders, and multi-day traders. The product is not geared toward fundamental investors.
Do you expect some customers who use the end-of-day product to switch to this new product?
People have different objectives. Some customers have strategies that involve accumulating information across multiple days, weeks, and months, rather than intra-day; they will still be happy with the end-of-day product. But other customers want to deploy this more on an intra-day basis. They may see a large iceberg order trade in a given name, and they want to take advantage of that information intraday, because that information may not be helpful tomorrow. So it just opens up the opportunity for new and more strategies to use this, versus the end-of-day.
Comparing Liquidity Lamp real-time with every 10 minutes and end of day, what is the difference in ‘lift’ from the user’s perspective?
Real-time is infrastructure-intensive – you need to purchase a ticker plant and consume full-depth market data feeds. End of day is the opposite – we just give you an AWS [Amazon Web Services] free bucket to grab the data files from, which you can do with commercial internet access on your laptop from the basement of your home.
The new intraday product has some infrastructure necessary because it is consuming real time market data, but it’s mostly distributing out to AWS. It’s a hybrid approach, but we take up a lot of the infrastructure cost and try to make consumption very easy for customers.
What’s the most important takeaway from the associated whitepaper?
One point is that we had some early adopters of Liquidity Lamp come back to us and say this is beneficial as a creative product to our existing P&L. Meaning, instead of creating brand new alpha, what it does is enhance our existing strategies. That was the draw.
Because of that draw, in the whitepaper, we took the early adopters’ approach. We said, “Let’s design a basic, plain vanilla stat arb strategy and then inject that baseline model with Liquidity Lamp information to see if it creates a better portfolio, a better strategy.”
And that’s exactly what we found. It reduces draw downs, it increases returns, it increases your Sharpe ratio and beats the S&P 500 on a relative basis, but it also defeats the baseline stat arb model on an absolute basis, which we were really pleased about.
Final thoughts?
We’re not traders at Exegy. From a customer perspective, that is an important consideration. Rather, we’re trying to take tools that are closely guarded at most HFT shops and democratize those signals so that more people can access this information to create their own alphas and improve their strategies.
Catching Complex Cross-Product Abuse Calls for a Trading Floor Mindset
By Mike Coats, CTO, TradingHub
In the analog days of trading, regulators were concerned with finding instances of the good old-fashioned pump and dump, insider trading, or market timing schemes. But digital transformation has reimagined the entire financial system, and while some bad actors have simply digitized legacy forms of market manipulation, most have used technology to originate new, more intricate forms of abuse. For banks, this means that the trade surveillance problem is now more complicated than ever before.
Regulators turn a keen eye to the cross-product problem
As markets have become more complex and interconnected, the potential for illicit activity has expanded. This is especially true when it comes to the opportunity for market manipulation presented by ‘cross-product’ trading patterns – correlated activity that spans multiple asset classes and trading venues, often across both lit and OTC markets. But no matter how complex this trade surveillance problem is, regulators expect market participants to have a handle on it. We can see this message clearly in the large number of enforcements in this space in recent years.
Despite the capital markets industry investing heavily in compliance and surveillance capabilities in the hopes of avoiding regulatory hot water, recent high-profile fines against Bank of America, JPMorgan Chase, NatWest, HSBC, and others, indicate that these efforts are falling short.
So why is cross-product abuse slipping through the gaps of some banks’ preventative systems? And what can be done to safeguard against the huge financial and reputational risks of regulator action?
Operating cross-product and cross-venue
The truth is that the market abuse problem is a web that’s spread across the entire capital markets. Today, virtually every sell-side trading desk makes money by employing cross-product strategies, often laying off risk arising from illiquid client-transactions with benchmark or listed products. Trading floors approach their activity by thinking in terms of ‘market risk’, and as such they trade across a variety of securities, products, and venues to manage that market risk, while calculating the relationships between various assets.
This means that taking a single-venue or single-instrument approach to trade surveillance – which is the default at a lot of banks – is no longer sufficient. These systems aren’t equipped to identify the instances of abuse which present the highest regulatory risk, because they’re not looking at the cross-product picture. Because today’s riskiest forms of market manipulation are happening across various interconnected places that traditional approaches are not capable of spotting, banks are left with costly blind spots.
Surveillance models now need to understand the relationships between instruments so that they can assess pricing impacts to indicate the intent behind any trading pattern. Buying one instrument and selling a related one – albeit elsewhere on the markets – is a position in the spread between the two, not two independent positions, so looking at any transaction independently misses the full picture.
An interconnected web
For example, in fixed income a nefarious trader might use a series of trades in a liquid instrument like government bonds to move the price of a much less liquid instrument like certain over-the-counter (OTC) derivatives. Many fixed income products are interconnected, therefore effective surveillance must consider trading activity in correlated products — such as cash vs. futures, or products with different durations.
Every single bond from government to corporate, and every single product from bond futures to swaptions are expressions of interest rate risk. This means they will manifest significant correlation in their price evolution. These ever-present relationships between different instruments are the very reason that traditional trade surveillance methods have proven inadequate to police such complexity.
What’s more, because trade surveillance approaches are so out of date based on today’s markets, any bank monitoring trades in isolation is now going to be dealing with an onslaught of expensive false alerts. This monopolizes a compliance team’s time, yet the bank will remain incapable of capturing the severe cases of wrongdoing because the cross-product picture is not being accounted for or it’s lost in the pile of false positives.
Rules-based efforts are futile
Traditional trade surveillance systems are using rules based approached to detect instances when a trader is trying to artificially move the price of an instrument to his or her advantage by executing trades or placing and then cancelling orders.
This approach reflects more straightforward equities abuse detection tactics and has resulted in institutions deploying technology platforms that flag every potential instance of manipulation based on lengthy, rules-based taxonomies that can include 60-70 or more categories of market abuse types. The result is an unmanageable deluge of alerts that flag any potential instance of market price movement – which costs millions of dollars just to investigate.
Yet despite most trading desks operating using cross-product strategies today, only one of those 60-70 categories is likely to be a row labeled cross-product abuse. This implies that cross-product abuse is still a rare and niche risk, instead of one of the most common market manipulation strategies that it has become today.
A different surveillance POV: using a trading floor way of thinking
There is no question about it – trade surveillance must adapt. And this adaptation cannot take the shape of simply bolting on additions to chronically unperforming legacy technology. To truly protect against the costly fines and reputational damages of regulatory action, the entire system must be reoriented to reflect how traders actually think, behave, and trade today.
Trade surveillance systems must think in the same way that those on the trading floor do: in terms of risk sensitivities, not rules. A bond and a bond future are basically the same expression of risk, although they might trade in a different way, over a different venue. In both cases, by buying or selling this product a trader is forming an opinion on what’s going to happen to interest rates.
Trades in different instruments are, in most cases, likely to form part of the same abusive strategy, so rather than treat them individually, a surveillance system must intelligently group them together.
Asset classes which require risk-based methodologies like OTC derivatives or fixed income products, are impossible to monitor using traditional surveillance methodology. But thinking in terms of holistic market risk can map how a trader’s positions across a combination of instruments and across a series of maturities are all linked, thereby determining the intent of a behavior and whether manipulation is likely to have occurred.
Market-risk models are able to put the trader’s mindset at the forefront to enable surveillance teams to capture hidden signals that traditional trade surveillance systems are blind to.
This approach not only reduces false positives dramatically, but it enables compliance teams to focus their resources where they really matter to protect against the growing reputational, financial, and legal risks of the most prevalent market manipulation types today.
It’s time to expand the trade surveillance lens
Our industry has built trade surveillance by thinking rules upwards instead of market risk downwards, resulting in the worst kind of complexity.
It’s like trying to spot a shooting star by looking through a keyhole – you’re not likely to see it, because you’re not looking at the whole sky. Now is the time to reimagine the trade surveillance approach so that banks can significantly expand the lens they’re looking through and capture the alerts which genuinely demand their attention.