Algorithmic Trading Redux

Nothing significant happened since I last posted here, the world of algorithmic trading appears to have morphed from one marked by a frenzied discovery into a world of glossy marketing and brand hype. Dotted with colorful strategy names like sniper, dagger, sonar, Tex and the likes! One has to wonder if the WWE has taken over promotional duties for the industry!

Nothing changes the fact that for the most part, institutional buyers and sellers have one of two primary objectives on mind; either to beat the "closing price" or outperform some "benchmark".

Obviously, successful strategies for beating the close are unlikely to be found at the other end of a broker connection, which leaves the option of trying to beat a benchmark. However last I checked, even that has been quite the elusive task for most brokers and DMA buysiders!

I spent the past year adapting my work on "strategic" or alpha generating strategies into "tactical" or best execution variants. Primarily driven by my strong belief that "capital commitment" is truly the last remaining area of growth left for the sell-side. One of the true measures of a modern desk's ability to deliver execution value to the buy-side.

That value manifests itself in the ability for desks to take on principal obligations such as "blind baskets" and dissipate them in the market on behalf of their clients. Algos taking on the brunt of that effort and unless they can consistently put numbers on the right side of the ledger, those services won't be offered for long.

To that end, I have since migrated commentary and musings on the topic to the VI4 blog. A place to showcase third-generation** algorithmic execution strategies...those redesigned mousetraps I keep referring to.


**Single purpose strategies designed to beat the "benchmarks"

Have algo BlackBox, will travel..

Excerpts are from an article originally published in "Wall Street Letter" that can be found on the globalcustody.net website here, reads:

To curtail costs, brokerages are likely to cut trading staff and pour more orders into automatic execution engines on program desks. Josh Galper, consultant with Vodia Group, said that use of algorithmic trading could actually approach 50% of all trading volume for sell-side desks. "Brokerages would receive orders via phone, but if there aren't enough traders to handle them and the orders aren't too complex, they can be routed to an algorithm through the program trading desk," he said, adding that the practice already occurs at several large sell-side desks.

Can't say that I'm surprised because from where I'm sitting, even active portfolio traders are constantly reaching over for their firm's algorithmic servers to take a load off their shoulders. Unless the trader can work a specific name against an active block, the best they can hope for day-in and day-out is to beat VWAP. Doing so for a handful of names might not be all that hard but given access to a good VWAP execution engine, there would be no contest.

But sell-side trading desks won't disappear anytime soon, though they may be whittled down. Sales traders on the desk will be fewer but more sophisticated, said Harrell Smith, analyst with Celent. "The buyside's understanding of algorithms is still imperfect, so the relationship with the algorithm provider will involve a lot of hand-holding for awhile," he said. Already, brokerages are hiring traders with more mathematical and quantitative backgrounds than just trading experience, said Michael Karp, founder of search firm The Options Group.

Quite the elusive task because a good algo which is defined as an alpha-generating piece of trade-execution logic, ultimately becomes nothing more than a system that's designed to beat all other market participants on any given day! A system that consistently (more than 50% of the time) beats VWAP is a system that beats the house! That kind of a system will gravitate towards the domain of "principal-traders" rather than "agency-execution" desks. While, hiring a sales trader who can code a trading-algo in your favorite portfolio trading platform might seem like the way to go, chances are that the piece of logic that constantly eats his lunch, will almost always be developed by the guy trading the firm's own capital from the backroom or a quantitative trader working on a buy-side desk somewhere in Greenwich.

Investment manager, seeking broker-dealer firm, must be big on analytics, algorithmic trading and above all "commitment"

Capital commitment, that is.

Capital commitment or the willingness of the sell-side to use its own money to take on a buy-side position that it would then attempt to principally trade away over a longer time frame, has traditionally been viewed as a premium service typically reserved for a select list of clients.

Like many others, this definition is going through a fundamental change. As access to decision-making information barriers fall, thanks to everything electronic, so will agency margins. Or so the conventional wisdom goes! With every facet of sell-side services under assault, the one value-added-service that has shown some resilience, is the sell-side's ability (willingness) to "stimulate" liquidity.

Capital commitment in its modern sense (portfolio acquisition and re-distribution services) is one key area that distinguishes the capabilities of different sell-side desks. Examples abound of traditional sell-side powerhouses' move towards committing more of their own capital to principal (proprietary) trading. Most recently this article appearing in the August 2005 issue of the Bloomberg Markets magazine, is quite telling.

Hence my true measure and probably the best proxy for "Quality of Execution:" the sell-side firm that consistently shows willingness to take on "blind bid baskets" is the one most capable of achieving "best execution."

The rationale here being that capital commitment in the form of blind-bidding, means more risk and additional inventory that needs to be internally-absorbed (principal positioning) or re-distributed (algorithmic mastery) and only those broker-dealers with the best channels, savviest traders and sharpest execution algos, can and will constantly show willingness to take on more capital commitment initiatives.

It also goes to show that these are firms that have long realized that the "vanilla institutional agency" business has forever changed.

DMA, shoring up the messaging floodgates

For anyone having doubts that messaging "network strain" is a major concern, this new exchange rule, could be an eye opener.

In essence, the CME's Globex complex, long considered at the cutting edge of electronic-matching-engines, is feeling the strain. A welcome price to pay for the success of the E-mini and other electronic contracts that have been showing steady growth for the past 7 years. The problem stems from the fact that more and more sophisticated front-end trader tools have been written to capitalize on the Globex API and its underlying infrastructure. To this end, the CME is proposing the creation of ratios or messaging-benchmarks that throttle the amount of modification (CFO) and cancellation (CXL) messages as a percent of matched orders. One preliminary example for the E-mini S&P 500, would be a ratio of 10 to 1 (or a limit of 10 CFO and CXL messages for every trade that executes). Values outside of the ratio bounds would incur higher fees (reads penalties). Those will either get swallowed (highly unlikely) by or get passed down by the brokers onto their DMA clients.

Employing similar rules on equity exchanges is most likely being entertained but quite unlikely to see the light. Equity traders engaged in statistical arbitrage, use algorithms that are constantly adjusting their orders for active long/short or pairs-trading strategies. Same goes for most smart-order-routing algos that handle size by seeking liquidity on different venues. In essence, a great deal of trading logic will be adversely impacted by instituting curbs on the number of modification and cancellation messages that one can economically send.

All of this is pointing to the increased cost of maintaining an infrastructure that can keep up with the tide of smarter systems and their aggressively active algos. A problem that will not be going away anytime soon, if anything this is just the tip of the iceberg.

DMA, staring down a double-barreled shotgun

The culmination of that tidal wave that was unleashed upon our industry some 10 years ago with the advent of FIX, ubiquitous networks and electronic access, boils down to this; DMA (direct market access) is the new game in town and everyone wants to play.

In an industry that is constantly searching for a middle ground and a better definition for the term "value-added," edgy, technologically savvy and alpha-hungry buy-side players, aka hedge funds, are aggressively bending the delicate balance of the intermediation landscape.

The harsh reality is that small to mid-size sell-side firms that don't have the budgets or IT resources of bulge-bracket heavyweights are facing the significant double whammy of increased infrastructure strain coupled with cutthroat DMA commissions. Forget getting paid "one" cent per share when half a cent is all the rage and a quarter per or less not that far behind.

Hedge funds especially ones engaged in statistically flavored trading strategies, generate more than fifty times the daily messaging traffic that a traditional manager does. FIX networks are being pushed to their limits and with some 1300 outstanding custom tags and more in the pipe, even the protocol itself is beginning to show signs of excessive bloating.

Today, if you are a sell-side firm where institutional agency commissions constitute a major chunk of the bottom-line, these are trying times and unfortunately things are not likely to get any easier. Firms everywhere are crunching these new numbers and adjusting their expectations accordingly. However no one really has the stomach to account for the unavoidable impact of a general slowdown in trading activity like the one that followed the dot-com bust.

Reality is, players on both sides should be bracing for that inevitable headliner that will most definitely sour the pot for the hedge-fund industry and everyone else that it touches. We've already witnessed a couple of shots-across-the-bow here in Canada and the U.S. however those were not significant enough to trigger the mainstream panic-alarms, yet. With some 9000 hedge funds operating in North America today and an estimated trillion dollars in assets, one should expect that the next LTCM type story would cause a significant and long drawn commission drought.

For anyone on the sell-side, struggling to make a good case for supporting the new DMA return on infrastructure business model, factoring in a worst-case-scenario for a potential top-line negative impact of anywhere between 30% and 40%, could be downright disheartening but nevertheless very well advised.

Block Crossing Networks, and one bites the dust!

Not to say that my previous post was prophetic or that I had an inkling that this (story) was in any way eminent, which I didn't, but hey the post does appear to have been perfectly timed even if it was all by pure chance!

In essence, Harborside+ or Liquidnet's closest competitor has seized its trading operations on Friday May 13th and is currently being investigated by the SEC. I'm not saying that this would become the fate of everyone in this space rather that the concept of block-crossing-networks will become less significant as more buy-side firms get their legs in algo-trading shape.

It's ten o'clock. Do you know where your ECNs are?

Instinet buys Island, forms INET
NASDAQ buys BRUT
NYSE buys ARCA
NASDAQ buys INET
Knight buys Attain
NYSE buys Knight
NSX buys NexTrade
NASDAQ buys NSX

Okay so the last three are still figments of imagination but aren't we coming back full circle to the oligopoly of yesteryear? Maybe Reg-NMS will have nobody left to regulate after all!!

Smarter, smart-order-routing

One of the algos that I frequently like to re-visit is the one dealing with Smart Order Routing (SOR). SOR is supposed to be one of the simpler, more straightforward trade execution algorithms, or is it? Reality is, SOR algo logic could be more elusive than conventionally thought. The key problem is that "smart" means different things to different participants. Chances are, if you use a sell-side SOR Algo for unmarketable orders, the algorithm will most likely be skewed towards an avenue favored by the algo provider for reasons that might not be readily obvious.

I spent some time evaluating SOR algos offered by sell-side firms for the routing of listed-options orders. I chose options orders because they are slow enough to allow for easier tracking and dispersed enough to generate meaningful realtime observations.

In the U.S. today, active option classes trade on six regional exchanges (the AMEX, BOX, CBOE, ISE, PHLX and the PSE) with the ISE, PSE and CBOE being the heavy weights in active names. I evaluated three different broker sponsored SOR algos for oversized limit-orders and found that for the most part they exhibited similar behaviors and wherever they diverged there was some explanation related to their provider's involvement with a particular execution avenue. The basic logic considers the prevailing spread and size then divvies up amounts accordingly. In some cases, that divvying up takes on an air of rationalization or as I would like to call it "tilting," towards one liquidity source over another.

Everything else being the same where multiple avenues exist you will always come across cases where "time priority" will not be observed for a smartly routed order!

Actually I think tilting is good only if one gets a say into how things get tilted. Which brings us back to why the buy-side needs full control over the Algos they use. Typically, for unmarketable orders, I would throw in extra weighing points towards avenues that show more aggressive market making traits. For instance, the price improvement feature (PIP) offered by BOX would get a heavier weighting (from me) over other avenues with deeper books. I would also tweak the weighing to favor tighter spreads, better bid/offer size ratios and relative participation. In essence I would not treat my SOR as a black-box rather an "adaptive" system, albeit the adaptation in this case is quite discretionary and not rooted in some artificial intelligence logic.

SOR algos are a "must have" for any OMS worth considering, furthermore SOR algo logic must be fully disclosed and allows for user-definable parameters with the ability to dynamically tilt the relative weight of the routing rules based on acquired observations.

In this paradigm, the trader gets to express his/her expertise into re-usable algorithmic rules and that's exactly where we want to be in this brave new world of algorithmic trading..

Wouldn't you please scratch my block-trading itch?

It's no secret that I'm not a big fan of electronic block-trading networks, simply because I believe that their existence goes against the core foundation of the medium that they exist within. Actually here and now would be the perfect place and time to declare that we probably have seen the zenith of those networks as an idea that has reached its prime and things should start pointing downhill from here.

Crazy talk you might say, with Liquidnet easily sailing beyond the billion dollar cap mark and venture funds salivating at the prospects of getting a piece of that action, you must be quite the fool to even go there!! Hey someone somewhere had to call the top and I just did.

I'm doing so for two reasons, one of them is as obvious as they come and that is; in my book of market wisdoms you must always, always bet against a closed-garden-variety incumbent of any industry that was built upon the concept of open access. I can't remember how many times I argued against the prospects of companies like AOL, CompuServe, Prodigy and the likes, as resellers of private-island-lots in that oversized pond called the web. You can chalk it all to basic island-phobia but even back when AOL had more followers than the Catholic Church, I lost my voice telling everyone that their business model was definitely going the way of the dodo! Like buying frenzies and pyramid schemes, that concept ends when you are the last one standing in line.

The second reason has to do with the concept of disruptive technologies and their impact on progress. Crossing-networks should for the most part be viewed as a disruptive technology. Or by definition a technological compromise for a bigger problem. These networks do not represent a cutting edge concept and they most definitely do not represent a conceptual advancement in how capital markets should work. They assume a vacuous existence where close neighbors engage in a perpetual garage-sale moderated by the enterprising family up the street. Sooner or later everyone has to take a drive to the mall and contend with elbowing the common folks for a spot in a cashier line.

It used to be that advocates of those networks would almost always invoke the "e" word (as in eBay) in reference to how auction markets should be like. They don't bother to reflect on the fact that it didn't take eBay long to realize that as a "network" it could not survive and only as a true "marketplace" it did.

It's hard for crossing-network fans to explain that for the most part they are driven by fear and contempt. Fear of those prying vultures and contempt for all of those liquidity disturbers, day traders and odd-lot scavengers. Whether through apathy, complacency or ignorance they forget that they are as much responsible for how the market behaves as the next guy!

Electronic block trading is an itch and no matter how hard or long you keep on scratching, if you like the outdoors, you still have to contend with those pesky mosquitoes.

As for the networks, you can never discount the abilities of luminaries such as Liquidnet's Seth Merrin to realize the fragility (measured in market years) of his incarnation. However, unless he can pull that mainstream transformation from a single-purpose-network into something more organic like say a full-participation ECN or maybe even a bona fide exchange, he would be well advised to borrow a page from the AOL book of tales.

Ask not what your broker/dealer can do for you; ask what you can do for market liquidity

Today's Report On Business magazine (an insert in the Canadian national newspaper The Globe and Mail), has a glossy plug for Markets Inc. the Canadian answer to stateside block-crossing-networks.

The piece starts with the all-too-familiar fable about how a manager for one of Canada's largest pension funds accidentally witnesses one of his supposedly anonymous blocks being shouted about on the trading floor of a broker dealer. A prelude to the solace and justification for why his fund became a key backer of Markets Inc.

That argument aside, the fact that the article incorrectly bundles crossing networks with traditional ECNs as one and the same with thrown in nationalistic references to home grown versus imported technology and...., you get the picture.

I do have a problem with block-crossing networks in general but in the case of Markets Inc. in particular, I believe that this is a company that appears to have been conceived and designed from the ground up for the sole purpose of becoming a buyout target. It brings zero real innovation but lots of inner-circle pedigree to the table. Positions itself as a thorn is someone's side (The TSX, traditional broker-dealers, U.S. crossing networks) and deftly slips onto the scene ahead of Canadian subsidiaries of established U.S. entrants (Liquidnet and Harborside+). Markets Inc., will work the regulatory hurdles, sign up some choice buy-side names, grab a pittance of liquidity, ensure it becomes the Canadian poster child of alternative networks and wait. Wait for the offers to come in.

Zero innovation, limited contribution to the dynamics of Canadian capital markets and one more of the closed-garden-variety clubby networks to contend with.

Capital markets especially in Canada, desperately need more liquidity injected into the system not three new single-purpose electronic players competing for the upstairs market!! To their defense, traditional broker dealers take those same block orders and for the most part throw big chunks of them back into the liquidity pool. Electronic crossing networks do not. In the U.S., no matter how big, companies like Liquidnet become, their impact on the liquidity pool will always remain tolerable, in Canada a similar impact will be profound.

Buy-side firms and large fund managers should realize that in capital markets as it is in waste management, recycling is their only hope for longevity!

It's all about Transaction Cost Analysis you fool

Transaction Cost Analysis is the new 'cause célèbre' for alpha-conscious buy-siders, their trading counterparts and freshly minted CFAs, the world over. More and more buy-side firms are asking for and getting access to, broker tools and TCA models, without having to make any commitments towards funneling trades back to their providers! They get to cherry pick the analysis and go with the most cost effective execution avenue at their disposal.

As for the brokers, the rush to expose their proprietary TCA models and algorithmic trading tools to the buy-side is a sure sign of how desperately they are trying to retain their client base. Fully suspecting that once the knowledge is transferred and with their own set of tools in hand, those same clients will ultimately be driving towards DMA sized commissions!

The question to ask here is this; if a buy-side desk had access to working execution strategies and direct connectivity to the sources of liquidity, would they be paying their brokers any extra for trade execution insight?

Broker sponsored TCA will remain all the craze until the buy-side feels comfortable enough walking around in their own sell-side-imitation shoes. Until then, if they can't gleam meaningful trading ideas from it, they can always swing TCA about as the menacing performance-evaluation stick it was meant to be.

Meanwhile and in absence of a meaningful payoff, the quality and performance of the free tools will degrade markedly which brings us back to the inevitable conclusion that the buy-side has no real choice but to develop their own tools and original ideas.

As for the sell-side, agency execution revenues and the cost of supporting DMA clients have nowhere to go but sharply down for the first and equally up for the second. However that is a topic for another post...

"Algo" a four-letter word!

Algorithmic trading is all about using systems to beat other market participants on any given day. Managing shortfalls, masking size and fulfilling a TCA mandate are all parts of the same argument. Incidentally, that is the same guiding principal behind (gasp) systematic day-trading!

Picking mainstream vanilla algorithms such as VWAP, TWAP or time-slice does not truly define a sophisticated algo trading strategy. Similarly, a smart-order-routing algorithm should be viewed as nothing more than basic functionality that has to be standard fare in all order-management systems and/or broker sponsored applications.

True algo trading is all about constantly refining adaptive systems to capitalize on changing market conditions and dynamic execution criteria. At its simplest form, the ability to combine two or more mainstream algos into an execution strategy, should be the norm. (example: apply VWAP between 9:35 and 9:55am then use a time-slice algorithm until 3:25pm followed by a VWAP algo into the close)

The point I'm trying to emphasize here, is that buy-side do-it-yourself practitioners need more control over their algo trading than simply using a pick-list of canned offerings from their broker or OMS vendors. Trading algorithms must be viewed and treated as unique intellectual property of a profound strategic value. No different from the value placed upon the talents of traders on a buy-side desk.

Others might just want to continue pegging their orders against popular benchmarks and paying their executing brokers for those services.

A "Buy Side Story"

In my previous posts, I briefly brushed upon the genesis of algo trading and opined on the probability that most buy-side desks will inevitably adopt a workflow utilizing some form of algorithmic strategy.

To better qualify this argument, a definition of the term "algo trading," is in order: "Algorithmic trading defines any systematic execution logic applied to an order at any point from inception until it's acknowledged by a source of liquidity."

To further define buy-side algo trading, we need to distinguish order types against a backdrop of broker interaction:

1. High Touch Orders: Are orders that are sent electronically to the sell-side with the expectation that a broker/trader will use discretion to add value to the execution process.


2. Low Touch: Are electronic orders that are subjected to some form of a sell-side algorithm (e.g. smart order routing) or the possibility of some form of human involvement that might introduce latency into the execution process.


3. No Touch: Are electronic orders that flow uninterrupted from the buy-side to a source of liquidity. Algorithms if any, are applied before the order is acknowledged by the sell-side.

No Touch orders are widely referred to as DMA (Direct Market Access) orders. Generally speaking, DMA orders are favored by the "Do-it-yourself and don't pay the middleman more than you have to," aficionados. Those are typically hedge fund managers and buy-side desks that are actively managing their execution costs, among other things. Interestingly enough, DMA started its life as a retail-trading feature (which to some degree, blurs the lines between the No and Low touch definitions since most brokers do apply some form of trading-limits and/or margin-validation logic against DMA orders).

Based on the definition above, I would also bundle buy-side crossing networks into the algo genre (others might disagree). From a buy-side desk perspective, traders would ideally expect their OMSs (order management systems) to provide for the ability to: a) Manually work an order (good old phone trading), b) Anonymously locate and allow for crossing a good sized block and finally c) Slice/carve/sweep orders based on programmatic rules and/or human intuition.

Last I checked on buy-side OMS capabilities all three could possibly be achieved with a great degree of jigging and duct-taping of several pieces that are not fully integrated. Ultimately somebody somewhere will come up with a decent buy-side broker-independent OMS that can keep up with the changing landscape. In all likelihood this system will have to borrow a great deal from a sell-side equivalent, which favors a complete unknown on the current buy-side vendor list (e.g. here). For the most part, the traditional buy-side vendor heavyweights are not being aggressive enough and are mostly relying on strategic relationships with brokers (here) and (here) rather than grabbing this bull by the horns and getting on the leading edge of this wave. Shades of a possible repeat to how tentatively then frantically they reacted to supporting electronic (FIX) trading not so long ago.

Furthermore, it's my opinion that to have a successful Low or No touch algo trading strategy, practitioners have to get into the habit of creating or at least tweaking algorithms into their own unique trading ideas. Canned or broker provided algorithms, will most definitely produce the exact opposite of the desired effect. For the most part they should be viewed as nothing more than analytic or marketing tools meant to attract order-flow business. Accordingly, the next wave of upheaval on the trading desk will become tilted more towards techno-savvy traders. Traders who can program their systems to do more than direct orders from point A to point B. How technically oriented becomes a function of how well designed and algo-scripting-friendly the next generation of OMSs prove to be....

Algorithmic trading and the VWAP Trap

The foregone conclusion that most buy-side firms will eventually get into the practice of algorithmically carving their own trades, is bound to change the rules of the game, yet again. To put it in the parlance of the street; doing a higher volume of no-touch and low-touch transactions, becomes the cornerstone of their bid to win the “Transaction Cost” game.

The expectation here is that buy-side Order Management System (OMS) vendors will be under increasing pressure to support algorithmic trading functions or risk facing obsolescence. No different from how those same vendors scrambled to adopt electronic messaging standards, long after they have become a staple of the sell-side bag of tricks.

Unfortunately history has shown that for the most part, buy-side order management systems, almost always leave a great deal to be desired. To their defense, those vendors are increasingly finding themselves being asked to match features that are traditionally reserved for the sell-side equivalents.

In their efforts to stay ahead of the performance curve, the buy-side will instinctively start bringing algorithmic trading in-house, putting added pressure on the viability of the agency role on the street. Lost in this picture is the realization that one of the biggest problems facing the adoption of algorithmic trading is not under but rather over-utilization.

Essentially, if everybody used standardized or highly correlated algorithms, they invariably become self-defeating.

Buy-side proponents of do-it-yourself algorithmic trading, are well advised to delve into the “art,” only if they are willing to invest in the “science” of it all. As the saying goes, a little bit of knowledge could be a dangerous thing and this is one area where less than mastery of the craft, will most certainly produce less-than-desirable results.

Prologue

In a research brief titled "Algorithmic Trading: Brokers Race to Mediocrity" that was published in Securities Industry News, October 2004 by analyst Randy Grossman (Financial Insights), he concludes that Algo trading has not yet lived to the hype of delivering transaction-cost savings to the buy-side desk. He further argues that this re-affirms the value of block trading specialists such as Liquidnet, Pipeline and Harborside+.

It's my opinion that the fundamental problem with Algo trading has to do with its synthesis from broker-performance-benchmarks to better-execution-mechanisms. VWAP or the poster child of algo trading began its life as a broker evaluation benchmark which was meant for aiding buy-side desks in their evaluation of their brokerage counterparts. Trade execution algorithms are a classic case of how the sell-side managed to adapt and quickly learned to benefit from the big stick of broker evaluation benchmarks. In essence, they took those same benchmarks that were designed to keep them in-check and ever so enterprisingly turned them into one of the better arbitrage opportunities of late.

By guaranteeing a performance pegged to a benchmark, not only did the sell-side appease the money manager, they also succeeded in creating valuable liquidity within their own pro trading books. To top it off, they also commanded beefy commissions for this premium service.

By demanding a pegged guarantee, the buy-side has succeeded in short-circuiting the agency relationship and placing their brokers on the same side of the market they were on. It is no wonder that traditional agency trading is taking a backseat to the more lucrative practice of principally trading institutional pegged inflows.

It won’t be long before the buy-side comes to the conclusion that if their brokers are making a tidy return selling execution guarantees, then they are most certainly not realizing their coveted “best execution.” Somewhere along the way, a healthy premium is being left behind..