CHAPTER 8

Losing $30,000 in Under a Minute

As I hit my stride in high volume scalping at Swift Trade, I gradually became increasingly comfortable with the idea of my per-trade size increasing. The trainees who insisted on attempting to focus on Bollinger Bands and MACD indicators instead of live Level 2 quotes and prints on the consolidated tape had begun to drop off like flies.

By then, I had accepted the extent to which I could effectively manipulate my own psychological tendencies. I had long admitted to myself that I couldn't truly change my own nature any more than I could fly—I couldn't force myself to feel differently or react differently after a winning trade or a losing trade. I could, however, force my mind to focus on specific ideas that I had discovered to effectively refocus my reactions to each of those situations. The most obvious example of this was that I managed to somehow reinstate in myself the controlled feeling of just the right amount of pressure that I had felt during the first week of training when I felt under pressure from the no-charts rule. In hindsight, a different strategy for this might work for others or a similar technique might also work. I can't claim to have done any further research on this besides the fact that it seemed to work very effectively on my own psychological predisposition at the time. As long as the bottom line remained green day in and day out, there was no real reason to change it at the time.

Meanwhile, Jack and another senior trader, Gordon, had begun their own experimentation phase of a different kind. Having already mastered the scalping game long ago, they continued to practice it as a core trading style and also began to branch out into other strategies. They had effectively treated their high volume scalping as their base income while augmenting it with new strategies as potential bonus pay. If they had failed with these strategies, the profits from their core method would at least hedge them—and the company—from any significant downside.

By this point in time, they had stumbled onto ideas from the senior traders at another branch of the company that centered around methods to capitalize on the relatively primitive algorithms (compared to those in use as of this writing) that had been employed by the dark pool algorithmic systems at the time. Recall that the core purpose of all dark pool platforms is to facilitate transactions of large amounts of shares with minimal market impact, which includes minimal visibility on the displayed market centers like the exchanges and ECN systems. On thinly traded, illiquid markets, the actual bids and offers from the sorts of large institutions who utilize the dark pools for their intended purpose tend to sit at prices all around the national best bid and offer, and often even in between. Essentially, the key to a dark pool's success was the ability to allow large traders (in the scale of the entire market), such as hedge funds and other major institutional market participants, to “hide their hand” from the market.

Without these routes, such participants would instantly move the market against their own favor by either posting their exceedingly large bids or offers for others to see, or by rapidly removing what little visible liquidity might be on the public markets, especially for thinly traded illiquid stocks. What the traders at another branch of Swift Trade had passed to Jack and Gordon was an idea that exploited the weaknesses in the dark pool systems that would effectively expose the large market participants' hand in the market.

On a more personal note, it had actually been a big week for me. The previous Friday afternoon, I had officially graduated from the trainee desk with old CRT monitors and that squeaky chair I had become accustomed to. It was my first week at my new position on a chair that was likely more comfortable and luxurious in my perception than it really was. In hindsight, it was almost an executive office chair without armrests. At the time, it felt like one of the most comfortable seats on the planet after becoming accustomed to the squeaky old units delegated to trainees. And, of course, the upgrade to a couple of 22-inch flat-screen LCD monitors with which to stare at NYSE Level 2 quotes and the consolidated tape prints felt like a luxury fit for kings compared to the blurry old tubes I had subjected my eyes to for six and a half hours a day.

As I sat at my desk trading names with high average daily volume figures, which are extremely liquid at all times, the stocks that Jack and Gordon were trading with their dark pool gaming strategy were on the opposite end of the spectrum. These stocks put up very low average daily volume stats, and typically exhibited a relatively wide spread (the difference between the national best bid price and the best offer price would be more than a penny apart at all times). This is typical of most low volume stocks, of course.

George, the only other trainee remaining from my class, took a break from his blurry CRT tube monitors to observe the senior traders with fascination.

“What are they doing?” George asked in a whisper as he approached my new trading desk. “Why would they scalp stocks like that?”

“They're not scalping them,” I answered based on what little information I had at the time about the actual strategy. At the time, I hadn't quite learned the intricacies of the strategy either, but I knew what they were doing with those illiquid names was nothing in the realm of the style we were trained with.

Gaming the dark pools, predatory trading, or whatever name you may choose to associate with the practice, was a very profitable style of exploiting a temporary opportunity in the system at the time. The strategy could be summed up with its main premise: to seek out large orders resting in the dark pool systems and to play the market against the large institutions that entered those orders. The generation of “dark pool traders” at Swift Trade, as they had come to be known among the Canadian proprietary trading firms, would begin by placing small orders onto the dark pool algorithmic routes to seek out potential blocks of much larger resting orders.

For example, Jack would throw out a few small orders of less than a thousand shares each onto a dark pool. His purpose was to locate any possible blocks of larger orders resting on the system that would indicate a large amount of buying or selling interest. This was not risk-free, of course. In this particular case, he would place a few small buy orders onto a dark pool route and had found that his orders were immediately filled. If he felt confident that it looked like a large seller had been in the market on the dark pool system, he would then begin to load up on a short position by swiping up the visible bids posted opposite the big potential seller's likely dark offer price. Effectively, he purposely absorbed (and therefore removed) the visible bid orders from the potential buyers in the market by single-handedly taking out the entire bid side on the market. He simply consumed the liquidity that the large seller would optimally hope would not budge (or at least would move minimally) in the process of waiting for their sell orders to be filled. This is because if the bid (displayed buy limit orders) disappears from the visible market, other market participants are more likely to post offers at the same or worse (lower, from the seller's perspective) prices if the national best bid had decreased.

Recall that the main purpose of a dark pool is to give large market participants better anonymity and minimal market impact, firstly by never displaying their resting orders and second by never showing the size of the seller's offer. If, for instance, a hedge fund is looking to dump a large amount of shares, the last thing the hedge fund needs is for the market to see that there is massive selling pressure opposite a proportionately small amount of buy orders. In theory, everyone with half a brain who actually saw the massive sell limit order posted on a regular ECN or other visible market center would immediately sell as well—either to short sell in anticipation of a large down move or to dump an existing long position for fear of the inevitable down move. All these events would be detrimental to the big seller as all or part of the seller's orders would be filled at lower prices as a result. So the very purpose of a dark pool, from the perspective of such large market participants as hedge funds, is to be able to transact such large amounts of shares without single-handedly causing the market to move unfavorably to the participant's own interests.

The merits of dark pool systems remain a widely debated issue as smaller market participants argue against the opacity that such systems introduce into the supposedly transparent and level playing field of such highly regulated markets as the equity markets. For those who gamed the early generation of dark pools in the U.S. equity markets at the time, however, it was a wildly profitable source of bonus income to pull profits by circumventing these opaque order matching systems.

Often in the process of playing the big players against themselves, Jack would load up with a massive position on these illiquid stocks. He and Gordon often joked that they should become majority shareholders for a few seconds on some of these low volume symbols. In this case, he loaded up a huge short position against a large institution attempting to use the dark pool to fill its large sell orders with minimal market impact. And once he sniffed out their orders, the impact that came was anything but minimal. Based on a couple of small, quickly filled buy orders, Jack aggressively turned around and took the market down a few levels by hitting every visible bid. Naturally, as the bids disappeared, new offers were posted at lower and lower prices, eventually stepping in front of the prices where he had found the big seller's dark resting sell orders. Eventually, the large market participant attempting to sell on the dark pool would be forced to lower its hidden sell limit price, which was often pegged to the visible national best bid and offer on many of these systems at the time.

In victory, Jack was able to unload the entire position (by buying to cover the massive short position he had built up) using buy orders on the dark pool route, which were immediately filled by the liquidity provided by the large market participant he had located with his initial test orders.

A generation of Canada's proprietary traders had built a major part of their careers on variations of this strategy. Most of the first generation of dark pool Swifties, of course, would later find trading much more difficult when new dark pool algorithms became increasingly sophisticated as the designers of the systems soon learned of the tricks used to game their technology. Some of these traders actually accepted offers from the firms that owned and operated dark pool systems to help them develop anti-gaming techniques in exchange for significant compensation packages. Others quickly latched onto new strategies, which is an ongoing process for many career prop traders. The rest, who failed to adapt, were often labeled one trick ponies and disappeared from the business as soon as the temporary edge had disappeared from the markets.

While the game was still good, of course, it was easy pickings for the so-called dark pool traders at Swift Trade who made a big part of their living out of picking the pockets of large institutions that attempted to use the controversially opaque market systems to hide their tracks. And the pockets of these market participants were deep enough to support entire trading floors full of recent college grads whose first foray into the trading world was the exploitation of this edge. It was an edge in a category that I call a “hard edge” in the markets—the type of edge that is typically transient but allows traders who are equipped to exploit it to do so with minimal skill, discipline, or experience.

That's not to say that a certain level of skill and talent was completely absent in this group of traders. Only that it was much easier for a trader to profit using such an edge than with “soft edges,” in which case skill, discipline, and experience is a much more significant deciding factor in one's net profitability.

That Friday afternoon, Gordon had loaded up a massive position in a dark pool play on a small illiquid NASDAQ stock. It was 3:55 P.M. in the New York (also Toronto) time zone. Every trader on the floor watched his or her own on-screen clock, well aware of the rule that Liz had driven into our brains since day one: Never hold a position past the market closing time at 4:00.

I was up on the day and felt relaxed, knowing that I closed out a week in the green. It wasn't much money at the time, but it felt like an accomplishment to me, given the constantly repeated “90 percent of all traders lose” mantra. That, and the hope that I would later be able to swing the kind of sizes that Jack and Gordon were doing by then.

Gordon, on the other hand, shifted forward in his seat and assumed a panicked position like an animal sensing an approaching predator. Liz walked at a more urgent pace than usual across the trading floor and stopped behind Gordon as she looked over his shoulder.

“It's 3:55, Gordon. Flat yourself, now.” She whisper-shouted the command as though the rest of the floor wouldn't notice as long as the actual volume was kept low. Of course, everyone on the floor, including most who pretended not to, had turned our attention to Gordon's predicament—especially as most of us were already flat by that time of day.

On the more liquid and heavily traded stocks, the rule against holding overnight positions was twofold: risk reduction and the company's own margining issues. For illiquid stocks like the names used by the dark pool traders, it was more than that: Illiquid stocks tend to have relatively wide spreads (large differences between the best bid and best offer, typically more than the one penny difference seen on more active stocks) during the day. After hours, it only gets worse. It's not uncommon to see these illiquid NASDAQ stocks show after-hour spreads of more than 50 pennies, maybe even a dollar or two.

Basically, a large position on one of these names, in either direction, would be at a massive loss if it were forced to be exited at market after hours.

And Gordon had piled into a massive position with less than five minutes remaining in the last trading day of the week. (There was no way Liz could convince Swift's head office to hold on to a floating loss of this size over the weekend.)

Panicked, Gordon pulled out all the stops and used every route he knew of to cover his position. He locked in small losses by hitting dark liquidity players who put out hidden orders between the spread, but it was better than closing out the entire position by aggressively hitting the opposite side of the national best bid and offer—that would be the last resort.

In those last five minutes of that Friday trading session, Gordon's exit trades were responsible for almost all the prints (completed transactions reported on the consolidated tape) on that illiquid little NASDAQ stock. In the process, he colored the trading floor air with a barrage of profanity that was as much a work of art as his manual destination-selection skills as he closed out lot after lot like a human smart router.

When the bell rang at 4:00 Eastern time, he was still caught in a fraction of his initial position. Liz placed a call to the head office and explained the situation, along with a quick summary of Gordon's track record over the past year. As soon as she hung up, he gave Gordon the okay to close out the rest of the position at market (to hit the opposite side of the after-hours NBBO at that point in time with the entire remainder of his position).

At 4:01, he got himself flat on the market and was down only $30,000, give or take a few hundreds and change. And I emphasize only. Had he stayed in his entire position from 3:55, he would have put a much bigger dent in that branch of Swift Trade Securities.

In an industry filled with stories of underhanded dealings, schemes, and betrayals, what I witnessed in the after-hours session that Friday afternoon was a whole other side to the business.

Liz picked up the phone again after Gordon had fallen silent with his head resting in his hands—most likely under the assumption that he had set himself up to be a sacrificial lamb for risk management. I recall my utter surprise as Liz hung up after a conversation with the branch owner, and told Gordon, “Don't worry about it. We'll cover the loss. You're back to zero for the month on Monday.” Without quite putting it into words, she gave him a look that made it clear that he'd better not do anything like that again, of course.

It turned out Gordon had made a lot of progress over the months preceding that incident, and the $30,000 was only a small fraction of his earnings to date for the year. The reset to zero was also fortunate for him as it was still early in the month.

It was a small fee to pay for a big lapse in risk management. Most traders who learn comparable lessons lose much more, and often their own funds.

Of course, as a beginner trader, the concept of a $30,000 loss in such a small period of time had a very different image for me. It wasn't so much the incident itself as it was the scale of it that was daunting.

In the years that followed, I learned through experience and observation that this is an industry where the sheer scale of income traders make during times of real hard edges, like dark pool gaming, may often be necessary for traders to cover the living expenses of the lean years when soft edges are all that's available to be traded and income levels are relatively lower or less consistent for some traders.