Man Vs. Machine: What's Right and Wrong with Stock Trading System
Smart Order Routing Systems are supposed to find the bets prices, but don't always do so these "top of the book" rules were created to make sure investors get the best prices, but that isn't always happening.
It's a brave new world of securities trading: new customers, new exchanges, new ways of buying and selling stocks, and new worries for the public.
And you can thank the Securities and Exchange Commission. Several years ago, the SEC, acting on directives from Congress, decided to foster competition and embrace electronic trading.
It certainly succeeded. As a result of changes in the regulatory structure, there is a whole new class of customers (high-frequency traders being the most prominent), new exchanges—each with slightly different rules that are designed to keep people trading in their space—and a whole new way of buying and selling stocks outside of exchanges (dark pools, crossing systems).
Some believe these changes were necessary, others think that they have opened up a whole new can of worms for the investing public.
"We have a lot of choices when we decide to trade a stock," says Gus Sauter, Chief Investment Officer at Vanguard, one of the country's largest mutual funds. "The decision we make will depend on the nature of the stock itself. Is it a large order or is it a small order—and across perhaps many stocks?"
One thing is for sure: There was no golden era of stock trading. The old system was expensive, slow, leaky (everyone knew everyone else's trading business), and had plenty of abuses.
Pros And Cons
In that context, then, there is much to like about the way trading is done today:
- The NYSE and Nasdaq no longer have a monopoly. There is plenty of competition for listings and trading.
- Trading volume has exploded. Consolidated, average-day volume (total trading in all NYSE-listed stocks) has almost tripled from from 2.1 billion shares in 200 to 5.9 billion shares in 2009.
- Trading costs have dropped dramatically. Before fixed commissions were abolished in 1975, it was typical to charge as much as a 1-percent commission. To trade 100 shares of IBM at $128 would have cost $128. Some even had minimum commissions in excess of that. The commission today varies, but $7.00 for a discount broker is common.
- Trading is faster and more convenient. In the old days, you telephoned your broker, and it could take hours—even days—before you got a confirmation on your trade and learned the transaction price. Today's online brokers measure time-to-fill orders in milliseconds, and you do it on a computer.
Progress, however, has come with a price. The new marketplace has a patchwork regulatory structure, insufficient transparency and—like the old trading system—abuses. Among the problems:
- Allegations that certain groups of players, particularly high-frequency traders, have unfair advantages over others.
- More fragmentation means markets do not always connect to each other very well, meaning the pool of liquidity (offers to buy and sell stock) is sometimes very shallow, which can create pockets of volatility and—on days like the Flash Crash of May 6—major price movements.
- Arcane rules that do not pass simple "smell tests" for fairness. For example, the markets are supposed to be populated by "market makers" who have an affirmative obligation to continually provide bids and offers for stocks, but their obligations are minimal. How minimal? In a stock trading $60, a market maker might only be required to post a bid—an offer to buy—of 100 shares for one cent a share. Such "stub quotes" are common today and happened during the Flash Crash.
For more insight into how the rules dont pass the proverbial "smell test," look at what happened in trading in Ciscoa few weeks ago and how the stock moved 10 percent in a millisecond.
Anatomy of a Trade: What Really Happened to Cisco?
At 10:41am ET on July 29, Cisco was trading at roughly $23.37. At that moment, there was a series of eight 100-share trades on the American Stock Exchange. Each of the trades was notably higher than the prior one: $23.67, $24.13, $24.16, $24.51, $24.59, $25.37, $25.56, and $26.00.
Once the stock printed at $26, it triggered a recently-enacted SEC rule requiring a five-minute halt in trading whenever a stock moves 10 percent in a five-minute period.
Remember, these trades happened essentially at the same time—literally, the same second.
How could this happen? How can a liquid stock like Cisco go from $23.37 to $26.00 so quickly?
Well, it is not because there is no one willing to sell Cisco at a lower price via another trading venue. And don't go blaming the AmEx, the NYSE, or the Nasdaq.
The culprit is very likely the structure of the trading system rules. Simply put: the market has fragmented into multiple pools of liquidity, which in reality are sometimes neither deep nor liquid. "You can trade at 10 different exchanges and 30+ alternative trading systems," says Vanguard's Sauter.
Tracing the Trading
Here is what probably happened. A market order (an order to buy at whatever the prevailing price) of uncertain size was placed to buy Cisco on the AmEx. There was obviously very little liquidity (offers to sell) on that venue, because the print on the AmEx was $23.67, well above the then-prevailing price of $23.37.
But wait: shouldn't the computers at the AmEx then have routed the buy order to the another exchange—Arca, NASDAQ, BATS, Direct Edge—that had a better price?
Yes, but under prevailing rules, trading venues (like the AmEx, which is owned by the NYSE) are required to go to other venues but only sweep the "top of book." In other words, the computer's only requirement was to take the one best offer—say $24.17 a share. After that it can move on to another venue, again taking the best, offer—say $24.51 a share—and then move on again.
If it ends up back at the AmEx and still hasn't filled the order, it keeps hunting there until it finds an acceptable offer, in this case, $26 a share.
Wait, you say. Why can't the system hunt for additional prices in any venue? Surely there were offers at the Nasdaq, for instance, to sell at $23.38 a share? There almost certainly were. The reason this happens is that exchanges—all of them—want customers to stay in their system. They offer money—rebates—for them to add liquidity.
On top of that, sometimes so-called Smart Order Routing Systems that are supposed to find the best prices don't always do so.
How could this situation been prevented? Obviously, had the computer that executed these orders looked a little bit deeper it seems likely could have found ample offers to sell shares at the prevailing price of $23.37.
But wait, what about the investor who bought Cisco at $26, when it was trading at $23.37 away from the AmEx? The NYSE ruled that the trades in Cisco will stand. Is that person out of luck? No.
If I was that person, here is what I would do:
- Call my broker and claim that they have violated their best execution obligation (this is a legal term, a bit fuzzy but not without some power) and demand an explanation.
- If they say their "Smart Order Routing System" sent the order to the best venue for execution and they did indeed fulfill their obligations, I would immediately declare it to be the "Stupid Order Routing System" and demand they pay the difference between what I paid and what Cisco was trading for on the other venues.
- If the order was done manually by a trader, who simply pushed the wrong button and sent the order to the wrong venue, or that he had a fat finger, or a thin brain, or a wide latitude, or anything else, I would still demand they pay the difference.
Still, the damage has been done. The NYSE has said the trades in Cisco will stand. Let's hope the SEC order-handling rules do not.
This is just one example where a regulation—in this case, "top of the book" rules—appears to be creating a problem that should not be a problem.
These "top of the book" rules were created to deal with the many new exchanges and to make sure that the investing public got the best prices, but as you see, that isn't always happening.