A Bad ‘Stop-Loss’ Trade Worth Fighting About

Here’s an example of one investor who should fight to have his trade unwound.

We get a lot of emails. We do in fact read them all. Here’s a snippet from one we received this morning that we would like to share:

“I am an active user of your website and an active retired trader of ETFs and stocks.

A trade occurred through Fidelity yesterday that you might find interesting. I had a position in XLV [the Health Care Select SPDR (XLV | A-92)] with a stop at 5 percent below the closing price yesterday.

At 4:00 p.m. EST, over 400K shares traded with approximately 100 shares trading 5 percent below market price. The price then went back to the closing price. This 100 share trade triggered my stop, but I sold at the closing price (for a small gain). Fidelity says it is a valid trade and the stop was triggered. I have not seen this occur before and was interested in your view.”

So, there are a few things going on here that should be an object lesson for traders of ETFs.

First, here’s the intraday chart of XLV going into the close so we can see what he’s talking about:

Source: Bloomberg

The official marked close for XLV was $59.68. He’s saying he had a stop 5 percent below that, which would be roughly $56.75.

The first important point to make here is why he had a stop. A stop in this case is a “stop-loss.” It’s an order you leave sitting in your brokerage account to sell your position (as a market order) should the ETF trade through a given price. That sounds like a nice piece of insurance—if the market goes in free fall, at least you get out, right?

Well, sometimes. In this case, our emailer had a best-case-scenario trade. The stock traded through the stop, but got executed far above where the trigger was. In most scenarios, that stop order turns into a market order, and in a declining market, once that happens, you’re entirely at the mercy of other traders. If XLV had been in a true free fall, he could have triggered his stop and then it would’ve been executed anywhere.

This is precisely how people got burned during the so-called flash crash in May 2009. People had stop-losses sitting on their accounts—sometimes forgotten stop-losses—far, far below current prices. When the hiccup came, those all turned into market orders, and they got executions vastly below rational fair prices. In this case, because the hiccup seemed to hit just one trade, the market recovered before his trade got executed.

There’s another kind of order he might have had on the books, which is a stop-limit order. Normally, I’m all about limit orders. They keep you from having unexpected surprises. But in a stop-loss, if you set your stop at say, 5 percent down, you then need to decide how bad a price you’ll take for your limit. In the case of XLV, his stop of $56.75 would then need a limit of maybe $56.60. The problem there is, if the market’s collapsing fast enough, you might never get executed, which defeats the entire purpose of the stop.

But the real mystery here is why his stop got triggered at all. After all, the scale on that end of day chart doesn’t go below $59, much less $56. With an ETF like XLV, you have to wade through thousands of trades at the close of the market, but on the chart above, you can’t see a single trade anywhere near $56.75 all day.

To figure out what’s going on, you have to look at the actual corrected tape from last night:

Source: Bloomberg

It’s not the 100 shares he suggests in the email, it was instead a reported trade of some 185,031 shares at $56.71, or roughly a $10 million trade.

The most important things in this display, however, are the arcane ones. The “*X” next to the trade means it was canceled. The “D” means this was a reported trade through the FINRA Alternate Display Facility or “ADF.”

And this is where things get confused. The FINRA ADF is something that all FINRA members have access to—that is, every broker/dealer in the country, no matter how small. It’s used to report trades that happen off the major exchanges. When you hear about “dark pools,” the volume from them comes in as individual ADF reported trades. Those trades are supposed to come in nonaggregated (one entry per trade) and time stamped based on when they occurred. If the trade is an error or is canceled, FINRA rule 6280 lays out a self-correction mechanism. If you cancel the trade within 10 seconds, you’re not a bad actor.

See how this could be problematic? It’s a system ripe for abuse and simple sloppiness. In this case, if we look for all trades of that size during the day, lo and behold:

Source: Bloomberg

That same amount was reported through the same facility 45 minutes later, and a price over the final close of $59.68. The “T” means the trade was reported by Nasdaq through the ADF, and the “FT” means it happened after hours; in other words, the actual trade did not occur until after the close.

We could speculate what’s happening here, but it would be just that, speculation. What’s clear is that the original canceled trade did not simply get punched in with the wrong price, because it would then have been resubmitted, and wouldn’t be marked as a Form T, after-hours trade.

Instead, it’s likely that (since the final trade was above the close) this was someone looking to buy their $10 million of XLV, who hit someone’s ask in a dark pool. That seller then said “whoops,” and the trade was unwound, leaving the buyer to scramble to get their position on after the close.

In other words, our emailer got hosed here. Luckily, he didn’t get hosed badly, but I don’t see this phantom 100 share “valid trade” that he’s talking about. The system did work—his stop was triggered by the tape and then executed. But the system should also include him running the tape back to Fidelity and getting his trade unwound as well.

And the moral of the story? If you’re going to get fancy with your ETF trading (and stops are fancy) then you better be willing to wade into the post-trade fight. Nothing like that is ever automatic. But nobody’s going to fight for your trades but you.

At the time this article was published, the author held no position in the security mentioned. Contact Dave Nadig at [email protected].