Testimony Of A High Frequency Trader

The so-called “flash crash” at the New York Stock Exchange on May 6th was naturally one of the the main issues at Golden Networking.net’s High Frequency Trading Leaders Forum 2010 last week. Among the key participants was Mr. Manoj Narang, founder and CEO of Tradeworx, one of the HFT firms that stopped trading on May 6th. Below is a transcript of the conversation between Mr. Narang and The Wall Street Journal‘s Scott Patterson that took place after the formal speeches.

“The market was ripe for a catastrophic event, because it was so saturated with stop orders, all it needed was a catalyst.”

Manjo Narang

“I want to call it Seis De Mayo because I believe it’s a vindication day for high frequency traders. I think that fingers were very quick to be pointed in the direction of High-Frequency Trading for this meltdown; I think that the post-mortem has not been written yet on this episode, but what’s abundantly clear is that like has been the case in every other crisis in recent memory, humans were intimately involved and fully responsible for this particular meltdown as well,” Mr. Narang said in his speech at the HFT Leaders conference.

After the speech, the following conversation took place between Manjo Narang and WSJ-reporter Scott Patterson who wrote a groundbreaking article on high frequency trading last year:

Scott: I remember, before May 6th, one of the predominant themes that I heard, in defense of high-frequency trading, was that, in 2008, when everything is falling apart, that one part of the entire market that stood up was the equity market, and that was a defense made by high-frequency trading, and high-frequency trading took the credit for that. On May 6th, obviously, you saw a completely different market, that revealed structural problems. I dont think there’s really much argument that high-frequency trading traders pulling out added to…

Manoj: You wanna put the statement of high-frequency trading traders about the services they provide against them?

Scott: I want to see how you react. The issue is when you have chaotic markets…

Manoj: You know you are not going to get me on this one?

Scott: I know, you are ready for this one. You have a chaotic market that messes up the models, the data slows downs, a lot of high-frequency traders pull out, I know they did, I talked to a lot of them, you look at the order books, and they just completely evaporate. It wasn’t just high-frequency traders, everybody got out, you see this as a problem, when you have a group of traders who are 60% of the market, who are highly sensitive to chaotic markets, and when you have that, they pull out, and that makes the decision even more chaotic. Is there a solution to that?

Manoj: I don’t think that’s a correct characterization. First of all, you are inferring that if humans would be running the show, they wouldn’t have pulled out. And that’s exactly what happened in 1987.

Scott: On the Big Board, they did slow down trades on May 6th, yet they had no cancelled trades. So you had a part of the market, which had human intervention, actually staying.

Manoj: And that was a critical part of what exacerbated the plunge, and let me explain why.

Scott: That’s not what the NYSE will say.

Manoj: Of course, they won’t say that, but I will; human intervention exacerbated the crisis. First of all, I will just reiterate that this crisis was precipitated by panic selling by humans. The reason why this happened was because was the market was in particular vulnerable state on May 6th. We just had had a huge run-up in the equities markets, we were in the midst of a 10% correction, even before the mayhem unfolded, and on top of that you had very vexing news coming out of Europe, lots of images on CNBC, people rioting in the streets in Greece, the prospects of the second phase of the crisis unfolding. In that sort of an environment, people tend to seed the markets with stop orders. That’s what happened in 1987, that’s what happened this time around. So, the market was ripe for a catastrophic event, because it was so saturated with stop orders, all it needed was a catalyst. And the catalyst occurred, according to a couple of Reuters reporters, because a large mutual fund complex, decided to do a $4.5B hedging transaction in e-mini futures contracts, which track the S&P 500 index, when the market was already in that vulnerable state.

Scott: Most of those were actually transacted on the way up.

Manoj: Right, but the orders were entered prior to the huge collapse, and that’s the important part. On an ordinary day, an order of that size, and that’s a pretty substantially sized order, would definitely have had a ½% or 1% price impact. But on this day, when volatility was already elevated, and the market was just looking for a reason to take profits, it had an outsize effect, and very likely triggered the first wave of stops, which then turned into market orders, which then led to a gigantic spike in volume. What happened then, in relatively short order, was that the Arca exchange (which is owned by NYSE), fell behind, in terms of its ability to process quotes. That’s not that unusual, that happens all the time, particularly with Arca. When that happens, faster exchanges, namely BATS and Nasdaq, are entitled to take a regulatory remedy, known as declaring called “regulatory self-help”. It means that the exchanges absolve themselves of their responsibility to route orders to Arca, at seemingly better prices, because they are declaring that those prices are steal and not real. After this, the Arca exchange ceased to be a functioning part of the market, effectively going offline. Shortly after that, NYSE made the fateful decision to take its exchange offline. Two of the biggest exchanges in the market were effectively closed for business, for a short period of time. In that sort of an environment, combined with the fact that all of a sudden, lots of stops were turning into market orders, what happens is that, because of the reduced liquidity, these market orders start slicing through the available liquidity, like a hard knife through butter, and there’s very little resistance in the way. Price formation in NYSE stocks is severely impacted, because, although NYSE itself has only 25% market share, it is still the primary price discovery market mechanism for stocks on the Big Board; you saw an impact that was far more pronounced in NYSE stocks. In fact, every stock that heavily declined, with one exception, was a NYSE stock. So that tells you of the wisdom of NYSE’s actions. So with half of the market offline, with clearly erroneous prices being transacted in the market place, certain high-frequency traders, myself included, made the decision that now it is not a good time to trade, because we are just going to exacerbate the problem.

So let me answer your question, Scott, by sending a question in your direction. When technology is not working, when system are screwed up, do you want the airline pilot to decide to take off, even though he knows his instruments are not working properly? I think the proper thing to do is to shut off. Really, the proper thing to do would have been for the exchanges to shut off, to prevent those erroneous prices to ever happen. NYSE’s decision would have been well-advised if there had been market-wide protocols in place to take a time out. But given that they weren’t, NYSE was extremely ill-advised to do that. Other things I would say:

First, you can ask why high-frequency traders were not there providing liquidity, when that’s their stated mandate, you can ask why high-frequency trading traders turned off when volatility was sky-high, given the contention that volatility benefits them. These are all valid questions, but you have to understand that strategies are not monolithic. People run different sorts of strategies.

A very common sort of strategy, Statistical Arbitrage, is a strategy that propagates price information from stocks that are moving to stocks that haven’t moved yet. It’s one of the primary mechanisms for liquidity transfer in the markets. However, what this involves is propagating price information. In an environment where the inputs are wrong, propagating this information is just going to exacerbate the problem. What I am trying to say is, if P&G is down 40%, it is not appropriate to propagate these values to other stocks, even though they are correlated to P&G. For those types of systems, the only responsible thing to do is to turn off because, even though they serve a very valuable function in an ordinary market, which is to propagate price information, on a cross-sectional basis, they are actually counterproductive, when the prices they are trying to propagate are clearly erroneous. So that’s the second point.

The third point I would make, and this is perhaps the most important point, is that liquidity really should be thought of as a commodity. It is a commodity, and in like any other market, it has supply and demand. Volatility is the symptom of the absence of liquidity. We saw that in May 6th, there was no liquidity, so markets were free to move in huge increments in short periods of time. What’s the relevance of what I am saying? In a supply and demand market, when there s a shortfall of supply or, equivalently an excess of demand, it becomes very profitable to be a supplier in the market. When oil prices are sky-high, producers of oil make a fortune. However, that does not imply that their activity exacerbates the mismatch between supply and demand, in fact, it does the opposite. So while oil producers are making money, oil prices being sky-high, by distributing oil into the market place, that tempers the price of oil and drives it back down by bringing the imbalance between supply and demand closer to equilibrium.

The same thing is true with liquidity. Liquidity is the opposite of volatility. When there is volatility in the market, it is because there is a shortfall of liquidity, there is a shortfall of high-frequency trading, because high-frequency trading is the backbone of liquidity in today’s markets. For that reason, when there is volatility, it becomes extremely profitable to be a high-frequency trader. Many of the folks that I know continued trading on that day had very profitable trades. Our own simulations in my firm indicate that we would have had a record day of profits if we kept trading. The reason we stopped, apart from the reasons we already mentioned, which were that two of the exchanges were down, and that price information shouldn’t be propagated if it is erroneous, we were also concerned about the risk of trade breaks. So if exchanges allow for these trades to happen (by the way, many thousands of trades were broken), so if you happen to buy at some of these prices, and then sell it at a higher price, and think that you made a tidy profit as a result of that, you can be sad to learn that your buy is broken but not your sell, and you actually finished the day short; if the problem is bad enough, you could actually violate your margin limits, and wind up with a huge overnight position that can’t be supported with the base of capital you are trading of.

So in those sorts of circumstances, even though you think you are going to earn a huge profit if you continue trading, it’s not responsible to do so, unless you are extraordinarily capitalized and can take that risk. So that’s the reason why my firm turned off, not because we didn’t want to be there providing liquidity, but the bigger point is that imbalances between supply and demand of liquidity, are driven by the demand side of the equation. What I mean by that is that, on May 6th, we saw a record number of shares changing hands. So clearly there was a lot of high-frequency trading going on. But also clearly, it was not nearly enough to sate the demand for liquidity. That because liquidity shortfalls are driven by the demand side, not by the supply side. Like any other supply and demand structure, the market cannot be assumed to be always in equilibrium. And there is no way regulators can force any market to be in equilibrium, that’s just purely a function of supply and demand. No matter what regulators do as part of liquidity obligations, no matter how much they force people to stay in the market, there will be times when herd-like behavior among long-term investors will all be stampeding for the exits at the same time, and simply there won’t be enough high-frequency trading to cover the demand for liquidity. Like I said, liquidity crises are not driven by the lack of liquidity, but by the demand of liquidity. So, I hope that comprehensively answers, from a number of different perspectives, the question you asked, but also, why it is a bit misguided on the part of regulators to try to prevent liquidity crisis from occurring. In order to prevent liquidity crises from occurring, you would need to prevent herd-like behavior among long-term investors, because that’s what causes bubbles, and that’s what causes liquidity crises.

Related by The Swapper:

The Rise Of The New Market Makers

May 6. 2010: “The Black Thursday”

Wall Street Collapse: Did Somebody See It Coming?

U.S. Stock Crash Compels Further Investigation of Wall Street Scam

Welcome Back to Earth, Mr. Market

Living In A Derivative World