Streetwise Professor

April 30, 2020

WTI-WTF? Part 3: Did CLK20 Get TAS-ed?

Matt Levine wrote a typically amusing piece highlighting the role of Trade at Settle (TAS) contracts in the 4/20/20 oil futures debacle:

But actually a lot of oil changed hands at those negative prices. Not because a bunch of investors came to the market all at once looking to sell, and no one would buy from them at negative prices, but for a more technical reason. Some oil traders use “trade-at-settlement” contracts: Instead of buying (or selling) oil futures at the market price at the time of your trade, you agree in advance to buy (or sell) them at whatever the official 2:30 p.m. settlement price is that day. 1  This is a good trade, for you, if your job is to obtain the day’s settlement price: For instance, if you run an index fund or exchange-traded fund that is benchmarked to that price, using TAS futures guarantees you the benchmark price. If you invest in oil futures and your boss fires you if you miss the benchmark, you might use TAS futures, that sort of thing. If you are a savvy oil trader attuned to minute-by-minute changes in supply and demand and trying to capture as much value as possible from your skills, you’ll probably just trade the futures at their current prices, selling if the price is too high and buying if it’s too low. But a lot of oil traders are doing something else, something a bit more passive, and for them the ability to guarantee the settlement price is useful.

He goes on to ponder whether the TAS mechanism could be manipulated, and whether manipulation could have contributed to the settlement fire that Red Adair couldn’t have put out:

The basic pattern—agree in advance to buy (sell) stuff at the official settlement price at some fixed future time, and then sell (buy) a bunch of that stuff in the minutes leading up to the official settlement time with the effect of pushing down (up) the price at which you are buying (selling)—is incredibly common, and the gradation from “sensibly pre-hedging the exposure you will get at settlement” to “sloppily pre-hedging the exposure you will get at settlement” to “manipulating the market to push down the price you will get at settlement” is blurry. If you type in a chat room “lol I’m gonna pound out 500 contracts to push down the settlement price and make fortune on my TAS trades, I am really ripping those muppets’ faces off, hope I don’t go to prison bro, hashtag fraud hashtag crime hashtag manipulation,” you will get in trouble. But if you don’t type that, and you quietly sell the 500 contracts and the price goes down, then as far as anyone knows that was just pre-hedging.

So could somebody have popped CLK20 with a TASer last Monday?

Funny you should ask. I wrote a paper on TAS manipulation a while ago. My interest was sparked by the CFTC’s action against Dutch trading firm Optiver, which the agency accused of doing exactly the kind of thing Levine writes about in crude, gasoline, and heating oil futures back in March, 2008. You can read the complaint–and listen to some actual “ripping those muppets’ faces off” trader braggadocio.

How does manipulation work here? First, to make manipulation profitable, there has to be an asymmetric price response to purchases and sales. If the manipulator’s purchases impact prices the same as sales, just buying and selling a lot can’t move prices in a profitable direction. Indeed, the manipulator would have to pay transactions costs (crossing the spread, brokerage, etc.) and this would cause the trading to be unprofitable.

The model in the paper derives conditions under which purchases and sales of TAS have a smaller impact on prices than do trades in the underlying futures. The basic idea is that if information is short-lived, or if there is intense competition among informed traders, new information will be incorporated into prices very quickly. Under those circumstances, informed traders will not want to trade TAS: their information will already be incorporated into the price by the time settlement occurs. Thus, TAS is a mechanism that allows traders to signal that they are uninformed: many “muppets” choose to trade TAS, and the informed don’t. Thus, the price impact of TAS trades is smaller than the price impact of regular outright trades: trades move prices because of the possibility that they are motivated by private information, so trades that are unlikely to be privately informed move prices less than trades that are more likely to be so.

This creates the asymmetry that makes manipulation possible: the manipulator buys, say, the TAS and then sells in quantity immediately before and during the settlement period and profits as Levine describes.

This type of manipulation is particularly pernicious because manipulative trades have persistent price impacts because they cannot be distinguished from informed trades (or liquidity trades, for that matter). Note that in Optiver, prices did not reverse after the firm’s trades,

This strategy is likely to be particularly profitable when markets are relatively illiquid, as in an illiquid market outright trades have bigger price impacts. Liquidity (measured by the bid-ask spread, quantity at the top of the book, price impact coefficients, etc.) has plummeted for everything since the CovidCrisis began, and the decline in CL liquidity has been particularly pronounced. Moreover, contracts close to expiration are less liquid anyways. Add to this the extreme physical constraints (which mean that small shocks to fundamentals have big price impacts) and the raging uncertainty about the logistical situation at Cushing, and it is likely that small volumes at the settle could have big impacts on prices.

To this I would add that an unexpected shortfall in buy orders (due to shorts exercising market power) at the settle could have price impacts, and exacerbate the price impacts of sell orders by exacerbating order imbalances.

Thus, the potential for a big asymmetry in price impact was pronounced on that fatal Monday.

In sum, it is not implausible that the market did indeed get TASed. Or at the very least, a TASer jolt contributed to the collapse. (Sort of like in this video!)

A final remark on the economic benefits and costs of TAS trading. TAS is a form of “cream skimming”–i.e., the skimming off of uninformed order flow. This tends to make the order flow in the regular continuous market more toxic, which reduces liquidity in that market. For this reason, other cream skimming mechanisms used primarily in equity markets (payment for order flow, dark pools, block trades) are frequently criticized. (This is why some regulators, particular in Europe, have attempted to curb such activity and force more trading into “lit” venues.)

I showed in my Market Macrostructure paper that things aren’t so simple. If the regular market isn’t perfectly competitive, the increased competition from a cream skimming mechanism can improve welfare. Moreover, there are distributive effects here: the uninformed traders who can utilize the TAS mechanism (e.g., those who are hedging exposures tied to the settlement price) benefit, while uninformed traders who can’t lose. Informed traders can lose too. Moreover–and this is a point that is almost always overlooked–some informed trading is essentially rent seeking (e.g., trading on information that will be released shortly anyways, and accelerating its incorporation into prices has little effects on resource allocation decisions). Reducing rent seeking informed trading is a good thing.

All in all, the role of the TASer is yet another piece of the 4/20/20 WTI WTF puzzle. The forensic analysis of this entire episode will be fascinating.

Print Friendly, PDF & Email

Powered by WordPress