Searching over 5,500,000 cases.

Buy This Entire Record For $7.95

Download the entire decision to receive the complete text, official citation,
docket number, dissents and concurrences, and footnotes for this case.

Learn more about what you receive with purchase of this case.

Federal Energy Regulatory Commission v. City Power Marketing, LLC

United States District Court, District of Columbia

August 10, 2016



          JOHN D. BATES United States District Judge

         City Power Marketing, LLC, an energy-trading firm founded by K. Stephen Tsingas, engaged in “virtual trading” in wholesale electricity markets. Virtual traders do not actually supply or receive electricity but instead stake out market positions that are effectively bets on how electricity prices will change over time. In other words, they engage in a kind of arbitrage. And their efforts are generally thought to improve the overall efficiency of energy markets.

         According to the Federal Energy Regulatory Commission (FERC or Commission), however, in July 2010 City Power engaged in a series of manipulative virtual trades that hurt the market while generating more than $1 million of profit for City Power. In essence, City Power found a way to place trades that had no risk of earning or losing money on the basis of price changes but that nonetheless triggered a financial credit for City Power from the market operator. After a lengthy investigation, FERC concluded that these trades constituted a fraudulent scheme that violated the Commission’s Anti-Manipulation Rule, 18 C.F.R. § 1c.2. FERC also concluded that by failing to reveal the existence of certain archived instant messages during the investigation, City Power had violated the Commission’s Market Behavior Rule 3, 18 C.F.R. § 35.41(b), which requires truthful communications by parties subject to FERC’s authority. FERC ordered City Power to disgorge its profits and assessed penalties of $15 million on City Power and Tsingas. Disputing their liability, City Power and Tsingas did not pay. As provided for in the Federal Power Act, FERC therefore filed this action seeking an order affirming its penalty assessment.

         City Power and Tsingas-whom the Court will collectively call “City Power, ” unless context indicates otherwise-have moved to dismiss, arguing that FERC’s claims under both the Anti-Manipulation Rule and Market Behavior Rule 3 fail as a matter of law. City Power also argues that this case should be treated like a normal civil action subject to the Federal Rules of Civil Procedure, not a summary review of agency action. The Court largely agrees with City Power on this latter point, and will follow the normal course of district-court adjudication. But the Court disagrees that FERC’s claims are unsound. Assuming the truth of FERC’s allegations, as the Court must at this stage, FERC has stated plausible claims under both the Anti-Manipulation Rule and Market Behavior Rule 3. City Power’s motion to dismiss will therefore be denied.


         I. PJM’s Wholesale Electricity Market

         This case concerns allegedly illegal trading in the wholesale electricity market run by PJM Interconnection, LLC (PJM). PJM is the independent, nonprofit Regional Transmission Organization (RTO) that administers the electric grid in a 13-state region that extends from North Carolina to New Jersey to Illinois and includes the District of Columbia. As part of administering the grid, PJM operates a wholesale electricity auction of the sort recently described by the Supreme Court:

These wholesale auctions serve to balance supply and demand on a continuous basis, producing prices for electricity that reflect its value at given locations and times throughout each day. Such a real-time mechanism is needed because, unlike most products, electricity cannot be stored effectively. Suppliers must generate-every day, hour, and minute- the exact amount of power necessary to meet demand from the utilities and other “load-serving entities” (LSEs) that buy power at wholesale for resale to users. To ensure that happens, wholesale market operators [such as PJM] obtain (1) orders from LSEs indicating how much electricity they need at various times and (2) bids from generators specifying how much electricity they can produce at those times and how much they will charge for it. Operators accept the generators’ bids in order of cost (least expensive first) until they satisfy the LSEs’ total demand. The price of the last unit of electricity purchased is then paid to every supplier whose bid was accepted, regardless of its actual offer; and the total cost is split among the LSEs in proportion to how much energy they have ordered. So, for example, suppose that at 9 a.m. on August 15 four plants serving Washington, D.C. can each produce some amount of electricity for, respectively, $10/unit, $20/unit, $30/unit, and $40/unit. And suppose that LSEs’ demand at that time and place is met after the operator accepts the three cheapest bids. The first three generators would then all receive $30/unit.

FERC v. Elec. Power Supply Ass’n, 136 S.Ct. 760, 768-69 (2016). The clearing price at a particular location, or “node, ” on the PJM grid is called the “locational marginal price” (LMP).

         PJM operates a so-called “dual settlement market, ” meaning that it runs two rounds of bidding for each operating day. It first runs a “day-ahead market, ” which “allows market participants to secure prices for electric energy the day before the operating day and hedge against price fluctuations that can occur in real time. One day ahead of actual dispatch, participants submit supply offers and demand bids for energy. These bids are applied to each hour of the day and for each pricing location [i.e., node] on the system.” FERC, Energy Primer: A Handbook of Energy Market Basics 95 (Nov. 2015). PJM takes all of the bids and offers, crunches the numbers, and determines what the clearing price-the “day-ahead LMP”-will be at each node. Generators who offered to supply energy for less than the clearing price are committed to supply that energy and are paid the day-ahead LMP; buyers who bid to purchase energy for more than the clearing price are committed to their purchases and pay the day-ahead LMP. Id.

         The next day’s actual supply of and demand for electricity, however, might be different from what the day-ahead market presumed. A generation unit might unexpectedly fail, affecting supply, or the weather might be much different than predicted, affecting demand. PJM therefore also runs a “real-time market” designed “to meet energy needs within each hour of the current day.” Id. (The real-time market is also sometimes called the “spot market” or “balancing market.”) As the name suggests, offers and bids in this market are made in real time. “Real-time LMPs are calculated at five-minute intervals based on actual grid operating conditions as calculated in PJM’s market systems.” Id.

         PJM’s wholesale market is not restricted to those engaging in “physical transactions, ” i.e., those who will actually deliver or actually receive electricity. Traders who neither have nor want actual megawatts can engage in “virtual transactions.”

A virtual transaction does not require generation to be dispatched or load to be served. Rather, it allows a market participant to arbitrage day-ahead versus real-time prices by either purchasing or selling a position in the day-ahead market, and then doing the opposite in an equal volume at the same location in the real-time market, thereby taking no physical position when the system is dispatched.

City Power Mktg., LLC, 152 FERC ¶ 61, 012 at ¶ 17 n.38 (2015) (“Penalty Assessment Order”).

         Suppose, for example, that a trader believes that the real-time LMP at a particular node will be higher than the day-ahead LMP. (Perhaps the trader has reason to believe that the temperature there will be higher than most forecasts predict, meaning increased air-conditioning use and hence increased power consumption.) The trader makes a virtual bid in the day-ahead market to buy 100MW at $25/MW. If his bid clears, then to offset his purchase, he must also sell 100MW at the same node in the real-time market. If his prediction turns out to have been correct, and the realtime LMP rises to $30/MW, he makes a profit: he “bought” 100MW for $2500 in the day-ahead market and “sold” it for $3000 in the real-time market. Likewise, a trader who correctly predicts that the real-time LMP will be lower can conduct a mirror-image virtual transaction, promising in the day-ahead market to supply 100MW in exchange for $3000, and then buying those 100MW in the real-time market for only $2500. These sorts of arbitrage transactions-the first is called a “decrement bid, ” the second an “increment offer”-are good not only for the traders but also for the market as a whole. Virtual trading encourages price convergence between the day-ahead and real-time markets, and provides price discovery, market liquidity, and increased competition. Penalty Assessment Order at ¶ 20 & n.48; see also PJM Interconnection, Virtual Transactions in the PJM Energy Markets 22-31 (Oct. 12, 2015) (explaining how virtual transactions help mitigate both buyer- and supplier-side market power).

         This case centers on another, more complex transaction called an “Up-To Congestion” transaction (UTC). Whereas the virtual transactions described in the preceding paragraph revolve around price changes at a single node, a UTC is concerned with the changing spread of prices between two nodes. It works like this: The trader selects a “source” node and a “sink” node, specifies a number of megawatts, and bids a maximum amount by which the day-ahead LMP at the sink might exceed the day-ahead LMP at the source. (The price difference correlates with the transmission congestion between the nodes, hence the name “Up-To Congestion.”) If the actual day-ahead price difference is less than the trader’s bid, the UTC transaction clears, and the trader must pay the day-ahead price difference times the number of megawatts specified. The trader will then receive the real-time price difference between the nodes (times the number of megawatts). Thus, if the price difference, or “spread, ” between the nodes is larger in real time, the UTC transaction is profitable. See Penalty Assessment Order at ¶¶ 2 n.7, 18.

         An example will illustrate. A UTC trader might pick source A and sink B, specify 100MW, and say that the day-ahead price at B will be no more than $40/MW greater than at A. After the day-ahead market clears, it turns out the price at B is $120/MW and the price at A is $90/MW. Because the difference is less than the trader specified, her transaction clears, and she must pay $3000: the actual difference ($30/MW) times the number of megawatts (100). Luckily for the trader, the price spread grows between the day-ahead and real-time markets. In the real-time market, the price at B is up to $130/MW; at A, down to $85/MW. Hence, the trader receives $4500: the real-time price difference ($45/MW) times the number of megawatts (100). Taken as a whole (and ignoring certain transaction costs for the moment), the UTC turned a profit of $1500.

         At the time of the events in this case, many traders were using UTCs as a “purely virtual product.” Penalty Assessment Order at ¶ 19. But UTCs had been “initially created as a tool to hedge congestion price risk associated with physical transactions.” Id. at ¶ 18. Apparently because of that original connection to physical transactions, PJM required UTC transactions, even virtual ones, “to be associated with transmission service reservations, which, once obtained, provided the right to flow electricity across the PJM system.” Id. at ¶ 22. There was normally a charge to obtain a transmission reservation, but there was a legitimate way for UTC traders to avoid that charge. For reasons that are not important, PJM did not require UTC traders to reserve the same transmission path (i.e., source and sink) that the UTC involved, and it did not charge a fee for transmission reservations where the sink was located in a neighboring RTO’s territory. This meant that, regardless of which nodes were involved in the UTC, a trader could always choose to obtain a free transmission reservation by specifying a transmission path with a sink in the neighboring RTO. Id.

         These free reservations, however, were not eligible to receive a “Marginal Loss Surplus Allocation” (MLSA) payment, a type of financial credit that is central to this case. When electricity is transmitted across the grid, some energy is inevitably lost in the form of heat. This is called “line loss.” To ensure that the market price at each node reflects the actual cost of providing energy at that location, the LMP that PJM calculates for each node incorporates a line loss component. In 2006 FERC instructed PJM to start using a new method for setting the price of the line loss component. Atl. City Elec. Co. v. PJM Interconnection, LLC, 115 FERC ¶ 61, 132 (2006). This new “marginal loss method” would result in more accurate price signals, and hence a more efficient allocation of electricity-generation resources, but (for reasons not worth detailing here) it would also lead PJM to “receive[] more payments than necessary to compensate for actual line losses, resulting in a surplus revenue.” Penalty Assessment Order at ¶ 24. PJM therefore had to establish a method for disbursing the “marginal loss surplus” to market participants. PJM originally proposed (and FERC agreed) that the surplus should be paid to load-serving entities (LSEs) in proportion to their share of total load, on the theory that LSEs pay for the fixed costs of the transmission grid. Atl. City Elec. Co. v. PJM Interconnection, LLC, 117 FERC ¶ 61, 169 at ¶¶ 12-13, 27-28 (2006).

         In what became known as the Black Oak proceeding, a group of virtual traders challenged both the marginal loss method of setting prices and the original MLSA payment scheme. The traders argued that because their virtual transactions did not result in actual power flows, they should not have to pay for line losses. Alternatively, they argued that if they had to pay for line losses, they should also get MLSA payments. In March 2008, FERC rejected their arguments, concluding that they should have to pay for line losses just like all other market participants and should not get MLSA payments because they did not contribute to the fixed costs of the grid. Black Oak Energy, LLC v. PJM Interconnection, LLC, 122 FERC ¶ 61, 208 (2008). In October 2008, however, FERC partially reconsidered its decision. FERC adhered to its position that, in general, virtual transactions should not entitle traders to MLSA credit. In doing so FERC repeatedly expressed concern that, otherwise, virtual traders might “conduct trades simply to receive a larger credit.” Black Oak Energy, LLC v. PJM Interconnection, LLC, 125 FERC ¶ 61, 042 at ¶ 38 n.46 (2008); see also id. at ¶ 43. But FERC concluded that because virtual traders placing UTC bids did pay transmission costs and therefore did support the fixed costs of the grid, it appeared discriminatory not to give MLSA credit for those transactions specifically. Id. at ¶¶ 48-49. PJM accordingly revised its tariff and began paying MLSA to traders for UTC transactions that cleared the market and were associated with a paid transmission reservation. Penalty Assessment Order at ¶ 25. By contrast, a UTC associated with a free transmission reservation (obtained by selecting a transmission sink in the neighboring RTO), was not eligible for MLSA. Id. at ¶ 22.

         One final and important point about MLSA: In some circumstances, the MLSA payment a trader received would exceed the cost of reserving the associated transmission (plus other transaction costs). That is, sometimes the MLSA payment could be large enough to cover the cost of placing a UTC trade, and then some. This was most likely to be true during hours of peak usage, when marginal losses, and hence MLSA payments, would be high. See id. at ¶ 52.

         II. Defendants’ Alleged Conduct

         A. Trading Conduct

         Defendant K. Stephen Tsingas is the founder and controlling owner of defendant City Power Marketing, LLC, an energy-trading firm he started in 2005. Penalty Assessment Order at ¶ 12. City Power’s trading in the PJM marketplace focused on UTCs, with City Power seeking to identify through intensive research pairs of nodes where the price spread was likely to widen substantially between the day-ahead and real-time markets. Id. at ¶ 43. But in July 2010, FERC alleges, City Power developed a new strategy-a strategy that at bottom was “a fraudulent UTC trading scheme to receive excessive amounts of MLSA payments.” Id. at ¶ 3. City Power’s trading activity itself tells part of the story, but further detail is supplied by online instant messages (IMs) sent between Tsingas (who went by the handle “traderyoda”) and a City Power partner named Tim Jurco (“jurco831”). Jurco saved these IMs and, as will be explained later, they eventually wound up in FERC’s hands.

         The story begins in late June 2010, when Tsingas observed that other virtual traders were making unusually large transmission reservations. FERC Enforcement Staff Report and Recommendation, City Power Mktg, LLC & K. Stephen Tsingas, App’x A to Pl.’s Ex. 2 [ECF No. 1-4] at 13-15 (“Staff Report”). (Traders could see each other’s transmission reservations but not their pricing nodes. Id. at 13.) Tsingas noticed that these large reservations were during peak hours and suggested to Jurco that the other traders were “doing cheap stuff to collect losses”-i.e., to get MLSA. Id. at 15. On July 3, while pondering the other traders’ strategy, Tsingas had an insight:

traderyoda: wonder what points they’re doing
traderyoda: or is it the rope-a-dope
traderyoda: that may be the trick
traderyoda: do both sides to collect losses
traderyoda: EUREKA
traderyoda: those bastards

Id. at 16-17. By “do both sides to collect losses, ” Tsingas was referring to “round-trip” trading: placing one UTC trade with source A and sink B, and a simultaneous trade with source B and sink A. This combination of trades was guaranteed not to earn or lose money on the basis of changing price spreads, for any profit the A-to-B trade might yield would be offset by an equal loss on the B-to-A trade. But, if the trader chose to pay for transmission, the transaction would generate MLSA. See Penalty Assessment Order at ¶ 45. Jurco saw the potential:

jurco831: nice
jurco831: load up [i.e., trade in large volumes]
jurco831: net flat [i.e., no profit or loss from price spreads]
jurco831: collect [i.e., get MLSA]
jurco831: that is dirty dirty
jurco831: but legal I guess

         Staff Report at 17. Later that day City Power began placing round-trip trades with paid transmission reservations. Sure enough, the MLSA exceeded the cost of transmission (and other transaction costs) and so City Power got paid for its “net flat” trades. All told, over the course of July 2010 City Power collected $455, 730 in net profit from round-trip trades. Penalty Assessment Order at ¶¶ 47-48.

         City Power developed a second type of “loss trade” in short order. This one entailed placing UTCs between two nodes named SOUTHIMP and SOUTHEXP, which were “import and export pricing points of the same PJM interface, and which ha[d] equivalent prices in both the day-ahead and real-time markets.” Id. at ¶ 49. The result of this equivalent pricing, as Tsingas observed in a message to Jurco, was that “SOUTHIMP-SOUTHEXP settles at $0 all the time, DA [day-ahead] and RT [real-time].” Staff Report at 22. Once again, then, a UTC trade would yield no profit or loss, but could generate MLSA. City Power traded SOUTHIMP-SOUTHEXP for roughly a week in early July and collected $106, 401 in net profit. Penalty Assessment Order at ¶ 49. Tsingas remarked on their success during this period:

traderyoda: these losses paid well the few days we had 2, 000 mw’s
jurco831: great
traderyoda: as in 100k plus
traderyoda: feels sleazy
jurco831: wow

         Staff Report at 25. But Jurco became concerned about this particular trading path:

jurco831: back to the losses thing - I feel really funny about the southimp-southexp but I think you’re right about the other deals

Id. Jurco worried that the SOUTHIMP-SOUTHEXP trades “could be great ammo” for PJM’s Market Monitor, the independent entity tasked with ensuring the competitive and efficient operation of the market. Penalty Assessment Order at ¶ 50; see also Elec. Power Supply Ass’n v. FERC, 391 F.3d 1255, 1260 (D.C. Cir. 2004) (describing market monitors). City Power stopped trading this path in mid-July. Penalty Assessment Order at ¶ 50.

         City Power’s third (and final) type of loss trade involved trading between nodes NCMPAIMP and NCMPAEXP. Although these two nodes did not always have identical prices, they had historically experienced only very small differences. Id. at ¶ 51. Unlike SOUTHIMP-SOUTHEXP, then, this path could result in spread gains or losses, but they would reliably be minimal. As it turned out, City Power’s NCMPAIMP-NCMPAEXP trades in the second half of July 2010 earned $100, 642 through price-spread gains. Id. at ¶ 52. Those gains, however, were wiped out (and then some) by the transaction costs of placing the trades. Id. (noting transaction costs of approximately $532, 060). Once again, it was the MLSA payments that made these transactions profitable: City Power’s net profit after receiving MLSA was $716, 227. Id.

         Throughout July 2010 Tsingas expressed amazement and concern about the volumes of transmission being reserved by others that he suspected were engaged in similar trading:

traderyoda: the amount of trans sold has gone up over 2 fold over ...

Buy This Entire Record For $7.95

Download the entire decision to receive the complete text, official citation,
docket number, dissents and concurrences, and footnotes for this case.

Learn more about what you receive with purchase of this case.