Metrics and Trading Strategies in World Natural Gas Spot and Forward Markets Hélyette Geman University of London & Johns Hopkins University Member of the Board of the UBS- Bloomberg Commodity Index To be presented at the Bachelier World Congress Brussels June 2 to 6, 2014
Commodities as Three Classes of Individual Commodities: Energy, Metals and Agriculturals Commodities have existed itdat least as long as humankind The years 2002 to 2005 for crude oil, then copper and other base bs metals, 2006 and 2007 for agriculturals saw gigantic rises in prices and the so-called financialization of commodities. The financial crisis sent all prices down during the second half of 2008. As of 2009, commodity prices have rebounded differently. Since 2013 and 2014, commodities have returned to their specific attributes, away from financialization Agricultural Commodities, as the stated subject of China (FT, May 29) and Natural Gas, because of the US developments, are currently receiving a major attention
CRB Index 1985 to 2014
US Natural Gas and Oil US imports of natural gas and crude oil have fallen 32% and 15% respectively in the past five years, narrowing the US trade deficit The International Energy Agency in Paris estimates that for the first time last year since 1982, the US produced more natural gas than Russia. It also predicts that in 2015, the US will become the first oil producer intheworld,ahead h d of SaudiArabia and Russia In Europe, big industrial consumers are turning to the international wholesale market as opposed to the national gas companies In Australia, Coal Seam Gas (CSG) is another major source of energy. New projects in East Coast Australia will increase LNG capacity by 150% and by 2016, Australia will be a major player and is likely to relieve pressure on Asian prices, Japan in particular North America will emerge as a gas swing country very soon
US Crude Oil & Fuels Production in 2013
Defragmentation of Natural Gas Markets Until recently, three regional markets could be identified in the world, with limited trade between them because of the cost of transportation of gas over long distances As a comparison, it costs only $3 to 4 to transport a barrel of oil around the world, i.e., less than 5% of the price of abarrelforprices over $60. In the case of NG, it may represent 100% of the price or more: less than $4 per Mbtu in the US, 8 to 12 in Europe, 16 to 18 in Japan Increased LNG is bringing these regions together, Americas and Europe in a first stage
Nymex Natural Gas Forward Curve, Oct 2009
NYMEX Natural Gas Prices 1991 to 2014
6 5 4 3 2 1 0 NG 14 Aug 2012 7 NG 1 Sep 12 1 Dec 12 1 Mar 13 1 Jun 13 1 Sep 13 1 Dec 13 1 Mar 14 1 Jun 14 1 Sep 14 1 Dec 14 1 Mar 15 1 Jun 15 1 Sep 15 1 Dec 15 1 Mar 16 1 Jun 16 1 Sep 16 1 Dec 16 1 Mar 17 1 Jun 17 1 Sep 17 1 Dec 17 1 Mar 18 1 Jun 18 1 Sep 18 1 Dec 18 1 Mar 19 1 Jun 19 1 Sep 19 1 Dec 19 1 Mar 20 1 Jun 20 1 Sep 20 1 Dec 20
4.8 4.7 4.6 4.5 4.4 4.3 4.2 4.1 4 3.9 Henry Hub Natural Gas Forward Curve May 2013 Jul 2013 Oct 2013 Jan 2014 Apr 2014 Jul 2014 Oct 2014 Jan 2015 Apr 2015 Jul 2015 Oct 2015 Jan 2016 Apr 2016 Jul 2016 Oct 2016 Jan 2017 Apr 2017 Jul 2017 Data from: http://www.cmegroup.com/trading/energy/natural gas/naturalgas_quotes_settlements_futures.html
Qatar to launch first market for LNG Qatar has launched the first market for trade in LNG based at IMEX, an energy exchange located in Doha, the capital of Qatar The reserves of North Field, between the territorial waters of Qatar and Iran, represent 25 trillion cubic meters of natural gas, or 25% of the world reserves Many liquefaction units have been built, where the gas is heated, then cooled ldto the temperaturet of -162 degrees, then stored dtemporarily Moreover, Qatar has got a fleet of 75 methane tankers, with a capacity of 265,000 tonnes and liquefaction capabilities onboard. However, the world gas market is now in excess supply of 100bn cubic meters, and LNG tankers are touring the world, contributing to a price collapse in European spot markets
Nymex Natural Gas 2005 to 2013
The Forward Curve The set {F T (t), T > t} is the forward curve prevailing at date t for a given commodity in a given location It is the fundamental tool when trading commodities, as spot prices may be unabservable and options not always liquid It allows to identify possible «carry arbitrage» : buy S, sell a future maturity T and pay the cost of storage and financing as long as the net cashflow is strictly positive The shape of the forward curve is atany date t in a one-to-one mapping with the convenience yield y It will reflect the seasonality in the case of seasonal commodities such as natural gas
Crude Oil Forward Curve Feb 2010 The Contango is nearly as steep as at the end of 2008, offering carry trade opportunities again
Shale Gas Basins in the US
Trading Strategies in US Natural Gas The most popular ones used to be calendar spreads Futures: liquidityidi has been very high 120 months out since 2007 at least The typical strategy is to be long Winter months, short Summer months and expect weather events to change the spread, as it happened during the months of January to March 2014 in the US In 2005, the trader Brian Hunter made $1.25 billion (and more than $1 million for himself) when the hurricanes Rita and Katrina destroyed d oil platforms in the Gulf of Mexico, creating a spike in winter natural prices very high by substitution effects, hence in turn a jump in thesizeof the Futures spreads he was holding while inventories for physical gas were very low The following year, he multiplied li the size of his positions i incalendar spreads by 10, but there was no hurricane
The Amaranth Hedge Fund Saga During the week of September 15 2006, Amaranth, a respected diversified multi-strategy hedge fund, lost 65% of its $9.2 billion assets When the fund transferred its energy positions to JP Morgan and Citadel, it was at a discount of $1.4 billion to their Sept 19 mark-tomarket value, which meant a lost of about $6 billion since the end of August for the fund's investors Besides the non- occurrence of hurricanes, gas storage facilities were quite full in August 2006 Commodities are not a two-sided flow like stocks
9,5 9 8,5 8 7,5 7 65 6,5 6 NYMEX Natural Gas Curve - March 2007 Nymex Natural Gas Price avr-07 juil-07 oct-07 janv-08 avr-08 juil-08 oct-08 janv-09 avr-09 juil-09 oct-09 janv-10 avr-10 juil-10 oct-10 janv-11 avr-11 juil-11 oct-11 janv-12 avr-12 août-12 nov-12 Contract months
Dynamic Strategies in Natural Gas and Other Seasonal Commodities Borovkova & G (Review of Derivatives Research, 2006) Extend their analysis of seasonal commodities forward curves and wish to disentangle the seasonal and stochastic component of the forward premia Introduce as a first state variable the average value of outstanding (liquid) Futures prices as the first state variable to represent the "backbone" of the forward curve A family of seasonal premia s(t,t) attached to calendar months, T = 1, 2,, 12 A term structure of stochastic forward premia p(t,t) γ (t, T) where the aggregation of s(t, T) and p(t,t) allows the traditional" cost of carry to be extended This representation is used to calibrate NG and electricity forward curves
The first state variable is as Seasonal Cost of Carry Model ln F the average forward price prevailing at date t, and defined 1 N N ( t ) = ln F ( t, T ) T = 1 where N is the most distant liquid maturity, and a multiple of N for seasonal commodities. Hence, F ( t ) does not have any seasonal feature. Next, we define the seasonal premia s(m), for M = 1,..., 12 as the set of long-term average premia (expressed in %) in ft futures expiring ii in the calendar month M over the average forward and assume these premia to be deterministic Lastly, we introduce a stochastic cost of carry net of seasonal premium γ(t, T), defined by F ( t, T ) = F ( t ) exp [ s ( T ) ( T t ) γ ( t, T ) ] Note that in contrast to γ, s(t) is an absolute quantity attached to the delivery month
Term structure of stochastic forward premium volatilities, UK natural gas futures 0.25 Term structure of convenience yield volatility η τ (t) 0.2 0.15 η τ (t) 0.1 0.05 0 1 2 3 4 5 6 7 8 9 10 11 12 Time to maturity τ
Natural Gas seasonal premia 0.3 Seasonal component, Natural Gas futures 0.25 0.2 0.15 0.1 0.05 0 0.05 0.1 0.15 0.2 1 2 3 4 5 6 7 8 9 10 11 12 Calendar month
12 Strategies Involving two Markets at the same time : HH and NBP Forward Curves on 8/13/2010 10 USD/M MMBtu 8 6 4 HH NBP 2 0 0 10 20 30 40 50 60 Months to maturity
12 HH and NBP Forward Curves on 7/31/2013 10 USD/MM MBtu 8 6 4 HH NBP 2 0 0 10 20 30 40 50 60 Months to maturity
Bo and G.(2014) Introducing Distances in Commodity Spot and Forward Markets Firstly, test whether h the locational l spread between the Henry Hb Hub and National Balancing Point can be better estimated by the difference of two Futures contracts or the difference between the Averages with the seasonality adjustment Use the result in favour of the latter (even looking for the optimal lag between the Future maturity and the spot date) to exhibit a consistently profitable strategy trading the spread between the two indexes and paying the LNG tanker cost
NBP (higher) and HH forward curves
Calendar Spread Strategies against a Gas Storage Facility G Mazieres Hbb Hubbert (2014) revisit the valuation of gas storage facilities Want to use a 2-dimension model for gas prices As well as more than one volumetric constaint in the problem Use the Least Square Monte Carlo approach initiated by Carriere (1996), Longstaff- Schwartz and many other papers Apply the technique of Tensor of Compactly Supported Basis Funtions by Hubbert and Mazières (2011) to a tensor of Radial Basis Functions Ehibi Exhibit the remarkable kbl results in terms of speed and quality of convergence
1- Contracts Gas storage/swing g valuation Optimal decision problem with daily optionality To find the optimal trade for every time & volume This is to find the maximum value for the contract Value=Payoff now + disc. Continuation Val. use LSMC Contracts American option: compare continuation value with exercise value Gas storage: compare different Values for Inject, Withdraw, Do Nothing Gas Swing contract: same as storage but o Compare, Value for buy trades: 0, ½ or 1 unit of gas (DCQ) o Multi period OCQ constraints & ACQ constraints 32
1- Contracts 6 Mths Swing and Volume Constraints 60 Volume Paths (cummulated trades over 6 months) purchas of gas Volume sed50 40 30 20 10 0 0 20 40 60 80 100 120 140 160 180 200 Days 33
2- Kernels Tensor of Radial Basis Function (TBF) TBF has all properties of RBF, allows the matrix decomposition Two dimensional model for the spot price TBF improves the speed for estimating i the Continuation Vl Value TBF: 34
5- Analysis Swing 4D: TBF & RBF vs benchmark Value (solid line es) CPU tim me (sec.) 360 340 320 300 Swing contract: Impact of No. of Sim. 15 Value (TBF) Value (RBF) Value (Roll.Int.) Std. error (TBF) Std. error (RBF) Std. error (Roll.Int.) 280 0 50 100 150 200 250 300 400 500 600 700 750 800 0 10 4 TBF based model RBF based model 10 2 Roll.Int. based model 10 0 50 100 150 200 250 300 500 750 No.Sim. in backward induction (In addition 5 x No. Sim. for the forward induction) St tandard err ror (dashed d lines) 10 5 X 5 X 5 Both the speed and the standard deviation of the error are greatly improved 35
References H.Geman and P.Vergel (2013) Investing in Fertilizers in Times of Food Scarcity, Resources Policy H.Geman, L. Thukral and J.Wright (2013) Mispricing and Trading Profits in Commodity ETNs, Journal of Trading H.Geman and W Smith (2012) Theory of Storage and Metal Forward Curves Dynamics, Resources Policy, H Geman and S. Sarfo (2012) Seasonality in Cocoa Spot and Forward Markets: Empirical Evidence, Journal of Agricultural Expansion and Rural Development H.Geman ( 2011) Volatility in Commodity Spot Markets: Speculation or Scarcity?, Swiss Derivatives Review H.Geman (2010) Commodities and Numéraire, Encyclopedia of Quantitative Finance H. Geman and S. Ohana (2009) Inventories and Volatility in US Oil and Natural Gas Markets, Energy Economics H. Geman and Yfong Shi (2009) The CEV model for Commodity Prices, Journal of Alternative Investments H. Geman and S. Kourouvakalis (2008) "A Lattice-Based Method for Pricing Electricity Derivatives under a Jump- reversionmodel", Applied Mathematical Finance H. Geman and C. Kharoubi( 2008) Diversification with Crude Oil Futures : the Time-to- Maturity Effect, Journal of Banking and Finance S. Borovkova and H. Geman (2006) "Seasonal and Stochastic Effects in Commodity Forward Curves", Review of Derivatives Research H. Geman and A. Roncoroni (2006) "Understanding the Fine Structure of Electricity Prices", Journal of Business H. Geman (2005) "Energy Commodity Prices: Is Mean Reversion Dead?", Journal of Alternative Investments H. Geman and S. Ohana (2009) "Inventory, Reserves and Price volatility in Oil and Natural Gas Markets,Energy Economics H. Geman (2005) "Commodities and Commodity Prices: Pricing and Modelling for Agriculturals, Metals and Energy", Wiley Finance H. Geman and V. Nguyen (2005) "Soybean inventory and forward curves dynamics", Management Science H.Geman (2004) Water as the Next Commodity, Journal of Alternative Investments H. Geman and M. Yor (1993) "An Exact Valuation for Asian Option", Mathematical Finance A. Eydeland and H. Geman (1999) "Fundamentals of Electricity options" in Energy Price Modelling, Risk Books H. Geman and O. Vasicek (2001) "Forwards and Futures on Non Storable Commodities", RISK H. Geman (2002) "Pure Jump Lévy Processes in Asset Price Modelling", Journal of Banking and Finance H. Geman (2003) "DCF versus Real Option for Pricing Energy Physical Assets" Conference of the International Energy Agency - Paris - March 2003 hgeman@hotmail.com