E-StorMTM forms a part of our EnergyMetricsTM suite of energy risk management and forecasting models used by NERA to help our clients with valuation and decision-making in connection with a range of energy assets, including gas storage, merchant power plants (thermal and hydro), interconnectors, and contracts (plain vanilla, options, swings, swaps, etc.). The power of EnergyMetricsTM is its ability to assess the aggregated risk and return of a mixed portfolio of different assets taking into account stochastic nature of prices for power, fuels, and CO2, as well as other variables. Combined with our expertise in the fundamental economics and competitive dynamics of energy markets, this powerful tool allows us to give clients an independent and insightful appraisal, not just mechanistic modeling results or software solutions. A screenshot of the main EnergyMetricsTM user interface is illustrated below.

Click image to enlarge 

E-StorMTM is a general option valuation model applicable to many types of commodities, but here we highlight its application to gas storage valuation. Rather than optimizing with "perfect foresight," E-StorMTM recognizes that the users of storage have to decide what to do on each day in conditions of uncertainty about future price evolution, as illustrated in the graph below.

Click image to enlarge

The model decides the optimal action each day (i.e., inject, withdraw, or do nothing), based on a probabilistic assessment of the expected value of the storage at the end of the day conditional on each choice (i.e., with more, less, or the same gas in store). To do so the model estimates an expected "continuation value" of the storage at the end of the day, which depends on the current spot price and the level of gas in store, as illustrated in the graph below. The model then picks the action that maximizes the expected sum of today's profit and the continuation value.

Click image to enlarge

The following steps summarize the algorithm used by E-StorMTM.

Step-By-Step Summary
The model uses the following step-by-step procedure:

  1. Forecast a price path using statistical methods. E-StorMTM is completely flexible regarding the choice of the price process. One of the methods we adopted is as follows:
  • Take the forward curve for quarterly gas contracts and extrapolate it into the future towards the random long-term level of gas price, maintaining the quarterly shape observed in the forward curve.
  • Vary this price curve by assuming conditions like 2005/06 (severe winter plus infrastructure shock) occur with a frequency of 2/50 (=1/50 winter + 1/50 infrastructure shock).
  • Add volatility using a "mean reverting Brownian motion," i.e., random fluctuations plus a tendency for high prices to fall and low prices to rise, based on observed measures of volatility and mean reversion in historical prices.
  1. Use Monte Carlo techniques to generate N separate price paths.

Set the values for the following "Terminal Conditions for the end of the modeling period (period t):

  • Level of storage
  • Continuation value at that time

4. Step back one period and calculate expected continuation values for each of the N prices at period (t-1):

  • Calculate continuation values for the current period (for each price and each level of storage), derived from releasing gas in the last period.
  • Conduct a regression of the continuation values against the prices, to establish an expected continuation value for each price and storage level
  • Decide what is the best action to take in period (t-1) -- inject, withdraw or do nothing -- in order to maximize the sum of the current cash flow and expected continuation value in the next period, given the current level of prices, for all feasible levels of storage.


5. Step back one period and repeat the calculations and regressions for period (t-2).

6. Repeat the process back to the current day.

7. Calculate the net revenues derived from the optimal series of actions for each price path.

The combination of the best actions on each day, for each price path, amounts to a series of energy trades (inject/withdraw/nothing = buy/sell/no trade) at the current spot price (adjusted for the costs of injection and withdrawal). The revenue stream to be earned from the storage is the sum of net revenues from these trades, which can be discounted to provide an NPV. With N price paths, we obtain N possible revenue streams and NPVs. We take the mean annual revenue in each year divided by the space of the facility as our prediction of the market price of the storage capacity.

Sean Gammons
Managing Director
NERA London
+44 20 7659 8564

Vakhtang Kvekvetsia
Associate Director
NERA London
+44 20 7659 8746

Name Title Location Phone Email
Sean Gammons Senior Managing Director London
+44 20 7659 8564
+49 30 700 1506 01
Vakhtang Kvekvetsia Managing Director London +44 20 7659 8746
Title Type Author
The Road to Mexico’s Self-Sufficient Energy Reforms Published Article Dr. Jeff D. Makholm, et al.
Fixing Local vs. National Public Interest Contests Requires a Full-Court Press in Con... Published Article Dr. Laura T. W. Olive