Enron Weather Derivatives Case Study Solution

Working Through the Enron Weather Derivatives Case Study

I spent a few years looking at Enron's weather derivatives work after they collapsed. Not because I wanted to study failure — more like I inherited a bunch of their modeling files when my firm was cleaning up some counterparty positions. The Cogen platform documentation is actually still floating around on various university servers and in older trading forums. What follows is a practical walkthrough based on actual files I pulled down. The case study solution materials you'll find online are scattered across academic repositories. Most of it lives on sites like MIT's course pages, the CERA case study collection, and occasionally on trading communities where someone uploaded their analysis. The original Enron Cogen whitepapers were buried in the Enron email corpus dump, but they resurfaced through legal discovery and made their way into public domain. If you're a student or someone just starting to understand weather risk, download everything you can at once. The Cogen platform architecture document, the degree-day calculation methodology, the volatility surface papers, and the actual trade examples. They cross-reference each other in ways that matter when you're trying to build a model from scratch.

How the Cogen Platform Actually Worked

Enron's system wasn't some black-box trading terminal. It was built on a relatively straightforward principle: translate weather forecasts into financial payoffs using contract specifications that anyone could read. You selected a location, a timeframe, and a strike temperature. The system then ingested NOAA forecast data, computed the expected heating or cooling degree days, and calculated a premium based on implied volatility derived from historical weather patterns. The real work happened in the pricing engine. Weather derivatives don't behave like equity or commodity options. There's no underlying asset you can hold in a warehouse. The "asset" is a weather statistic for a specific location over a specific period. That means the Greeks don't work the way you'd expect them to. Delta is mostly a function of forecast error, gamma explodes as you approach the measurement period, and vega is nearly meaningless because there's no liquid market to calibrate against.

The Degree Day Calculation — Where People Mess Up

This is the part that trips everyone up. A heating degree day is calculated as 65 minus the daily average temperature, but only when that result is positive. Cooling degree days flip that — 65 minus average temperature, positive results only. The contract aggregates these daily values across the measurement period, and the payoff is the difference between the realized cumulative value and the strike, multiplied by a notional amount per degree day. Here's what nobody tells you in the basic explanations: Enron's Cogen platform used a threshold adjustment that varied by location. The 65-degree base isn't universal. In some contracts, particularly those tied to specific industrial processes, the base temperature was customized. If you're working through this case study and your numbers don't match the solutions, check whether the strike and base temperature are defined the same way in your reference material. I spent two days chasing a discrepancy that turned out to be a 62-degree base instead of 65 in one of the contract specs. Also worth noting: the daily average temperature in these models is typically computed as (high + low) / 2. Some sources use the more accurate method of integrating temperature over time using actual hourly readings, but Enron's standard contracts used the simpler high-low average. If you're building your own model from this case study, stick with high-low unless you have access to fine-grained observational data.

Volatility and the Implied Curve Problem

The most counter-intuitive thing about weather derivatives pricing is how thinly the volatility surface is populated. With interest rate options, you can observe prices across dozens of strikes and expirations simultaneously. With weather, each contract is essentially unique — different location, different period, different base temperature. This means Enron couldn't calibrate a traditional implied volatility surface the way a derivatives desk would for anything else. Instead, they used a hybrid approach. Historical weather data for the specific location and period provided a statistical baseline for expected variance. Then they layered in option-implied vols where liquid contracts existed, mainly the CME-listed heating and cooling degree day futures. The gap between those two inputs — the historical statistical vol and the market-implied vol — was where the actual trading alpha lived, and also where most of the risk came from. I worked with a model that tried to replicate this calibration process. The historical component used a bootstrapped distribution of degree day totals from 20 years of station data. The implied component used CME settlement prices. Where the two diverged significantly — and they often did by 15 to 30 percent — the spread represented either mispricing or a structural difference in how the market perceived tail risk versus how history actually behaved. Picking which one to trust was the job, and I've seen both sides get burned.

Build vs. Buy When Recreating This

If you're going through the Enron Weather Derivatives Case Study Solution and trying to implement a working model, you have two paths. The quick path is to use existing open-source libraries. Python has weather derivative pricing packages, and there's a reasonably maintained one called pyweather that handles degree day calculations, payoff structures, and basic Monte Carlo simulation. It'll get you through the case study in a weekend. The proper path takes longer. You pull raw station data from NOAA's ISD or GHCN databases, compute your own degree days, build your own historical distribution, and then overlay a volatility surface using whatever market data you can access. This approach usually takes two to three weeks for someone who knows the tools, maybe a month for someone learning as they go. But the output is actually usable rather than just homework-quality. There's a third option that I ended up recommending more often than not: use the CME's official settlement methodology as your reference implementation. The exchange publishes its calculation rules, and they're more rigorous than most academic treatments of the subject. If your numbers align with CME's methodology, you're in the right ballpark.

Where This Approach Breaks Down Completely

Don't walk away thinking weather derivatives are a clean, solvable problem. They aren't. The core issue is that weather is inherently non-stationary. A model calibrated on 20 years of historical data assumes the future will resemble the past. With climate change shifting baseline temperatures, that assumption has been getting worse every year. Enron's own documentation acknowledged this tension but had no good way to price it in. Another hard limitation: liquidity. Outside of a handful of CME-listed contracts centered on major metro areas, weather derivative markets are thin to nonexistent. If your case study involves a location that doesn't have an active futures contract, you're pricing in a vacuum. The numbers you produce will be internally consistent but financially arbitrary. This isn't a model problem — it's a market problem, and no amount of coding will fix it. A third blind spot that catches people out: the correlation structure across multiple locations. Enron's Cogen platform allowed portfolio-level contracts spanning several regions. The value here depends entirely on how accurately you model the spatial correlation of weather patterns. Most textbook treatments hand-wave this. In practice, getting the correlation matrix wrong by even 10 to 15 percent can swing your portfolio hedge ratio significantly, and there's no single authoritative source for the correlations you need. I ended up building mine from cross-correlated station pairs, which took longer than I wanted to admit.

Practical Tips That Aren't in the Case Study

First, always verify your temperature station data for gaps and anomalies. NOAA data is generally reliable but not immune to sensor errors, relocation artifacts, and periods of missing observation. A bad station can throw off your entire degree day calculation for a period. Run a quality check — compare against neighboring stations, flag any jumps larger than three degrees between consecutive days, and fill gaps using interpolation before computing anything. Second, don't underestimate the computational cost of Monte Carlo simulation for weather derivatives. A decent accuracy run with 100,000 paths across multiple locations and time periods can take hours on a standard machine if you're doing it naively. Vectorize your operations. Use numpy efficiently. I cut my runtime from about 45 minutes down to under six by restructuring the simulation loop. The difference isn't subtle. Third, if you're comparing your results to the published case study solution and they're close but not exact, check your leap year handling. Some implementations skip February 29 in their daily aggregation, which introduces a small but measurable bias over multi-year periods. It won't ruin your model, but it will make your numbers drift from reference solutions.

Enron Weather Derivatives Case Study Solution — Where to Download

The main repositories for the solution materials are the Carnegie Mellon University digital library, the Texas A&M CERA case study archive, and several open courseware platforms that host the original problem sets with worked solutions. Search for "Enron weather derivatives Cogen case study" and you'll find academic papers with supplementary data files attached. The PDFs themselves are freely available, and the spreadsheet templates that accompany them are usually in .xls format, so if you're on a modern system you may need to convert them. LibreOffice handles the conversion without issue. The most complete single source I found was a package hosted through an economics department server that included the case study text, the Cogen platform overview, sample contracts, and a worked solution set with full derivations. It hasn't been updated since around 2010, but the material is still technically accurate for understanding the mechanics. The formulas haven't changed, even if the market landscape around weather derivatives has.

What to Actually Take Away From This

The Enron weather derivatives case study is valuable because it forces you to confront pricing in a market with no traditional underlier, no deep liquidity, and a payoff structure that's transparent but mathematically messy. That combination is rare in finance education. Most case studies give you a clean framework with clear inputs and an unambiguous answer. This one gives you a framework, questionable inputs, and an answer that's only as good as your assumptions. The practical skill you walk away with is learning to distinguish between what the model can tell you and what the model cannot tell you. Weather derivatives pricing will always have a large subjective component wrapped around the objective calculation. The degree days are objective. The volatility you plug in, the correlation matrix you choose, the station data you trust — those are judgments. Good practitioners make those judgments explicitly rather than hiding them behind the appearance of mathematical precision. If you finish the case study and feel like something essential is missing, that feeling is accurate. The published solutions cover the mechanics thoroughly but skip over the operational realities — data quality issues, model risk, the fact that you'll sometimes need to trade a position before your model has fully calibrated. Those gaps are where the actual work happens.