Screen-print style poster of a risk gauge over an LNG carrier

CASE STUDIES / RISK · AUG 2​025

Why historical and simulated VaR give different answers.

The study measures risk in the same LNG portfolio using historical price changes and simulated price paths.

$46.1M / $64.5M

ONE-DAY VAR · HISTORICAL / SIMULATED · 5% LOSS TAIL

View as slides (PDF) →

Use the same portfolio.

01 · PORTFOLIO

The model contains 24 contractual positions forming 12 paired journeys. US supply serves Germany and Qatari supply serves the UK, with HH, TTF, Brent and NBP indexation.

CONSTRUCTED EXAMPLE BOOK · INDEX VALUES OF 20 JULY 2​025 · HISTORY JAN 2​024 TO JUL 2​025 · 1,000 SIMULATIONS · PORTFOLIOS ARE NOT RE-OPTIMISED ACROSS PATHS

THE BOOK · 3.5 TO 3.9M MMBTU PER CARGO

FOB purchases · Calcasieu Pass6 CARGOES · 110% HH + $1
DES sales · Brunsbuettel6 CARGOES · 95% TTF
FOB purchases · Qatar6 CARGOES · 9% BRENT
DES sales · South Hook6 CARGOES · 85% NBP
Fleet2 VESSELS · 171K CBM EACH

Compare the two methods.

02 · TWO METHODS

Both methods estimate one-day losses while keeping the operating plan fixed. Historical VaR uses observed price changes; Monte Carlo uses 1,000 simulated scenarios. The 5% VaR marks the loss exceeded in 5% of scenarios. Expected shortfall averages losses beyond that threshold.

$0M$25M$50M$75M$100M46.164.5+39.7%VAR · 5%58.282.1+41.1%VAR · 1%51.476.3+48.4%SHORTFALL · 5%61.190.5+48.2%SHORTFALL · 1%HISTORICALMONTE CARLO · ONE-DAY LOSS, $MSAME BOOK, SAME DAY · HISTORY JAN 2​024 TO JUL 2​025 VS 1,000 SIMULATIONS
Historical and simulated estimates of portfolio loss, in $M.

Examine cargo-level risk.

03 · PER CARGO

The study breaks down the results by cargo and delivery period. The estimates differ because the cargoes carry different pricing indices and timings.

$0M$2M$4M$6M123456789101112CARGO 7 HAS THE LOWEST ESTIMATEIN THIS HISTORICAL COMPARISONCARGO ID · AS LABELLED IN THE SOURCE CHART1% VAR5% VAR · HISTORICAL, PER CARGO, $M
Historical risk by cargo. The source gives conflicting route descriptions for cargo 7.

Compare the same cargo under both methods.

04 · CONSISTENCY

Historical and simulated estimates can rank individual cargoes differently. Their price histories, correlations and delivery periods differ. The study holds the operating plan fixed when calculating these risk measures.

$0M$5M$10M123456789101112SIMULATION IS HIGHER AND FLATTER · HISTORY SWINGS FROM CARGO TO CARGOCARGO · 5% VALUE AT RISK PER CARGO, $MHISTORICALMONTE CARLO
Risk by cargo at the 5% loss tail. Simulated values are read from the source chart, based on 1,000 price paths.

Test a revised UK price formula.

05 · THE LEVER

Changing the UK leg to 50% NBP plus $4 reduces the reported historical VaR from $46.1M to $41.8M. Expected shortfall falls from $51.4M to $46.8M, while forward profit changes by +0.18%.

$0M$10M$20M$30M$40M$50M46.141.8PORTFOLIO4.73.6CARGO 125.05.0CARGO 44.14.9CARGO 6BEFOREAFTER · GREEN LOWER RISK, AMBER HIGHER5% VAR BEFORE AND AFTER RE-INDEXING · FORWARD PROFIT CHANGES BY +0.18%
Reported profit rises from $169.2M to $169.5M. The source inconsistently labels the original UK formula as 80% or 85% NBP.

06 · RESULT

Reported one-day VaR at the 5% loss tail: $46.1M historically and $64.5M under simulation.

The study holds the operating plan fixed for the risk calculation. Historical and simulated estimates use different price scenarios. The UK reference formula differs between source slides, so the revised-formula comparison is reported as a separate source result.

What this comparison shows.

07 · TAKEAWAYS

✓

Compare the two methods

Both methods estimate one-day losses while keeping the operating plan fixed. Historical VaR uses observed price changes; Monte Carlo uses 1,000 simulated scenarios. The 5% VaR marks the loss exceeded in 5% of scenarios. Expected shortfall averages losses beyond that threshold.

✓

Test a revised UK price formula

Changing the UK leg to 50% NBP plus $4 reduces the reported historical VaR from $46.1M to $41.8M. Expected shortfall falls from $51.4M to $46.8M, while forward profit changes by +0.18%.

!

Scope of the result

The study holds the operating plan fixed for the risk calculation. Historical and simulated estimates use different price scenarios. The UK reference formula differs between source slides, so the revised-formula comparison is reported as a separate source result.

How the study was calculated.

08 · CALCULATION

1

Set up the comparison

The model contains 24 contractual positions forming 12 paired journeys. US supply serves Germany and Qatari supply serves the UK, with HH, TTF, Brent and NBP indexation.

2

Calculate the alternatives

The study breaks down the results by cargo and delivery period. The estimates differ because the cargoes carry different pricing indices and timings.

3

Review the result

Reported one-day VaR at the 5% loss tail: $46.1M historically and $64.5M under simulation.

Modelled results depend on the contracts, prices and operating assumptions used.