The 4-layer pressure model: how Sirius measures freight cost pressure

4-layer contribution to composite reading - illustrative example

Layer
contributions

~120

Composite
reading

Fuel

Capacity

Surcharge

Values are illustrative. Actual layer contributions vary by lane, modality, and market conditions.

The formula

The Sirius index uses a weighted linear combination of four pressure layers. The formula is:

FCPI = (fuel × w_fuel) + (capacity × w_cap) + (surcharge × w_surch) + (fx × w_fx)

Base = 100 at January 2026. Weights sum to 1.0 per lane. Weights differ across the 10 trade lanes.

Each of the four terms represents a pressure layer reading normalized to the January 2026 baseline. The weights express how much each layer contributes to the composite for that specific trade lane. The result is an index where 100 represents the cost pressure level of January 2026 and every point above or below 100 represents a percentage point deviation from that baseline.

The formula is intentionally simple. The complexity lives in the input construction - how each layer reading is derived from its source signals - and in the weight calibration, which requires both domain knowledge and empirical validation.

Fuel pressure layer

The fuel pressure layer captures the cost of energy inputs to freight transport. Different modalities use different fuel types, and the fuel layer weights these by modal mix per lane:

Ocean freight: Very Low Sulfur Fuel Oil (VLSFO) and MGO bunker prices, sourced from the Singapore bunker market as the primary reference point for major Asia-linked trade lanes.
Air freight: Aviation jet fuel (ATF/Jet-A1), using the IATA weekly global jet fuel price index as the primary published signal.
Road freight: Regional diesel prices (EU, US, Australia, Southeast Asia, Korea), sourced from government and public agency data releases.

The fuel layer is the most responsive of the four to sudden market events. A geopolitical shock that disrupts oil supply will typically appear in the fuel layer within 1-2 weeks. This is why fuel carries the highest weight on lanes where modal mix is heavily ocean- or air-freight-dependent.

The fuel layer is computed as the trade-weighted average of these fuel input prices, normalized to the January 2026 baseline. A reading of 115 on the fuel layer means the fuel cost inputs are 15% above the January 2026 level.

Capacity pressure layer

The capacity pressure layer measures how tight the available freight capacity is relative to demand on the lane. Unlike fuel, which is observable as a price, capacity pressure is a derived signal built from multiple indicators:

Ocean capacity: Container vessel utilization rates, blank sailing announcements, vessel transit data from the IMF PortWatch chokepoint monitoring system (Hormuz, Suez, Malacca), and port throughput data from major hubs on the lane.
Air capacity: Belly capacity availability, cargo load factors, and regional air cargo demand indices sourced from public carriers and aviation data agencies.
Chokepoint restriction: When a major chokepoint is restricted or diverted (as in the current Hormuz situation), an explicit capacity pressure adjustment is applied based on the estimated throughput reduction and the proportion of lane volume that transits that chokepoint.

Capacity pressure typically builds more slowly than fuel pressure and resolves more slowly too. A vessel utilization spike driven by a carrier blank sailing program can persist for 8-12 weeks. This slower dynamics is why capacity weight tends to be highest on lanes with structural tightness - where the balance between supply and demand is already thin before any crisis event begins.

Surcharge pressure layer

The surcharge pressure layer captures carrier-declared surcharge activity beyond the fuel and FX mechanisms. These are surcharges that carriers apply in response to specific operational cost events or market conditions:

General Rate Increases (GRI) and Peak Season Surcharges
War Risk premiums on affected corridors
Emergency Operational Surcharges (EOS) related to congestion or disruption
Port Congestion Surcharges (PCS) at major hubs
Piracy and security surcharges on affected routes

The surcharge layer is the most event-driven of the four. It can spike rapidly when a carrier declares an emergency surcharge and resolve equally quickly if the carrier withdraws it. This volatility means the surcharge layer tends to carry moderate weight in the formula - enough to capture genuine market signals without allowing individual carrier pricing decisions to dominate the index.

Surcharge data is sourced from publicly observable carrier announcements and rate news sources. Only surcharges with verifiable public documentation are incorporated. Unverified or unconfirmed surcharge reports are excluded to maintain data integrity.

FX pressure layer

The FX pressure layer captures the cost impact of currency movements on freight invoices. Freight is largely priced in USD, but the underlying operational costs of shipping to and from different regions involve local currencies. When those currencies move significantly against USD, the effective cost pressure changes even if the USD-denominated index rate appears unchanged.

The FX layer uses trade-weighted currency pairs relevant to each specific lane. For the Far East to Europe lane, the EUR/USD and CNY/USD pairs carry high relevance. For the Middle East to Far East lane, the AED/USD and USD/JPY pairs are primary. The weights within the FX layer reflect the currency composition of costs and invoices on that corridor.

The FX layer tends to carry lower weight than fuel and capacity on most lanes, reflecting that FX pressure is a secondary amplifier rather than a primary driver of freight cost pressure in most market conditions. However, during periods of significant USD strength or weakness, the FX layer can contribute materially to the composite reading - particularly on lanes where non-USD operational costs are proportionally high.

CAF vs FX pressure: The FX layer in the Sirius model is not the same as the Currency Adjustment Factor (CAF) on your carrier invoice. The CAF is a specific billing mechanism with its own update cycle and formula. The FX layer in Sirius measures the underlying market pressure that eventually flows through the CAF. The two move in the same direction but with different timing.

Per-lane weight calibration

The four-layer weights are not uniform across all 10 trade lanes. They are calibrated per lane to reflect the structural characteristics of freight cost composition on that corridor. Lanes with higher ocean concentration and longer transit times carry higher fuel weight. Lanes crossing active conflict zones carry higher surcharge weight. Lanes with significant cross-currency invoice exposure carry higher FX weight.

The initial weights were set through a combination of:

Analysis of the composition of freight invoices on each lane, decomposing fuel, capacity, surcharge, and FX components as a proportion of total variable freight cost.
Sensitivity analysis against historical price data to identify which layers have historically explained the most variance in actual freight rate movements on each lane.
Domain expert calibration to ensure the weights reflect known structural features of each corridor that may not be fully captured in historical data.

The weights are fixed between RL loop updates (see the next section) and are published in the Sirius Pro report for transparency. Users who believe their specific trade profile differs significantly from the default weight set can apply the layer readings directly against their own weight assumptions.

How weights update via the RL loop

The Sirius model includes a reinforcement learning feedback loop that compares weekly forward projections against their eventual realized outcomes. Over time, this comparison reveals whether the current weight set is producing accurate projections or whether systematic deviations suggest that the weights need adjustment.

The RL loop operates on a quarterly update cycle, not a weekly one. This is deliberate. Updating weights weekly would cause the model to over-fit to recent events and lose its long-run calibration. A quarterly update cycle allows enough data to accumulate for statistically meaningful comparison before any adjustment is made.

The RL adjustment mechanism is bounded: individual weights cannot move more than a defined threshold per quarter, and the total weights must continue to sum to 1.0 per lane. This prevents the model from drifting into configurations where one layer dominates the composite due to a single quarter of atypical market behavior.

Each quarterly weight update is published with a changelog describing which lanes were adjusted, by how much, and what forecast deviation drove the adjustment. This maintains the transparency and auditability principle that underpins the Sirius model design.

RL loop vs crisis adjustment: The RL loop adjusts the structural weights that reflect the long-run cost composition of each lane. It does not adjust the current-period readings for known crisis events. Crisis event pressure is captured in real time through the signal inputs to each layer. The RL loop is a slow-cycle structural calibration, not a fast-cycle event response mechanism.

What a reading of 118 means

A Sirius composite reading of 118 on a trade lane means that the weighted combination of fuel, capacity, surcharge, and FX pressure inputs is 18% above the January 2026 baseline level. This is the correct and complete interpretation of the number.

Several things it does not mean are worth stating explicitly:

It does not mean your freight bill is 18% higher. The Sirius index measures variable cost pressure on the inputs to freight pricing. Your actual freight bill change depends on your contract structure, your carrier's specific surcharge formula, and the proportion of your freight cost that is variable versus fixed.
It does not measure base freight rates. Those are separately contracted and not included in the index. The Sirius index measures the variable surcharge layer only - typically 30 to 52% of total freight cost depending on modality and lane.
It is not lane-specific to your origin-destination pair. Each trade lane is a corridor average across multiple port pairs. Individual routes within the lane may experience higher or lower pressure.

The practical implication of a reading of 118 for a freight manager: if your variable freight cost components (fuel surcharge, currency adjustment, capacity surcharge) were at a certain level in January 2026, they are approximately 18% higher now. If those variable components represent 40% of your total freight spend on that lane, the reading translates to approximately a 7% increase in total freight cost - before considering any fixed component changes, carrier-specific formula differences, or modal mix effects.

For budget use: The Sirius Pro report includes a P&L impact calculator that translates a composite reading into a freight cost delta for a given spend level, modal mix, and contract structure. The calculator applies the appropriate transmission coefficients rather than applying the index change directly to total spend.

Reading reference

Reading	Interpretation
100	At January 2026 baseline. No elevated variable cost pressure.
105-110	Moderate elevation. Within typical seasonal variance. Monitor.
110-120	Significant elevation. Review contract renewal timing. Check modal alternatives.
120-135	High pressure. Active cost management warranted. Review routing and carrier mix.
>135	Crisis-level pressure. Typically associated with active chokepoint disruption or major supply shock.
<100	Below January 2026 baseline. Cost pressure has moderated. Review locked-in rates for renegotiation opportunity.

Thresholds are for orientation only. Actual response thresholds depend on your freight cost structure and budget sensitivity.