How to Build Economic Baskets From Supply-Chain Exposure
August 20, 2026
Altsets
Research by Altsets Research
Construct baskets around measurable customer or supplier relationships and weight members by economic exposure instead of using only themes or market capitalization.
Data used:Altsets Supply Chain Intelligence: 90k+ entities, 400k+ relationships, 20+ years of history.
Key findings
- A relationship-defined basket can use supplier revenue percentage as an auditable inclusion and weighting input for companies exposed to the same customer.
- Within the bounded Nvidia example, normalizing 27.88% SK Hynix exposure and 17.62% Micron exposure produces illustrative relationship weights of about 61.3% and 38.7%.
Build an economic basket by grouping companies with direct exposure to the same customer or demand source, then comparing how large that exposure is relative to each supplier. This is more measurable than starting only with a thematic label such as artificial intelligence, electric vehicles, semiconductors, or cloud infrastructure.
Nvidia provides a simple worked example. Altsets estimates: SK Hynix to Nvidia at 27.88% of SK Hynix revenue and Micron Technology to Nvidia at 17.62% of Micron revenue. Both suppliers therefore have a meaningful mapped revenue relationship with the same customer. A relationship-weighted basket can use those percentages as one input instead of assigning equal weights or relying only on market capitalization.
What makes an economic basket different
The basket is defined by a relationship, not by a story. A traditional "AI suppliers" basket might include companies because analysts associate them with artificial intelligence. A relationship-based basket can require a measurable connection to one customer, customer group, product cycle, or economic driver.
That makes the inclusion rule auditable. The basket can still be wrong as an investment strategy. The point is that its membership comes from a defined economic screen.
Start with a precise exposure rule
For example: Include suppliers with a mapped Nvidia relationship and visible supplier revenue percentage above a chosen threshold.
With only the two supplied examples, both SK Hynix and Micron would qualify above a 10% threshold. That does not imply they are the only qualifying Nvidia suppliers. It means they are the qualifying names within this bounded example. The distinction between a sample and a complete universe should stay explicit.
One simple weighting method
A researcher could normalize the two supplier revenue percentages inside the example basket. The total displayed exposure is: 27.88 + 17.62 = 45.50
Normalized exposure weights would therefore be approximately:
- SK Hynix: 61.3%
- Micron Technology: 38.7%
Those are not portfolio recommendations. They illustrate how relationship exposure can replace equal weighting inside a defined research basket. A different method could cap weights, combine exposure with liquidity, or use relationship size instead.
Why market-cap weighting answers a different question
Market capitalization measures company value. It does not measure how much of a supplier's business is tied to the basket's economic driver. A very large company with a small relationship can dominate a market-cap-weighted basket even if the selected driver represents only a small part of its business.
Supplier revenue percentage reverses the emphasis. It asks how concentrated each supplier is on the chosen customer or demand source. Neither weighting system is universally better. They represent different research objectives.
Relationship size can prevent percentage-only distortions
A high supplier revenue percentage can belong to a relatively small company or relationship. Adding relationship size helps test absolute scale. In the Nvidia example, SK Hynix to Nvidia is estimated at 21B USD, while Micron to Nvidia is estimated at 9.8B USD.
Both relationships are large in absolute terms as well as meaningful in supplier revenue percentage. That makes them stronger examples than a high percentage attached to a very small relationship.
Baskets can be built around customers, suppliers, or paths
The same framework can be adapted. A customer-demand basket groups suppliers exposed to the same customer. A supplier-dependency basket groups customers exposed to the same supplier.
A geographic basket groups relationships tied to a defined country screen. A second-order basket groups companies connected through a defined two-hop path. The inclusion rule should match the economic question. Mixing several unrelated rules into one basket makes the output harder to interpret.
Historical snapshots can turn a basket into a research signal
If the same relationships are available through time, the basket can be reconstituted at prior dates. That opens questions such as which suppliers are becoming more exposed to the customer, which relationships are shrinking, whether the customer is broadening its supplier base, and whether market performance changes alongside customer concentration. Those questions require point-in-time historical values. A current snapshot alone should not be presented as historical evidence.
A repeatable economic-basket workflow
- Define the economic driver.
- Define relationship direction.
- Define the minimum exposure rule.
- Build the qualifying company set.
- Choose a weighting method.
- Add relationship size as a scale check.
- Cap concentration if the research design requires it.
- Rebalance only on defined dates.
- Use point-in-time data for any backtest.
- Keep basket construction separate from investment recommendation.
The customer read-through guide explains why supplier revenue percentage is useful for customer-driven research. The shared-customer portfolio guide approaches the same network from the opposite direction by looking for common demand overlap inside an existing portfolio.
For metric definitions and point-in-time limitations, read the Altsets supply-chain data methodology. Browse Supply-Chain Data Use Cases for other methods.