Can Two Quant Strategies Be Crowded Into the Same Customer Without Looking Correlated?
September 14, 2026
Altsets
Research by Altsets Research
Yes. Different models can own different stocks and still converge on the same customer, supplier, or bottleneck, creating hidden dependency crowding that ordinary strategy labels and normal-period return correlations may not reveal.
Data used:Altsets Supply Chain Intelligence: 90k+ entities, 400k+ relationships, 20+ years of history.
Key findings
- The supplied network shows how one strategy holding Micron and another holding SK Hynix can both depend materially on Nvidia even when the securities and alpha signals differ.
- Portfolio-crowding research shows that construction choices can create common exposures and systemic risk, while a supply-chain graph adds a dependency-space definition of crowding below the ticker and strategy labels.
Yes. Two quant strategies can look diversified by signal, sector, or historical return correlation and still be crowded into the same economic dependency. If one strategy buys Micron and another independently buys SK Hynix, both strategies can end up relying on Nvidia demand even though they were generated by different models.
Different alphas can converge on the same outside company
Imagine one strategy selects Micron because of earnings revisions while another selects SK Hynix because of momentum. The strategies appear differentiated because the signals are different and the securities are different. The supplied Altsets data shows another layer: Nvidia represents 17.62% of Micron revenue and 27.88% of SK Hynix revenue in the displayed relationships.
That common node creates economic crowding that the strategy labels do not reveal. A major Nvidia demand shock can hit both positions through the same customer path. Historical return correlation may capture some of that dependence, but it can remain modest during normal periods and rise when the shared customer becomes the active source of information.
Crowding should be measured across strategies, not only inside one portfolio
Traditional crowding research studies how many managers or portfolios converge on the same assets and how portfolio-construction processes can amplify common exposures. A multi-strategy quant fund has another version of the problem: different models can select different stocks that ultimately depend on the same external company.
A supply-chain crowding matrix can measure overlap between strategies in dependency space. Each strategy can be mapped to its important customers, suppliers, or bottlenecks using the holdings it currently owns. Two strategies receive a higher crowding score when their holdings repeatedly converge on the same outside nodes even if they share no securities.
Relationship direction changes what kind of crowding exists
Two suppliers sharing a major customer are crowded into a common demand source. Two customers sharing a critical supplier are crowded into an upstream bottleneck. Those structures should not be collapsed into one generic overlap count because the shock mechanisms differ.
Directional metrics can refine the score without becoming portfolio weights. The Nvidia exposure of Micron and SK Hynix can indicate that the shared-customer overlap is economically meaningful. Structural-only relationships can still identify common paths but should not receive invented magnitude. The crowding model should preserve the same evidence rules as the underlying relationship data.
Hidden crowding can explain why supposedly independent strategies fail together
A common frustration in multi-strategy portfolios is that several models become correlated at exactly the wrong time. The reason can be more than broad risk-off behavior. If several independently designed strategies select companies exposed to the same customer or supplier, one network shock can synchronize their losses.
That makes dependency crowding useful as a risk-control layer. A portfolio allocator can cap how much aggregate strategy risk points toward one external company, or it can charge a higher internal risk budget when several strategies converge on the same node. The strategies remain independent at the alpha layer while the portfolio manager recognizes their shared economic plumbing.
The conclusion is that strategy diversification can fail below the ticker level
Yes, two quant strategies can be hiddenly crowded even when they own different stocks and use different signals. Map each strategy's holdings into customer, supplier, and bottleneck space, then measure whether the supposedly independent books converge on the same external companies. That gives the allocator a second definition of diversification that historical strategy correlation alone may not reveal until the shared dependency is activated.
The covariance-model guide explains why network connections can matter more during specific shocks than in average return correlations. The portfolio external-company guide explains how one company outside the portfolio can become economically important across several holdings.
For relationship definitions and evidence limits, read the Altsets methodology.