Traditionally, hedge fund "activity" or "success" has been measured by alpha. Alpha, loosely speaking, is what cannot be "explained" by correlations in fund returns to benchmarks or factors. If a manager invests all capital in an S&P 500 ETF, for example, the returns of the S&P 500 should explain all of the fund's returns; the alpha should be zero. That manager might have great returns (because the index gained, say, 30% that year) but he would be considered "unskilled" or even downright bad because his alpha is non-existent.
Modeling global inter-lender syndicated loans as a network reveals that like other complex systems (like those found in nature) the inter-bank network displays small world properties. Translation? The failure of a random institution is unlikely to cause other than some limited local pain, but if certain institutions disappear the network suffers cascading failures, community dissolution, and financial crises. We can identify these institutions ex ante, that is, before something bad happens. Unlike current ad hoc methodologies used to identify important banks, network methodology identifies dynamically systemically important firms. That way, institutions are not always "too big to fail" and therefore insulated from market discipline. The full paper may be found here.