Reading Fund Styles from Public Disclosures

What quarterly long-only holdings filings can and cannot reveal about hedge-fund styles, and how SurgeFlow's Whales boards read consensus, conviction, and position deltas across four markets.

Category: Whales

What this is and why it matters

Hedge funds rarely explain how they invest. Regulators, however, force fragments of the picture into the open. In the United States, institutional managers above a size threshold must file Form 13F each quarter, listing their long positions in US-listed equities. Japan, Hong Kong, and China operate their own shareholder-disclosure regimes. Each filing is a snapshot of one side of one book at one moment in time.

A fund's "style" is the repeatable way it tries to make money, and it matters to a reader of disclosures because the same holding carries different information depending on who holds it and why. A merger arbitrageur's stake in a takeover target is a view on deal completion; a deep-value manager's stake in the same name is a view on the underlying business. Reading style from public holdings is therefore an act of inference, not observation. This article covers the classic style taxonomy, the arithmetic of holdings-based inference, and — just as important — what the data cannot tell you.

The classic hedge-fund style taxonomy

Most hedge-fund strategies fall into four broad families. The labels are coarse, and real funds blend them, but they remain the standard map. - Long/short equity — buy stocks expected to outperform, sell short stocks expected to lag. Net market exposure is a dial rather than a constant, ranging from strongly net-long to market-neutral. - Event-driven — positions built around corporate events: announced mergers, spin-offs, restructurings, distressed debt. Returns hinge on whether the event completes, not on the market's direction. - Relative value — paired positions exploiting pricing gaps between related instruments: convertible bonds against the issuer's stock, dual-listed share classes, an index against its constituents. Each leg hedges the other. - Macro and managed futures — top-down positions in rates, currencies, commodities, and equity indices, usually expressed through futures and forwards; systematic trend-following is the managed-futures branch.

The mechanics: what a long-only snapshot can and cannot reveal

Quarterly long-only filings support a small set of honest computations. Write V_i for the market value of holding i. The position weight is w_i = V_i / sum_j V_j — the share of the disclosed book in one name, the basic proxy for conviction. Concentration is summarized by the Herfindahl index HHI = sum_i (w_i^2), and the effective number of positions is N_eff = 1 / HHI: a book of a dozen large weights reads very differently from a book of two thousand slivers. The position delta is dq_i = q_i,t − q_i,t−1, the split-adjusted change in shares held between consecutive filings — accumulation or reduction. Across many filers, breadth is B = H / F, where H is the number of tracked filers holding the name and F is the total number of tracked filers.

What the snapshot cannot reveal is just as concrete. Short positions never appear, so net exposure NE = (L − S) / E — longs minus shorts over equity — is not computable from filings; you observe L and nothing else. Most derivatives, swap exposure, futures, and non-US listings are absent from 13F, which is why macro and managed-futures funds leave almost no equity footprint. A relative-value pair shows only its long leg and masquerades as a one-sided bet. And US filings can arrive up to 45 days after quarter end: the position you are reading may already be gone. Staleness is not a data defect to be cleaned away; it is a property of the instrument, and honest use of the data models it rather than ignores it.

How SurgeFlow applies it

SurgeFlow's Whales boards at /whales aggregate institutional disclosures across markets into three holdings-only reads. Consensus counts how many tracked filers hold a name — the breadth B above, developed further in the institutional-ownership consensus score. Conviction weighs a position against the size of the filer's disclosed book and its own history — the w_i and concentration lens. Position delta tracks quarter-over-quarter accumulation and reduction — the dq_i lens. Together these approximate style tilts from holdings alone: a concentrated, slow-turnover book reads as fundamental and long-horizon; a broad, fast-churning book reads as systematic; clustered stakes in announced deal targets hint at event-driven exposure.

Equally important is what the site does not do. SurgeFlow does not classify funds into styles with a labeled model — public data contains no ground truth to train on. It does not reconstruct short books, leverage, or derivative overlays, because they are not disclosed. And it does not treat any of these reads as a forecast: disclosure data is stale by construction, and on SurgeFlow the signal_date of every whale signal equals its disclosure date — never backfilled to the position date — so nothing in the boards assumes information before the market could have had it. All of it is research material, not investment advice.

What SurgeFlow data this uses

Whales coverage spans SurgeFlow's four markets: the US, China, Japan and Hong Kong. Disclosure feeds differ by regime: SEC 13F quarterly snapshots for the US; EDINET large-shareholding reports for Japan; HKEX disclosure-of-interests notices for Hong Kong; CSRC-mandated shareholder disclosures for China. These regimes are not interchangeable — 13F is a periodic portfolio snapshot, while threshold-crossing regimes report only when an ownership line is crossed — so cross-market comparability is limited, and the boards key everything by disclosure date. Reference and price data come from the site's standing providers: FMP for the US; Tushare for China and Hong Kong; and J-Quants plus EDINET for Japan.

Explore it yourself

Everything described above is queryable. A free beta API key is available at /account, good for 2,000 requests per day, and the full endpoint catalogue is published as an OpenAPI specification at /api/public/openapi.json. Pull the whale boards, join the consensus and conviction reads against your own universe, and test whether holdings-based style inference holds up out of sample. The data is offered as research material for your own analysis, not as a recommendation of any security.

Sources and further reading

Primary, academic, and SurgeFlow data references for the claims above: - Sharpe, W. F. (1992), "Asset Allocation: Management Style and Performance Measurement", Journal of Portfolio Management. - Fung, W., and Hsieh, D. A. (1997), "Empirical Characteristics of Dynamic Trading Strategies: The Case of Hedge Funds", Review of Financial Studies. - Mitchell, M., and Pulvino, T. (2001), "Characteristics of Risk and Return in Risk Arbitrage", Journal of Finance. - Fung, W., and Hsieh, D. A. (2004), "Hedge Fund Benchmarks: A Risk-Based Approach", Financial Analysts Journal. - Brunnermeier, M. K., and Nagel, S. (2004), "Hedge Funds and the Technology Bubble", Journal of Finance. - Griffin, J. M., and Xu, J. (2009), "How Smart Are the Smart Guys? A Unique View from Hedge Fund Stock Holdings", Review of Financial Studies. - Cremers, K. J. M., and Petajisto, A. (2009), "How Active Is Your Fund Manager? A New Measure That Predicts Performance", Review of Financial Studies. - Agarwal, V., Jiang, W., Tang, Y., and Yang, B. (2013), "Uncovering Hedge Fund Skill from the Portfolio Holdings They Hide", Journal of Finance. - SurgeFlow routes: /whales (institutional disclosure boards), /research/institutional-ownership-consensus-score (consensus methodology), /account (API keys), /api/public/openapi.json (API specification).