Voss Capital's 22.4% quarter beat the Russell 2000 by 0.9 points on a book 100.6% net long

The Value Fund's 22.4% was 0.9 points above the Russell 2000's 21.5% ^[1][3][16]. The offshore feeder's 23.2% was 1.7 points above it ^[2][17]. The onshore fund cleared the Russell 2000 Value (+17.2%) by 5.2 points and the S&P 500 (+15.2%) by 7.2 ^[4][18][19]. The comparison that fits a net-long small-cap book is the first one, and it is the narrowest.

On June 30 the Voss Value Master Fund was 178.7% gross and 100.6% delta-adjusted net long ^[6][5]. If gross and net are on the same basis, that implies a long book near 139.7% and a short book near 39.1% ^[20][21]. The net-to-gross ratio of 56.3% ^[22] is a long way from market-neutral. A book that net long takes small-cap moves nearly one-for-one. So we think most of the 22.4% is the index's 21.5%, and the 0.9 points is what stock selection and the shorts added. Or rather, the more defensible version: that holds if the book's beta is near one.

The counter-thesis is concentration. The ten largest longs weigh 73.9%, an average of 7.4% each, against 3.5% each for the ten largest shorts ^[7][23][24], a ratio of 2.1 to 1 ^[25]. A long book built that way can behave very differently from the index in either direction. The excerpt of the letter does not report the fund's beta or list its positions.

The letter's own diagnosis is a market split between AI-infrastructure winners and everything else ^[9], traded in what it calls a "flow-amplified, non-fundamental tug of war between AI inevitability and existential doubt" ^[10]. Its example: when a semiconductor titan prints a beat-and-raise, software and the "quality" factor may reflexively gap down 3% as hot money grosses up momentum and de-grosses elsewhere ^[11]. Whether the fund sits on the receiving end of that 3% depends on which names are in its longs and shorts.

The letter's revision data cut against a clean two-bucket market. Net EPS revision breadth for small and mid caps troughed at -28% in 2023 and finished the Q2 earnings season at +10% ^[12], a 38-point swing ^[26]. Consensus earnings for large caps in every GICS sector have been revised higher since May ^[13]. The same letter says the five hyperscalers account for roughly half of Russell 3000 next-twelve-month capex ^[14]. It also says traditional machinery remains in contraction, while residential housing, non-data center commercial construction and auto production sit near the bottom of their post-2019 ranges ^[15].

The master fund is small beside the firm: roughly $260.7 million of about $2.00 billion in firm assets, or 13.0% ^[8][27]. We would change our view if the rest of the letter shows the short book made money in the quarter, or that the ten largest longs are low-beta names. Either would put more of the 0.9 points down to selection.