Savana US Small Caps Active ETF (ASX:SVNP) fell 7.65% in September, trailing the S&P SmallCap 600 by 4.92 percentage points as higher US bond yields and a concentrated AI rally split the market. The manager says its valuation discipline kept it away from the month’s strongest semiconductor performers, while highlighting a 52% gain from its Everforth investment.
- SVNP fell 7.65% in September versus a 2.73% decline for the S&P SmallCap 600
- Only five of the ETF’s 42 holdings finished higher during the month
- US 10-year Treasury yield rose 54 basis points to 5.29%, its highest level since 2007
- AI semiconductor stocks rose sharply while most small-cap shares weakened
- Everforth position gained about 52% between May entry and September exit
September loss widens SVNP’s benchmark gap
The Savana US Small Caps Active ETF (ASX:SVNP) lost 7.65% in September, turning a difficult month for US small caps into a sharper setback for the fund. The S&P SmallCap 600 fell 2.73% in Australian-dollar terms, leaving SVNP 4.92 percentage points behind its benchmark.
The weakness was broad across the portfolio: only five of 42 holdings finished higher. Savana said the S&P 600 declined on 13 of 21 trading days as higher US bond yields and narrow market leadership pressured much of the small-cap universe. SVNP’s year-to-date return remained positive at 6.15%, but that trailed the index’s 9.35% gain.
The result contrasts with the fund’s July outperformance, when SVNP gained 2.21% and beat the benchmark by 5.44 percentage points during a semiconductor sell-off. September showed the other side of that exposure: the manager’s limited participation in AI-related semiconductor names became a material drag on relative performance.
Bond yields and AI stocks create a two-speed market
The US 10-year Treasury yield climbed 54 basis points during September to 5.29%, its highest level since 2007. Savana said the trailing one-month correlation between daily S&P 600 returns and movements in the 10-year yield fell to negative 0.64, compared with a 10-year average of positive 0.11.
At the same time, about 90% of semiconductor stocks rose, compared with just 23% of companies across the broader S&P 600. The contrast was especially visible in MaxLinear, which Savana said rose approximately 400% year to date and gained about 53% in September alone. The stock’s monthly return had a correlation of negative 0.96 with SVNP’s excess return over the benchmark between June and September.
Savana argues that the AI rally has become powerful enough to distort benchmark comparisons, even though the relevant semiconductor names represent only a relatively small part of the index. Its June benchmark shortfall also came during a broad small-cap rally led by valuation expansion, illustrating how quickly relative performance can turn when leadership narrows or widens.
MaxLinear illustrates the valuation risk
The manager’s concern is not that the AI investment cycle is necessarily a bubble. Rather, it says prices for some small-cap beneficiaries have moved well ahead of fundamentals. Savana cited MaxLinear’s approximately 400% share-price gain against revenue growth of about 22% and gross-profit growth of about 23% since January.
Over the same period, MaxLinear’s price-to-sales multiple expanded from approximately 2.1 times to 14.3 times, while last-twelve-month earnings per share remained negative. Savana said those valuations can continue to work if growth, margins and AI investment remain strong, but leave the shares more sensitive to disappointment.
Everforth trade delivers a 52% gain
The letter offered Everforth Inc. (NYSE:EFOR) as the counterexample to thematic momentum. Its share price fell from about US$125 in late 2021 to US$17 by July 2026, and dropped 63% between January and June even as last-twelve-month revenue declined only 1% and normalised earnings per share fell 6%.
SVNP entered the position on 8 May at approximately US$20.32, when Everforth was trading at roughly 5.8 times earnings, or a 17% earnings yield. After the company announced a US$115 million, three-year US Army contract and reported second-quarter adjusted earnings per share of US$0.91, up from US$0.69 in the first quarter, the fund exited on 8 September at about US$30.92. The position appreciated roughly 52% in four months.
Savana says the trade was not based on predicting the contract or earnings recovery. Its stated approach is to find situations where negative expectations are already heavily reflected in the price and the potential reward from stabilisation outweighs the downside if deterioration continues.
Valuation discipline faces its next test
SVNP’s since-inception return stood at 10.76% annually after fees and costs, compared with 3.55% for the S&P SmallCap 600, according to the letter. September nevertheless underlined the cost of avoiding fast-rising themes: a valuation-driven portfolio can lag materially while momentum remains concentrated in a handful of stocks.
The immediate question is whether fundamentals begin to catch up with AI-related valuations, or whether the narrow rally continues to dominate benchmark returns. SVNP’s future relative performance will remain tied not only to its stock selections, but also to whether market leadership broadens beyond the semiconductor names that shaped September.
Bottom Line?
September exposed the short-term price of SVNP’s valuation discipline; the next test is whether earnings breadth can replace narrow AI momentum as the market’s main driver.
Questions in the middle?
- Will higher US bond yields continue to pressure the wider small-cap universe?
- Can AI semiconductor valuations keep expanding if revenue and earnings growth do not accelerate?
- Will SVNP’s oversold-value strategy produce more recoveries like Everforth as market leadership broadens?