Renaissance Technologies' portfolio: Q3 2026 holdings and October returns

Executive summary

Renaissance Technologies' institutional equity fund reportedly gained 6% in October 2026 while MSCI World fell 0.8%; the adviser's 13F for 30 September, filed on 12 November, shows $73.4bn of US-listed holdings, with Nvidia, Meta, Amazon, Apple and Broadcom cut to almost nothing during the quarter. The October figures come from Investing.com on 10 November 2026, citing Bloomberg; Renaissance does not publish them.

SEC filings also show the cost of moving differently from equities. A registered fund's stake in the Renaissance Institutional Equities Fund (RIEF) fell 19.9% in reported value from May to October 2025 while MSCI World net rose 20.9% in US dollars. That is an estimated investment return only if no cash flows affected the stake.

Copio's assessment: the available observations show that RIEF can move differently from world stocks, but do not establish its correlation or contribution to long-run growth. Its brochure describes a strategy exposed to model failure and leverage as well as market falls. A lower beta target offers no guaranteed floor under an investor's capital.

What is in the fund?

Renaissance runs three fund families from one adviser, according to its Form ADV Part 2A brochure dated 30 March 2026. Medallion trades a short-term strategy across asset classes. RIEF trades US-listed equities with a net-long position, and the Institutional Diversified Alpha funds (RIDA) trade global equities. Regulatory assets under management were $96.4bn at 31 December 2025.

Source: Form ADV Part 1, Schedule D, section 7.B.(1), amendment of 1 July 2026; the last column is the share owned by Renaissance and its related persons. Gross asset value includes positions financed with borrowing, so it overstates investors' capital in a leveraged fund.

Outside investors hold most of RIEF, while Medallion's capital is overwhelmingly Renaissance's own and its related persons'. The form counts 705 beneficial owners in RIEF, 193 in RIDA and 549 in Medallion. The brochure lists $5m starting minimums for outside investors in RIEF and RIDA; Medallion's $25,000 minimum applies to employee and employee-related investors.

The 13F filed on 12 November 2026 lists 3,151 positions worth $73.42bn at 30 September, against $72.62bn at 30 June. It aggregates Renaissance's reportable holdings, so it cannot be read as RIEF's portfolio. It contains only long positions in US-listed shares and ETFs: no options appear, and the filing format excludes short sales, futures, swaps and borrowing.

The reported long book contains many small positions. Grouping share classes by issuer, the largest position, Micron Technology, was 2.21% of reported value; the ten largest issuers were 12.72% and it took 136 issuers to reach half. The next eight were Alphabet, Advanced Micro Devices, United Therapeutics, Intel, UnitedHealth, Exelixis, VeriSign and SpaceX, a new position of 4.1 million class A shares worth $620.5m.

30 June to 30 September 2026, share of reported 13F value: Nvidia 1.95% to 0.01%, Meta 1.57% to 0.002%, Micron 0.34% to 2.21%, AMD 0.64% to 1.64%, UnitedHealth 0.21% to 1.11%. Source: Renaissance 13F information table for 30 September 2026 and for 30 June 2026; positions and weights (CSV).

Nvidia went from 7,091,256 shares to 23,170 and Meta from 2,023,153 to 1,577. Nvidia, Microsoft, Apple, Amazon, Alphabet, Meta, Broadcom and Tesla together fell from 7.90% of reported value to 2.22%, and almost all of what remained was Alphabet and Tesla. Micron's share count rose more than sevenfold and UnitedHealth's sixfold. Across the whole book, 591 positions worth $6.76bn were new and 580 worth $4.03bn at 30 June had gone.

None of this says how RIEF was positioned on 1 October. A long position in the 13F may sit alongside a larger short in the same stock, or be offset by index futures, and the strategy trades continually, so a quarter-end snapshot cannot show net exposure in October. The filing is evidence of what the long book held on one day, which is all we use it for.

Why does Renaissance use this structure?

The brochure describes models that "use publicly available financial data to identify and implement trading decisions" (page 11). RIEF's stated objective is "a long-term risk-adjusted return that, on a gross basis, exceeds that of the Funds' benchmark index (the Standard & Poor's 500 Index with dividends reinvested) while attempting to maintain a relatively low beta to the Index of 0.4 or lower and lower volatility than the Index."

The structure behind that objective is leverage. RIEF is "expected to have a long-term average leverage of approximately 2.5 to 1" and is designed to "be no more than $100 net long for each $100 of equity" (same page). Short sales and derivatives are intended to reduce the long book's market sensitivity. Leverage magnifies exposure; it does not itself keep beta low. The brochure's targets describe the intended structure, rather than measured current exposure.

RIDA aims at "maximizing long-term return while attempting to meet a standard deviation target" (page 11), which the brochure puts between 8% and 16% a year, and expects "most returns over the long term to come as alpha as the Funds are intended to have low correlation to the U.S. and global equity markets" (page 19). Both families may use stock index futures or options "for risk reduction purposes".

The brochure records failures of those controls. RIEF's ability to keep beta at 0.4 or lower "depends on the accuracy of the Funds' beta models, which, in recent volatile markets, have not performed as expected", and "in recent volatile markets, the RIEF Funds' leverage has exceeded 3 to 1" (page 18). On hedges it warns of "a substantial risk, however, that such hedges will not be effective in limiting losses" (pages 15-16).

The annual management fees and performance allocations differ by fund (pages 5-6). RIEF charges 0.20% to 1.50% of net asset value a year plus 0% to 10% of net gains; RIDA charges 0.85% to 1.00% plus 10%; Medallion charges 4% plus 36% to 44% of net profits. Renaissance may "suspend or delay redemptions" in extraordinary circumstances (page 17).

Historical performance versus MSCI World

No primary document we could find publishes a monthly or annual return series for RIEF or RIDA, so their geometric return, Sharpe ratio and Sortino ratio against MSCI World cannot be calculated. We can compare two six-month changes in a registered fund's reported stake value, subject to cash-flow assumptions, alongside press-reported figures for 2026.

The filings come from the Macquarie Focused Access Fund, formerly CPG Focused Access Fund, a registered fund of hedge funds that has held RIEF since 1 November 2018. It values the stake at RIEF's net asset value, "net of management and incentive fees", and reports its cost and fair value twice a year. Unchanged cost is consistent with holding the same investment, but cannot rule out distributions or offsetting transactions. We treat the value changes as return estimates conditional on no such cash flows.

RIEF filing estimates, assuming no cash flows and net of its fees, against MSCI World net USD: November 2022-April 2023 +11.4% and +12.3%; May-October 2025 -19.9% and +20.9%; October 2026 +6.0% (press-reported) and -0.8%; January-October 2026 +12.3% (press-reported) and +10.9%. Source: Macquarie Focused Access Fund schedules of investments at 31 October 2022, 30 April 2023, 30 April 2025 and 31 October 2025; MSCI World net USD (index 990100); FRED TB3MS; windows and inputs (CSV).

In the first window RIEF roughly kept pace with world stocks. In the second it went the other way: the stake was worth $112,963,785 at 30 April 2025 and $90,500,264 at 31 October, with the cost unchanged at $67,888,474. Its N-PORT for 31 July 2025 gives conditional estimates of -7.0% in May-July and -13.9% in August-October, against MSCI World's gains of 11.9% and 8.0%.

A 19.9% loss needs a 24.8% gain to recover. Multiplying the two conditional filing-window estimates turns $100 into $89.21 in RIEF, against $135.72 in MSCI World net and $104.40 in modelled Treasury-bill accrual. These separated windows omit the intervening returns, so the calculation describes neither a continuous investment path nor full-cycle performance.

October 2026 reportedly favoured RIEF. Investing.com reported RIEF up 6% for the month and 12.3% for the year to date, and RIDA up 4.6% and 22.4%, citing Bloomberg's sources; Hedgeweek added that many quantitative funds lost about 1% and Kirkoswald Asset Management "recorded its worst monthly loss on record". These are press figures, not fund documents.

Calendar October differs from the press article's trailing month. Over calendar October, MSCI World net fell 0.8%, MSCI USA net 1.0% and the Nasdaq-100 price index 6.7%, by our calculation from MSCI and Nasdaq closes. The 10.3% Nasdaq-100 fall in the Investing.com article is a trailing one-month figure into November; from its 6 October close to 5 November the index fell 14.7%.

Over January to October, the reported 12.3% for RIEF compares with 10.9% for MSCI World net and 3.1% for three-month Treasury bills. The same press figures imply RIEF was up about 5.9% at the end of September, against 11.8% for MSCI World, so one month turned a lag of almost six points into a lead of 1.4.

Copio's reading

Is it optimal for geometric growth?

The available evidence does not establish whether RIEF improves long-run compounded growth. The objective is expected log wealth over decades, subject to the same withdrawal needs and borrowing limits. The relevant comparison is world equities and cash, alone and combined with RIEF, after all costs. The filings identify three questions that a full return record would need to resolve.

• Is the risk paid for? RIEF targets a beta of 0.4 or lower while expecting average leverage of about 2.5 to 1. Its stake's estimated 19.9% fall during a rising equity market shows that low market sensitivity can coexist with substantial losses. Whether expected stock-selection gains compensate for that risk needs a full cycle of net returns.
• Does combining it with equities improve growth? That depends on its expected net return, volatility and joint losses with the rest of the portfolio. At RIDA's separate volatility target of 8% to 16%, the small-return approximation of half the variance implies a drag of about 0.3 to 1.3 percentage points relative to the arithmetic mean. That approximation is not a measured fund result.
• Do fees, financing and liquidity change the answer? RIEF charges up to 1.5% a year plus 10% of gains, before a fund of funds' additional fees. Borrowing costs and suspended redemptions can further impair an investor's outcome. The brochure records leverage above 3 to 1 in volatile markets, increasing the capital exposed to adverse position moves.

Is it crash aware and protected?

The strategy targets lower market sensitivity, with no promised capital floor. In a linear scenario with beta fixed at 0.4, a 30% equity fall implies about 12% loss through market exposure alone, before stock-selection gains or losses. A beta target is not a reliable crash-loss forecast. The brochure itself says the beta models have not performed as expected in recent volatile markets and that the funds "may be exposed to losses in times of general market downturns".

The 2025 valuation fall illustrates risk outside a broad market decline. RIEF's stake fell 19.9% while equities rose sharply, so its losses need not coincide with market falls. Index futures and options are permitted for risk reduction, but the brochure does not disclose a guaranteed portfolio floor or a fixed crash payout for investors.

Is it diversified without sacrificing growth?

The public observations suggest different exposures, but cannot establish reliable diversification. RIEF reportedly gained while world stocks fell in October 2026; its registered investor's stake lost value while they rose in mid-2025. Opposite returns in selected windows do not measure correlation, and the overlapping 2026 figures are not separate observations.

We cannot show a correlation matrix, because the four windows above are not a return series. The adviser-wide long book spans 2,931 issuer groups, with the largest at 2.21%; that does not establish RIEF's own diversification. Concentration risk sits in the models and in leverage, and RIEF's documents say positions can be "difficult to exit quickly without affecting their prices".

Copio does not grade Renaissance in its portfolio checker, because a leveraged long-short book of single stocks cannot be reproduced faithfully with UCITS ETFs. Outside investors face minimums starting at $5m, subject to exceptions, and copying the 13F would omit the shorts, futures and borrowing that determine the actual funds' exposures.

For a holding intended to diversify equities, compare its losses, fees and return contribution over rising and falling markets, using the same dates. These windows provide examples to investigate, rather than a sufficient test. A household's spending needs also matter: accepting lower growth can be reasonable when it reduces the risk of a forced sale.

Our studies of Cliff Asness and AQR's diversification and Mark Spitznagel's tail hedging examine two other ways professional investors try to get returns that do not fall with stocks. Which assets are uncorrelated with stocks gives twenty years of euro data on the options a household can actually hold, and Sharpe ratio versus CAGR explains why a smoother ride does not by itself mean faster compounding.