Renaissance Technologies
Founded by Jim Simons (d. 2024); the firm is led by co-CEOs Peter Brown (computational linguist, ex-IBM). The most secretive and most successful quant shop in history.
The archetypal systematic shop. Its legendary Medallion Fund is closed to outside money (since 2005, employees/family only) and famously opaque — Medallion's signals are NOT in the 13F. What IS visible is the long US-equity book underlying the external institutional vehicles, RIEF (Renaissance Institutional Equities Fund) and RIDA, which run a more capacity-tolerant, lower-turnover statistical strategy. The 13F surfaces real AI-trade exposure: a boosted NVDA stake, Micron (memory/HBM for AI), and a sizeable Palantir position — but read it as model output, not human conviction.
- Not open — Not an open-ended vehicle — analysis only, not an available allocation.
- Partial 13F — The disclosed book is longs-only and ~45 days lagged — a slice, not the strategy.
Top holdings
Largest reported 13F position Q1 2026; biotech, non-AI.
top-5 — AI/data-analytics; trimmed materially in Q1 2026 (profit-taking by the model).
~2.5M shares — Stake boosted ~189% QoQ in Q1 2026 — the clearest AI-trade signal in the visible book.
top-10 — Memory/HBM supplier to AI accelerators; position reduced in Q1 2026.
New initiation in Q1 2026 ($243-281 range); Mag-7 exposure.
Recent moves
Q1 2026 13F (filed May 14, 2026): boosted NVDA ~189% QoQ, new Apple stake, trimmed Palantir/Micron/SanDisk. Source: Seeking Alpha / Holdings Channel.
Loading disclosed positions…
Our take
The headline caveat: the 13F you can read is NOT the strategy that made Renaissance famous — Medallion is closed and secretive, and its signals never appear here. The visible RIEF/RIDA book is genuinely AI-exposed (NVDA boost, Micron, Palantir), but it is systematic model output, rebalanced quarterly, not a conviction bet — RenTec is the canonical case for the systematic-vs-discretionary caveat. Honest limit: 13F shows only long US-listed equity; no shorts, derivatives, or the leverage that defines Medallion.
The 13F is the most-misread document in this fund's public footprint.
The +189% QoQ NVDA increase filed May 2026 (871K → 2.53M shares, ~$441M), alongside a new 3.08M-share Apple stake, is model output on a quarterly rebalance across a very long tail — and it follows an ~85% NVDA cut the prior quarter, so it is round-tripping, not a conviction AI bet. Size the gap: ~$63.9B of long US equities against $96.4B regulatory AUM (Form ADV, 2025-12-31, read via a third-party mirror), against $165B including leverage as disclosed in 2021. Shorts, derivatives, non-US and the leverage that defines Medallion are all invisible. What is genuinely durable is organizational rather than signal-level: a single-team culture with unusually low key-person risk. Simons retired from management in 2009 and died in 2024; the firm kept compounding under Peter Brown. What is not durable is the assumption that the process transfers to size. Medallion's capital cap and forced annual profit distribution are the firm's admission, in structure rather than prose, that its alpha decays in dollars.
- 2026-07At 31 Mar 2026 the larger AI-adjacent marks were memory, not accelerators: MU $730.7M and SNDK $508.0M, both above Nvidia. July 2026 landed exactly there — MU -28.7% and SNDK -46.6% while NVDA was +0.3%.
- 2026-05On that same Q1 2026 13F ($63.93B, filed 14 May), Nvidia is the tenth-largest long at $440.7M, behind UTHR $1.06B, PLTR $1.02B, AAPL $780.6M, KGC $779.7M and MU $730.7M. A tail position, not a lead one.
Thesis
Renaissance is two firms under one brand. Medallion — began closing to outsiders in 1993, employee-only since 2005, capped near $10B, charging 5/44 — built the legend. RIEF/RIDA/RIDGE are the capacity-tolerant, long-biased vehicles outsiders could reach, and they are a different product. The instructive fact is not that Medallion wins; it is that the same team, data and codebase produced a reported +76% in Medallion and roughly -20% in RIEF in the same 2020. The edge is short-horizon and capacity-bound, and it does not survive a slower wrapper.
Systematic quant run as a research monoculture — one shared codebase, no PM silos — feeding two capacity regimes. Medallion: very-high-turnover, heavily-levered, short-horizon signal harvesting across futures/FX/equities. The external funds: lower-turnover statistical equity, long-biased by construction rather than market-neutral, aiming to beat the S&P 500 gross. The visible 13F (~$63.9B, ~3,213 positions, 2026-03-31) is external-vehicle residue, not Medallion.
Assessment
- Capacity discipline enforced structurally: Medallion is capped and distributes profits annually rather than compounding AUM into decay.
- Single research team on one shared codebase avoids the multi-PM silo problem, so signals compound across the firm instead of being hoarded per book.
- Low key-person dependence, demonstrated: performance persisted through Simons' 2009 management exit and his 2024 death.
- Separating the capacity-constrained strategy from the scalable one is intellectually correct — most shops instead dilute one fund until the edge is gone.
- The accessible product and the famous product are not the same strategy, and the brand halo transfers even though the returns do not.
- RIEF is long-biased against an S&P benchmark, so a large share of its return is beta — attribution to skill needs the beta stripped first.
- Medallion's ~39% net annualized figure is not verifiable from any primary filing; it circulates via books and press, and every downstream comparison inherits that unverified anchor.
- October 2025 is the clearest evidence of crowded-factor exposure: RIEF -14.39% and RIDA -15.6% in one month, alongside a broad quant unwind.
- The 2021 IRS settlement of up to $7B over Medallion's tax treatment is a governance and disclosure signal, not merely a historical footnote.
Record
Two records, never to be blended. External vehicles: RIEF +22.7% in 2024 (best since 2011) and RIDA +15.6% (best since its 2021 inception). Then October 2025 broke them — RIEF -14.39% in that single month, leaving it -7.5% YTD; RIDA -15.6%, off 10.3% YTD (Institutional Investor). That is the firm's worst setback since Simons' 2024 death, and it clustered with a broad quant-factor unwind, which reads as crowded exposure rather than proprietary edge. Trade press later put RIEF near +9.64% for full-year 2025 on a Nov–Dec recovery, still trailing the S&P 500's ~12% — treat that figure as unconfirmed. Medallion's record is unverifiable from primary sources (employee-only, no public filing), so ~39% net annualized is reported, not established. Its 2008 gain of about 98% against an S&P down 38.5% is the counter-cycle datapoint arguing the short-horizon edge is real, just unbuyable.
Risks & fit
- Signal decay: short-horizon statistical edges erode as more capital and better data reach the same patterns; Medallion's cap is the tell that this is already binding.
- Factor crowding in the external funds — the October 2025 drawdown clustered with peer quant shops, implying shared exposures rather than distinct alpha.
- Leverage: the gap between $96.4B regulatory AUM and the $165B leveraged figure disclosed in 2021 implies gross exposure, and gap-risk, well above net capital.
- Opacity means an outside observer cannot audit attribution, risk limits, or drawdown control on the inside fund at all.
- Structural conflict: the best capacity is reserved for employee capital, so outside investors are by design in the lower-alpha tier.
The core claim — that the edge is capacity-bound and dies in the scalable wrapper — would be wrong if RIEF delivered materially positive alpha to the S&P 500 over a full five-year cycle net of fees once equity beta is regressed out, with drawdowns uncorrelated to the peer quant-factor cohort. One strong year like 2024 does not do it; 2020 and October 2025 both cut the other way. Evidence that Medallion's own net returns compressed materially post-2020 would falsify the separate claim that the short-horizon edge is intact — but absent primary disclosure, that test is unrunnable from outside.
Analytically useful as the canonical case in three things: why 13F reading fails for a systematic shop, why capacity is the binding limit on quant alpha, and why a closed fund's brand should never be credited to its open siblings. Medallion began closing to outsiders in 1993 and has been employee-only since 2005 — not an allocation question for any reader. The external vehicles disclose little publicly, so any evaluation of them works from press-reported rather than audited numbers.
Medallion charges 5% management and 44% performance — terms rational only because capacity is capped and the capital is largely employees' own. Fee terms for RIEF/RIDA/RIDGE are not reliably disclosed, so net-of-fee comparisons rest on reported returns.