Research note · No. 01

How many memecoins make an index?

Choosing N and the rotation frequency for $EVERYTHING
THE OWNERS — 14 July 2026 · data: CoinGecko, GeckoTerminal

The questionWhy 500 is the wrong number here

The S&P 500 holds 500 names because that is roughly where adding another stock stops reducing your risk in a way you can feel. Below about thirty you are still exposed to single blow-ups; past a couple hundred the curve is flat and you are just holding the market. 500 is a round number safely inside the plateau.

Memecoins are a different asset — far more volatile, far more correlated with each other, and living in a much smaller liquid universe. So the number is not 500. This note answers, with data rather than vibes: what is the memecoin equivalent of 500, and how often should the basket rotate?

The dataTwo sets, two jobs

Stated up front. The Solana set is today's survivors — names that died are not in it, so every absolute return is biased upward, and the 200-day window is a memecoin downtrend. We do not publish absolute returns as a pitch. We publish the shape of risk versus N, which is regime-independent — and a downtrend is the honest place to measure diversification, because that is when concentration hurts. Past behaviour is not future behaviour.

Finding 01The diversification curve

The average single memecoin runs at 203% annualized volatility. Pool equal-weight baskets of size N and measure how much of that risk survives:

0% 25% 50% 75% 100% 1 5 10 15 20 NUMBER OF CONSTITUENTS (N) N=10 · 78% of variance removed N=2 · 59%
FIG.1 — variance removed vs. constituent count. The first coin removes 59%. The knee sits at N≈10 (78%). Ten to twenty-two buys only four more points.
Equal-weight baskets, averaged over rotations · Solana · 200 days
Nportfolio vol% of one coinvariance removed
1203%100%0%
2130%64%59%
5107%52%72%
898%48%77%
1095%47%78%
1590%44%81%
2286%42%82%

The first coin you add removes 59% of the variance. By five names you have 72%; by ten, 78%. Ten to twenty-two buys four more points. Per-coin: 8→10 cuts volatility by 3%, 10→15 by roughly 1% each. After eight to ten names, each new constituent does almost nothing for your risk.

This is the same curve equity researchers have drawn since Evans & Archer (1968) — most diversifiable risk gone by a couple dozen names, then flat. Memecoins hit the plateau earlier than stocks, for two structural reasons: they share one beta (meme risk-on / risk-off, so they move together and pooling removes less independent risk), and the liquid universe is small (reaching for name #15 or #20 means reaching down into thinner books, and on a taxed on-chain index every thin name costs slippage to hold and rotate). The memecoin "500" is not 500. It is about 10.

Finding 02The product's own risk curve agrees

The curve above is equal-weight theory. Running the actual index rules — top-N by trailing volume, capped at 25% per coin, weekly rebalance — over the same window:

Index rules (vol-weight, 25% cap, weekly) · read the differences, not the levels
Nannualized volSharpemax drawdown
1490%−8.1−100% wiped out
2315%−7.9−100%
3113%−4.5−96%
573%−2.1−67%
1073%−2.1−66%
2071%−2.0−64%

Sharpe is negative because the window is a downtrend — that is the regime, not the method. The story is the same, sharper: N=1 is ruin (the single-coin index lost everything when its one name collapsed). Volatility falls off a cliff to N=5, then is flat from 5 to 20. Two more points of drawdown protection for doubling the basket from 10 to 20 is not a trade worth its cost.

Finding 03The eligible universe is under 10 today

Screening the live Robinhood pools (liquidity ≥ $150k, 24h volume ≥ $500k, excluding stablecoin pairs), the memecoins that qualify right now are roughly four to six names — CASHCAT, VIRTUAL, Elves, TRASH, with a couple more on the edge. The chain is two weeks old; the universe grows weekly.

10 is a forward cap, not a fill requirement. Today the index holds every screened name up to ten — which is all of them. As the chain matures and more names clear the liquidity bar, the index grows into the ten-slot ceiling that the Solana math identifies as the plateau. We do not pad the basket with junk to hit a number; we hold the eligible set and cap it where diversification stops paying.

Finding 04How often to rotate

Rebalance frequency · size-10 index · 200 days
rebalance everySharpemax drawdownturnover
1 day−2.31−69%5.1×
3 days−2.27−68%3.9×
7 days−2.13−66%3.0×
14 days−2.17−67%2.4×
30 days−1.99−59%1.5×

Rebalancing daily is the worst outcome on every axis — it churns the book (5.1× turnover), and each rotation on a taxed token pays slippage and tax, so the trading is a pure drag. But you cannot simply pick "monthly," because memecoin metas move in days, not months — a new leader can 10× or a holding can rug inside a week. The answer is a three-part rule that buys responsiveness without the churn:

The decisionN = 10, rotate on three clocks

N = 10. It sits on the knee of the diversification curve — about 78% of the achievable variance reduction, where going further pays basis points of risk for real liquidity and rotation cost. It is well clear of the danger zone (N=1–3, where single-name blow-ups dominate) and it matches the size of the liquid memecoin universe instead of forcing the basket into thin names.

Rotation: daily inflow-only, weekly hard rebalance on a 15-point drift trigger, and immediate event-driven ejection. Free continuous nudging, disciplined weekly realignment, no waiting on the calendar when a name breaks.

We will revisit both. The right N is a function of how many liquid, uncorrelated names exist to hold. As Robinhood Chain matures — more names, deeper books, and eventually enough history to run this study on Robinhood's own data instead of borrowing Solana's shape — the number may move. If the liquid universe grows and correlations fall, 10 could become 12 or 15. This is the answer as of July 2026, and it is dated on purpose.

DATA — CoinGecko (Solana daily price & volume, 22 names × 200 days) · GeckoTerminal (Robinhood Chain live pools).
METHOD & RAW FILES — 5_CHAIN/research/: sol_universe.json, hist/, n_opt.json, divcurve.json, rh_screen.json.
This is research, not investment advice. Memecoins can go to zero. See risk & disclosures.