DR-DDEI-HIST-0.1 · Mechanism Laboratory

DDEI HISTORICAL RECONSTRUCTION

Had these rules been applied to the available historical data, what would have happened? Research backtest. Not a live index. Not an investment forecast. Votes do not change history.

RESEARCH · SIMULATION · NOT INVESTMENT PERFORMANCE · NOT INVESTABLE · £0 REAL CAPITAL · PAPER

POINT-IN-TIME RECONSTRUCTION · NO LIVE DDEI · MODEL NOT FROZEN · DR_MODE=PAPER · REAL CAPITAL £0 · TOKEN NOT ISSUED · TRADING DISABLED · COMMUNITY VOTES IGNORED

What this is

Under this research methodology, reconstitution uses the 12 point-in-time ordinal snapshots (dead names included while they still existed). Weekly prices are Yahoo/Binance reconstructions (class B). Size is an ordinal/Zipf proxy (class C), not vendor free-float. CoinGecko public history stops at 365 days and was not used as the panel.

Earliest defensible return date: 2017-12-31. Latest: 2026-08-31.

2017-12-31 is the first research snapshot where a diversified (non-BTC-only) Yahoo weekly price panel exists (ETH-era). 2013–2016 membership research exists, but price coverage is too thin for an honest multi-asset total-return path. CoinGecko public history is limited to 365 days and was not used.

Time machine

Historical reconstruction under this research methodology. Not an investment forecast. Not a live DDEI.

Quarter

Reconstitution date 2025-12-31

2025-12-31

Literal names: 27
N_effective: 19.5
Largest: BTC 10.0%
Top-5: 39.8%

Input mix this snapshot — reconstructed prices (B): 68% · size proxy (C): 100% · missing prices (D): 28%.

Asset	Research weight	Change
BTC HOLD	10.00%	0.00%
ETH HOLD	10.00%	0.00%
XRP HOLD	7.64%	0.19%
BNB GROW	6.48%	0.92%
SOL SHRINK	5.70%	-0.62%
DOGE HOLD	4.69%	0.12%
ADA HOLD	4.35%	0.11%
TRX HOLD	4.06%	0.10%
AVAX HOLD	3.82%	0.10%
LINK HOLD	3.62%	0.41%
SHIB HOLD	3.03%	-0.33%
XLM GROW	2.92%	0.81%
BCH HOLD	2.82%	-0.03%
HBAR GROW	2.73%	0.67%
LTC HOLD	2.64%	-0.01%
DOT HOLD	2.57%	-0.39%
LEO HOLD	2.50%	-0.25%
NEAR HOLD	2.37%	-0.21%
AAVE ENTER	2.31%	2.31%
ICP HOLD	2.12%	-0.19%
RENDER HOLD	2.07%	-0.13%
FET HOLD	2.03%	-0.12%
OKB HOLD	2.00%	0.05%
CRO HOLD	1.96%	0.05%
OP HOLD	1.89%	0.05%
ENA HOLD	1.86%	0.05%
ONDO HOLD	1.83%	0.07%
ATOM EXIT	0.00%	-2.02%
INJ EXIT	0.00%	-1.73%

Normalized path

HISTORICAL RESEARCH · NOT AN INVESTMENT FORECAST.

WHAT HAPPENED TO DDEI? · Terra

Evidence clock T0 public: 2022-05-07. LUNA weight before collapse — raw 2.3% · simple cap 5.1% · sqrt+10 4.0%.

LUNA contribution: raw -2.3% · cap -5.1% · sqrt -4.0%. Index move: cap -26.8% · sqrt -26.2% · BTC -21.2%.

WHAT HAPPENED TO DDEI? · FTX

Evidence clock: 2022-11-02. FTT weight — raw 0.6% · cap 1.5% · sqrt+10 2.1%.

FTT contribution: raw -0.6% · cap -1.4% · sqrt -2.0%. Index: cap -25.7% · sqrt -26.1%.

WHAT HAPPENED TO DDEI? · COVID

Systemic crash window. Sqrt -50.9% · simple cap -51.0% · BTC -40.3% · ETH -54.1%. Diversification did not remove market-wide crypto risk.

£1,000 from common inception 2017-12-31

Illustrative reconstruction of £1,000 left in each series. Not advice.

Series	End	Peak	Low	CAGR	Max DD
BTC	£4785	£7496	£197	19.8%	-80.3%
ETH	£2143	£4145	£74	9.2%	-93.8%
60/40 BTC/ETH	£4495	£7451	£141	18.9%	-85.9%
Raw size proxy	£4217	£7159	£132	18.1%	-86.8%
Simple 10% cap	£2441	£4933	£83	10.9%	-91.7%
Sqrt+10%	£2907	£6468	£88	13.1%	-91.2%

Simple index challenge

DOES SQRT+10 BEAT SIMPLE 10% CAP: INSUFFICIENT EVIDENCE

Sqrt raises N_effective, but the size basis is a Zipf proxy and holdout confirmation is too weak to freeze a complex weighting family.

Mean N_effective — simple 14.8 · sqrt 17.9
Max single-name — simple 10.0% · sqrt 10.0%
Mean one-way turnover at reconstitutions — simple 25.3% · sqrt 28.2%
Emergency overlay vs quarterly-only: NEUTRAL

Latest research snapshot (not live DDEI)

2025-12-31 · 27 names · N_eff 19.5 · largest BTC 10.0% · top-5 39.8%

Freeze decision (not automatic)

SQRT+10 historically supported: INSUFFICIENT EVIDENCE
Beats simple 10% cap: INSUFFICIENT EVIDENCE
90-day seasoning supported: INSUFFICIENT EVIDENCE
Quarterly rebalancing supported: INSUFFICIENT EVIDENCE
Ready to freeze: NO
DDEI live ready: NO

Historical data inventory

Source	Range	Price	Mcap	Free-float	Failed names	PIT
PostgreSQL MarketDataObservation (live ingest)	2026-08-31 only	A	B	NONE	NO	YES
DdeiAsset / DdeiAssetMarketObservation	empty at reconstruction start	NONE	NONE	NONE	NO	NO
ASSET_MASTER + SNAPSHOTS (ddei-universe.ts)	2014-12-31 → 2025-12-31 (12 research snapshots)	NONE	C	C	YES	PARTIAL
Yahoo Finance chart API (weekly closes, cached)	ticker-dependent; BTC/LTC from 2014-09; broad alts typically from 2017-11	B	NONE	NONE	PARTIAL	PARTIAL
Binance public klines 1d	BTC/ETH from 2017-08; FTT from 2019-12; LUNAUSDT cut 2020-08-21→2022-05-13	B	NONE	NONE	PARTIAL	PARTIAL
CoinGecko public API	rolling 365 days only (error 10012 beyond)	B	B	NONE	NO	NO
Coinpaprika / CryptoCompare public	blocked on current public plans	NONE	NONE	NONE	NO	NO
HISTORICAL_EVENTS catalog	2014 Mt. Gox → 2025 Bybit	NONE	NONE	NONE	YES	PARTIAL

Missing historical data

Vendor point-in-time free-float market-cap history. Cannot distinguish observed free-float vs circulating cap. Weights use ordinal/Zipf size proxies. Proxy: SNAPSHOT ranks + Zipf/BTC-dominance overlay (class C)
Daily/quarterly observed market-cap panel 2013–2026. Cannot reconstitute quarterly on contemporaneous cap. Reconstitution only at 12 research snapshots. Proxy: Buy-and-hold between snapshots using weekly prices (class B).
Institutional custody / venue-depth / insider allocations historically. Modern qualification gates cannot be applied without hindsight. Proxy: Not applied in the historical engine.
Non-survivorship total-return panel for delisted names without Yahoo/Binance prints. Some dead names have membership (snapshots) but no price path. Proxy: Excluded from return contribution; still counted in membership/eligibility tallies.

Architecture tournament (structural, not this path) · DDEI overview