# Chimera (three thirds) -- the frontier in three equal lines: efficient core, stacked gold, trend
#
# Idea: a FRONTIER build and a game of simplicity, played at once. Three
#   lines of equal weight, each a different source of return, and two of
#   them capital-efficient: a 90/60 efficient core (stocks plus the duration
#   that pays in a deflationary crash), a stacked gold-and-equity fund (GDE:
#   90 % US equity plus 90 % gold on one dollar), and managed futures for
#   the crisis that unfolds over months. Three thirds of CAPITAL, 143 % of
#   notional, and the diversifiers held at full size without the equity
#   budget paying for them: that is the frontier, reached with the fewest
#   moving parts. It is also this collection's demonstration of GDE, the
#   one fund that stacks gold rather than bonds on an equity core.
#
# Lineage: the third child of Typhon and Echidna, after Hydra
#   (hydra-five-engines-ucits.txt, one head per engine) and Cerberus
#   (cerberus-three-heads.txt, three even thirds of NOTIONAL). Chimera is
#   three even thirds of capital, lion, goat and serpent on one body, and
#   like Cerberus it carries one non-UCITS line (GDE): the buyable route is
#   in the last section.
#
# What 100 EUR holds (notional 143): 60 US equity (31 through the core,
#   30 through GDE), 20 US Treasuries (intermediate ladder, futures), 30
#   gold (futures, inside GDE), 33 managed futures. Regime coverage by
#   weight: growth 60, deflation 20, inflation 63, crisis 63. Read the
#   equity share before anything else: this is a growth book with two
#   hedges at full size, not a risk-balanced one.
#
# Measured (USD, 1996-03 to 2026-09, sim:on, 180-day rebalancing) against
#   the S&P 500 over the same window:
#     CAGR 12.8 % vs 10.4 %; volatility 13.1 % vs 19.0 %; Sharpe 0.98 vs
#     0.61; max drawdown -30.5 % vs -55.3 %; longest recovery 1.9 y vs
#     6.1 y; worst rolling 5y +4.1 %/yr vs -8.2 %, median rolling 5y
#     +11.5 %; beta 0.55, up capture 65 %, down capture 58 %.
#   Stress windows (total return): dotcom 2000-03..2003-03 -12.8 % (S&P
#     -44 %), 2007-10..2009-03 -31.7 % (gold and equity fell together for
#     four months before the trend line caught up), COVID 2020-02..03
#     -19.6 %, the 2022 rate shock -13.6 % (the bond overlay and the
#     equity both down, gold flat, trend +24 %). The gold bears of
#     2011-2013 and 2013-2015 cost nothing: +2.8 and +8.2 %/yr, the equity
#     inside GDE carrying the line.
#
# The unlevered alternative, measured and NOT taken here. The same three
#   lines with plain gold (40 NTSX / 30 IGLN / 30 KMLM, notional 120)
#   returns 10.7 %/yr at 9.8 % volatility, Sharpe 1.08, a -19.4 % drawdown
#   and a better result on every stress window (dotcom -5.5 %, 2008
#   -21.4 %, 2022 -10.2 %). That book is the all-weather answer; this file
#   is the frontier one: the stacked gold line buys 2.1 points of return
#   for 3.3 points of volatility and eleven points of drawdown, at the
#   same Sharpe as the old four-line version of this file. Whoever wants
#   the calmer book moves 33 GDE to 30 gold and the freed capital into the
#   core; the change is one line.
#
# Why no broad commodity line. The earlier version of this file held 15 %
#   of a commodity index next to the trend fund. Managed futures already
#   trade the same contracts, long AND short, and the trend fund carried
#   the 2022 inflation year on its own; a long-only basket adds the roll
#   cost and the 2014-2020 bear for a cover the sleeve provides. Dropping
#   it is what made three clean thirds possible.
#
# Where it fails. A prolonged stagflation with gold AND trend both out of
#   phase (1970s-style; the trend line's backcast reaches 1996 only, so
#   that decade is not in the test); a slow lost decade in the equity
#   core, which 60 points of equity feel in full and the two hedges soften
#   without profiting from; and model risk on the trend line, whose
#   pre-2020 history is a donor chain of other managers' NAVs (see
#   pkg/datasets/simdata/KMLM.csv's header and
#   docs/specs/trend-reconstruction-design.md for the error budget). GDE's own
#   backcast is a 90/90 composite financed at the overnight rate; its
#   pre-2022 history is an engine, not a fund.
#
# Buy route (EU retail): NTSX -> IE000KF370H3 (the UCITS twin); KMLM ->
#   iMGP DBi Managed Futures (LU2951555403, EUR class) or the EUR-hedged
#   AQR class (LU2622190622). GDE has no UCITS twin: a retail account
#   synthesizes the line with physical gold (IGLN, IE00B4ND3602) plus a
#   2x MSCI USA UCITS ETF (CL2, FR0010755611) in roughly equal parts, at the
#   cost of the daily-reset drift of the levered fund; cerberus-three-heads.txt
#   carries the priced version of that synthesis and the access ledger.
#meta rebalance:180
#meta sim:on
34    NTSX            # IE000KF370H3  US Efficient Core 90/60: the growth engine and the deflation duration
33    GDE             # stacked gold and equity, 90/90 on one dollar (no UCITS twin: IGLN + CL2 synthesis): the inflation and currency hedge at full size
33    KMLM            # managed futures (UCITS: iMGP DBi LU2951555403): the crisis that unfolds over months