Reference alpha 1: a boring trend that is honest about itself

Not investment advice

This is an example of a method, not a recommendation, an offer or a call to trade, and not a promise of profit. Simulated and past results do not predict future returns. Trading can lose money.

examples/reference_alpha/ holds a deliberately simple, low-return strategy, so you can see what a real (modest) alpha looks like after every check of Monte-Neo: diversified time-series trend following with volatility targeting.

• Rules fixed in advance. Three look-back windows (63, 126, 252 bars) are averaged, never picked by results. Positions are sized by trailing volatility. The portfolio rebalances monthly in four weekly tranches, with a no-trade band.
• Expectations from the literature, not a promise. A net Sharpe of about 0.3 to 0.7 on long, diversified histories, drawdowns of 15 to 30 %, and years without profit.
• When it works: long directional moves, crises with sustained trends, many weakly correlated markets.
• When it fails: ranges and whipsaws, sharp V-shaped reversals, everything moving together, one instrument, high costs.
• How to "retrain" without overfitting: never re-tune the windows on recent performance. Volatility estimates update by themselves. Any rule change is a new version, registered with monte-neo register and counted in the ledger.

What the verifier says on three synthetic markets (12 instruments, 12 years):

Market	Verdict	Net return	Sharpe	Deflated Sharpe
Planted trends	PASS_WITH_WARNINGS	+18.4 %	0.30	0.84
Random walks	PASS_WITH_WARNINGS	+11.7 % (luck on one path)	0.22	0.77
Choppy, mean-reverting	REJECT	−32.9 %	−0.65	0.02

The random-walk row is the lesson: one lucky path can look like an edge. Across eight more seeds the random walks average −2.4 %, and the Deflated Sharpe never clears 0.95. The full rules, the tables of when it works and when it fails, the decision table for a drawdown and the commands are in examples/reference_alpha/README.md.