XNAS Inversion: Why US CEOs Trick Insider Trading Models

The Raw Observation

Sigma's BUY universe (Dec 2024 -- Jan 2026) included all markets where insider disclosure was available: XPAR, XNAS, XAMS, XWBO, XBRU, XHEL, XOSL, XSTO, XETR. The U.S. dominated by filing count via SEC, logically given market size.

First backtests isolating XNAS alone: Sharpe 0.01, CAGR -5.0%. Isolating XPAR alone: Sharpe 1.85, CAGR +29%. The divergence proved categorical.

Audit 101 covered 1,229 XNAS BUY rows in the OOS period (Jan 2025 -- May 2026). The raw mean r90 return on this universe is +5.13%. But the V13.5_stack picker selects the worst sub-segments, inverting the signal.

The Investigation: Role, Cluster, Capitalization

We segmented the XNAS cohort (323k filings, 56% of the 595k total dataset) by:

• Insider role: CEO vs CFO vs BOARD vs OTHER
• Signal type: cluster (multiple insiders same week) vs solo
• Market cap bucket: micro, small, mid, large, mega
• Transaction side: BUY vs SELL
• Time horizon: T+30, T+60, T+90

The result is unambiguous:

V13.5 awards CEO/CFO +2.5 and CHAIR +1.5 in the score. On XNAS, those are exactly the two roles with negative mean return. Board-of-directors buys, weighted only +0.5 by V13.5, are the strongest positive role signal on the US tape. The picker is upside down.

On clusters:

V13.5 adds +2 to score when isCluster is true. On XNAS, clusters underperform singles by 8 percentage points. Likely because US clusters are often forced disposals (Rule 144) or 10b5-1 plan executions, not conviction buys.

Hypothesis: US vs EU Compensation Regimes

Three converging factors explain the inversion:

1. RSUs and forced post-vesting liquidation. In the U.S., CEOs receive the majority of compensation in Restricted Stock Units. These remain illiquid until vesting, at which point executives must sell to diversify and cover taxes. This is not an economic signal -- it is a mechanical tax and risk-management transaction. In Europe, compensation packages remain anchored in cash and direct bonuses. Insider sales are far less "mandatory" and tend to be more discretionary.

2. Staggered blackout windows. SEC regulation imposes blackouts but the window varies: "no trading during the month following earnings" (company-specific). Insiders concentrate purchases in fixed, predictable windows, reducing information asymmetry. French and European regulators enforce looser windows.

3. Defensive vs. opportunistic signaling. A U.S. CEO rarely buys for genuine economic conviction. Often it is PR after a drawdown: "see, I'm confident!" This defensive buy predicts nothing.

The Worst XNAS Picks in OOS

The 15 worst OOS picks qualitatively: Bitcoin Depot (-63.8, -56.8, -56.8), Abundia Global Impact (-61.7 twice), Firefly Aerospace (-56.4 twice), Eos Energy (-55.9 twice), Spruce Biosciences (-53.9 x4), ConnectM Technology (-54.5), PagerDuty (-53.6).

All have pctMcap < 0.005 (insignificant amounts) and either a crypto-adjacent business model or a recent SPAC/dilution pattern. The picker treats these as legitimate insider conviction. They are noise or unwind activity.

The Decision: Exclude XNAS from V14e

Given this breakdown, we chose a restricted-universe strategy: V14e = EU_strict, excluding XNAS and other unvalidated North American venues. The resulting universe (38,484 total declarations, 12,384 OOS rows over 2 years) posts Sharpe 0.53, DSR -0.01, CI95 [-0.46, +4.20], CAGR +17.1%, maxDD -27.0%, win rate 48.7%.

This mirrors the treatment we already apply to CNMV (Spain) and CVM (Brazil): we do not recommend these markets until independent OOS validation is available.

Note: the results above are validated on the EU_strict universe (8 European markets: XPAR, XAMS, XWBO, XBRU, XHEL, XOSL, XSTO, XETR), after exclusion of XNAS following audit 101 (signal inversion on the US tape).

Implication for Retail Traders

If you trade U.S. markets using insider data, do not extrapolate European insights. The mechanisms are fundamentally different, and the majority of tradeable signals remain concentrated in Continental and Nordic Europe.