Week 43: The Priced-In Exit and the Earnings Hostage Problem

Week 43 delivered near-flat realised P&L (+$2.68) as disciplined asymmetric exits on Moderna offset Intel losses, but the portfolio now holds concentrated earnings-event risk across two Tech names reporting four days apart. The engine's directional accuracy on actionable BUY/SELL calls remains 18-43%—essentially random—while blended accuracy inflates to 86-93% by counting HOLD predictions as correct by construction.

The week's defining narrative was not what we traded but what we avoided. Moderna spiked to BUY conviction midweek on Phase III cancer vaccine momentum, but the priced-in score flagged 2.33—the hottest reading in the active book—before the market confirmed the signal with a 17% reversal. We exited the $1,000 position for +17.35% before the decay, a clean demonstration of asymmetric exit discipline working as designed.

Intel provided the counterpoint: Monday's spike drew two $1,000 paper longs at ~$92 entry, conviction peaked briefly, then decayed back to HOLD as news flow normalised. Both positions now sit underwater at $86.66, and the original signal thesis that justified entry has evaporated, leaving only binary earnings-event exposure heading into the Oct 29 after-close print. The engine does not predict earnings surprises—it trades news-flow inflection—and the inflection that triggered the longs has already reversed.

The portfolio closed Week 43 with just one live conviction: Palantir at 11.5, a $1,000 long position reporting earnings Nov 2. Four of seven active names report next week, and our entire live exposure concentrates into two Tech names four days apart, both hostage to guidance-driven prints in a macro environment where forward commentary moves stocks more than the quarter itself.

Portfolio Highlights

MRNA REDUCE High

Exited $1,000 position for +17.35% after priced-in score flagged 2.33 risk before the 17% reversal on cancer vaccine news. Directional accuracy at 18.2% validates the current REDUCE signal—the engine got the exit right but struggles with near-term timing.

INTC HOLD Medium

Two $1,000 longs opened at ~$92 now underwater at $86.66 heading into Oct 29 earnings. Original BUY signal decayed to HOLD as news normalised, leaving only event risk. Directional accuracy at 50% on small sample offers no edge on the binary print.

PLTR BUY Medium

One $1,000 long held into Nov 2 earnings at conviction 11.5, but directional accuracy flags caution: 37.5% at 1-day, 28.6% at 5-day. The engine sees news-flow momentum; the risk is a guidance-driven print that trades on forward commentary, not the quarter.

ARCT WATCH Low

Street upgraded to strong-buy with $22.29 target vs. ~$10 price, but the engine shows zero conviction. The divergence is structural: small-cap biotech catalysts often do not flow through news aggregation until binary readouts hit, and the pipeline trades inflection, not fundamental value.

Outlook

Next week concentrates binary earnings risk into two Tech names four days apart with weak directional accuracy and decayed signal conviction on one. The adversarial case is straightforward: neither INTC nor PLTR positions are justified by current signal strength, and the pipeline has no edge on earnings surprises. The counter-case is equally valid: a clean INTC beat-and-raise on Oct 29 would flip the framing entirely. The honest read is we do not know, and neither does the engine. Elevated probability that volatility spikes Tuesday evening and again Friday evening, with portfolio outcomes driven by binary event risk rather than predictive signal. PATH remains dark with one article and zero conviction. PLUG and TEM are flat watches with no catalyst flow.

This commentary is generated by an AI system analysing quantitative trading data. It does not constitute financial advice. Past performance does not guarantee future results.