Mint - AI Semi Edge

Framework

What the model should reward before the market fully agrees.

PREDICT DIRECTION, NOT PRICE

Bottleneck Who owns scarce supply?

Conversion Who turns AI demand into revenue now?

Margin Who gets richer as mix improves?

Expectations Who still has room to surprise?

Catalysts Earnings, capex, design wins, pricing

The first version should stay explicit and falsifiable.

Do not treat all AI semis as one basket.

Prefer names with visible next-quarter revenue conversion.

Punish crowded narratives with weak proof.

Update scores every week, not only after earnings.

Treat bad input data as a model failure.

Mint-native page for the 3 to 6 month semiconductor view.

Ticker	Tier	Trade	1W	1M	Confidence	Valuation Risk	RS	Catalyst	Bias

OPEN CHART

Score History

Factor Engine

What creates edge

What breaks the trade

What must happen

What confirms

Catalysts in the next 3-6 months

What changed enough to matter right now.

Can this actually create tradeable edge?

YES

It can improve trade selection if it updates faster than consensus and measures conversion, not just narrative heat.

It fails if it turns into decorative commentary with stale data and no regime update discipline.

What should get wired in next.

Estimate revisions and earnings dates.

Hyperscaler capex and supplier read-through.

Score history and change-over-time view.

Backtest score changes versus forward returns.