One AI Prompt to Rank Your Portfolio's Hidden "Iran War" Exposure

The US-Iran conflict has wiped $12 trillion from global markets in a month.

Oil is above $106 a barrel and the Strait of Hormuz is 90-95 percent shut down.

Every investor I talk to is thinking about the same thing right now. But here’s what I’ve noticed: few are thinking about this from a P&L basis ticker by ticker level across their portfolio.

“Do I own energy stocks?” is the wrong question.

The real risk is hiding in the second and third-order cascade (!)

For example ... Shipping costs are spiking. Which means input costs ripple through food and packaging. And consumer spending compresses as gas eats into discretionary budgets. You also get into rates exposure with energy being a core inflation component the Fed uses to determine rates policy.

I wanted to use AI to map this full cascade across a real portfolio and rank every holding from most to least exposed.

The practicality of this exercise is to use just 1 single prompt and uncover hidden macro level exposures customized to YOUR portfolio.

Here’s the plan:

1. Map the first, second, and third-order effects of the oil shock
2. Score each holding in a real fund’s portfolio against every layer
3. Rank from most exposed to least exposed
4. Find the surprises hiding in plain sight

I used Bill Ackman’s Pershing Square portfolio (it's public) as the demo portfolio as an example for this exercise.

The Prompt

I ran this on Perplexity Computer with the 13F data attached as a CSV.

If there are parts of this analysis you want to customize, swap in your own portfolio and adjust the scenario assumptions.

Pershing's Portfolio I used: pershing_square_13f_q4_2025.csv​

Prompt: oil_shock_analysis_prompt_ackman.pdf​

The Result

Full output here: Link​

What Perplexity Computer came back with was a brief report alongside 2 quick visuals so you can see the rank-order of the most impacted stocks in your portfolio.

We have 2 graphics this prompt produced:

1. A heatmap in rank order mapped by 1st order vs 2nd order vs 3rd order impacts
2. A graphic showing impact vs percent of portfolio (I highly recommend running this exercise with both the names of your tickers alongside the % of your book)

The most interesting finding: Howard Hughes Holdings ($HHH) is far more exposed than most people would guess.

Oil-driven inflation threatens to push the Fed back toward tightening. That spikes mortgage rates. HHH has $1.6 billion in future condo revenue dependent on buyer confidence and mortgage affordability. Plus $1.2 billion in variable-rate debt.

Here are the key findings from the scoring:

• $HTZ scored 8.0 (CRITICAL) but it’s only 0.50 percent of the portfolio, so it’s immaterial
• $UBER scored 7.0 (ELEVATED) at 15.9 percent of the portfolio. This is the biggest dollar-weighted risk. Drivers face an extra $150/week in fuel costs. Uber is burning cash on fuel subsidies to prevent a driver exodus, while consumers simultaneously cut back on rides
• $QSR scored 6.3 (ELEVATED). Tim Hortons runs its own trucking fleet with direct diesel exposure. Food commodity costs across all brands are oil-linked through fertilizer, animal feed, and packaging. Plus 4.2x net leverage with 43 percent variable-rate debt

Here's the big picture of this exercise and how you can apply this going forward:

The technique works for any macro shock. Swap out the cascade layers for whatever the crisis is. Tariffs, banking stress, China-Taiwan. The framework is the same.

Bill Ackman runs a concentrated book, but you can see the value of this exercise get augmented if you have say 100 names in your portfolio.

For any times of macro crisis, speed to action is critical and having this sort of granularity via AI techniques like this gives you precision on timely decision making.

Try this with your own portfolio. Upload your holdings CSV, adjust the scenario, and see where the risk is hiding.

Personal

I've been thinking about something that's hard to articulate but I know many of you feel it too.

The job of an investor used to be deep focus. Read the 10-K. Build the model. Sit with the thesis. Let conviction form slowly.

Now I have 10 Claude Code sessions running, 12 tabs open, and a backlog of ideas I generated faster than I can execute them.

What used to be hard is now easy. And because it's easy, you feel pressed to do more of it.... Always more... Always faster.

I think the future of this job looks less like a research analyst and more like a portfolio manager of ideas. You're allocating attention across a dozen threads, context switching constantly, triaging what deserves deep work versus what you delegate to AI.

That's a fundamentally different skill than what most of us were trained on.

I'm still figuring it out. If you've found a system that works, hit the reply button. I'd genuinely love to hear it.