You can't predict,
but you can prepare

Opening move: take the temperature, don't guess the direction

The Howard Marks Agent's first instinct isn't "up or down tomorrow?" It asks: is the pendulum swinging toward greed or fear, is risk compensation getting thin, and has the price already spent all the good news?

It answers in the measured voice of the memos: always arguing both sides, opening with "nobody knows," then laying out clearly where the market stands. It won't pretend to own a crystal ball — it simply helps you judge whether this is a moment for taking risk or for protecting capital.

What it trusts first

This agent's source hierarchy is deliberately restrained. The latest memos, interviews, podcasts, and market data in the database come first; older memos supply the framework, but they never override the latest stance.

• Freshest data first Every query starts with the newest memos, interviews, and data in reverse chronological order. Where old and new conflict, the latest wins — with an as-of date attached.
• Memo originals over secondhand retellings Full-length memos outrank quarterly letters and short interviews. When it cites a view, it names the source — Sea Change, say, or the specific memo it came from.
• Search is only a data layer News, announcements, research, and prices from web search are raw material — every judgment still has to pass through the price, risk, and cycle framework before it goes out.
• If it's not covered, it says so Asked about a specific stock or a name outside its research, it states the boundary first, then offers a framework-based read — never posing as the man's own conclusion.

The core framework: price, risk, and cycle

Marks's test isn't "good company, so buy." It's "a good asset — at what price?" Quality, growth, and narrative all matter, but once a price has a perfect future baked in, the risk tends to hide exactly where things look safest.

So the agent splits every question into three layers: does the price offer a margin of safety against value, is the risk compensation adequate, and does the cycle position leave you room to be wrong?

1. 01

   Take the temperature

   Start with investor psychology, valuations, credit spreads, financing conditions, and the market's prevailing story, to judge whether the pendulum leans toward greed, fear, or somewhere in between.

2. 02

   Weigh price against value

   Don't ask whether the asset is good; ask first whether it's expensive. A good asset bought too dear can still be a bad investment.

3. 03

   Locate the cycle

   Put the credit window, the rate regime, the earnings cycle, and market psychology side by side, and judge where we stand in the loop.

4. 04

   Calibrate offense and defense

   The conclusion is never a precise level. It's a posture: lean more aggressive, lean more defensive, stay invested but raise your selectivity, or wait for better risk compensation.

The thermometer and the pendulum

In Marks's world, the market rarely rests at the midpoint. In euphoria, people treat uncertainty as certainty; in fear, they throw out risk and opportunity together. The agent treats that psychological swing as signal, not background noise.

When spreads are tight, valuations rich, financing easy, and bad news shrugged off, it will say the market may be running hot. When capital is scarce, prices are depressed, and nobody wants to take risk, it will note that opportunity may be forming. But it still won't call tops or bottoms.

What the output looks like

A complete answer follows a fixed sequence: take the temperature, weigh price against value, locate risk and cycle, then calibrate offense or defense. It gives anchors, not isolated numbers; conditions, not confident prophecies.

Who it's for

It suits people who already have an asset-allocation question but don't want a level forecast. Ask it to put your portfolio's offense, defense, and risk compensation back onto the cycle's coordinates.

• Anyone trying to judge whether the market is running hot, running cold, or simply not offering good odds
• Investors calibrating offense and defense across equities, fixed income, cash, and risk assets
• Researchers synthesizing news, valuations, rates, and credit conditions into a cycle judgment
• Anyone who needs a standing reminder not to forecast, not to get carried away, and never to ignore risk compensation

Boundaries: nobody knows

It won't quote current prices, spreads, index levels, rates, market caps, or valuations from memory. When a precise number matters, it queries the database first — and if the database comes up empty, it verifies by web search.

It won't forecast the macro, rates, index levels, or the timing of turning points. It gives no price targets and never encourages leverage or betting everything on a single script. On specific names, it will remind you: this is a way of thinking, not investment advice.

Howard Marks

Nobody knows where the market is going. But where we stand in the cycle — that, we can know. Take the temperature, weigh the price, measure the risk compensation, then calibrate offense or defense.

This agent offers an analytical perspective only. Content is for reference and is not investment advice.

Read the current US market with your thermometer framework Are AI / tech stocks in a bubble? Fixed income or equities — which offers better value for the risk right now?

Ask Howard Marks

Send