Why Better Predictions Start by Counting the Right Things

May 28, 2026

10 min read

3 views

71%

The mistake of measuring the story instead of the mechanism

What if the most important thing in your model of the world is not the thing you can see most easily, but the thing you have been trained to ignore?

That is the hidden challenge connecting medicine, investing, and forecasting. In cholesterol testing, it is tempting to focus on how much cholesterol is floating around. In markets and decisions, it is tempting to focus on the confidence of our beliefs, the elegance of our story, or the loudness of our conviction. But all of these can be misleading if they measure the wrong layer of reality.

A simple idea sits underneath the confusion: outcomes are driven less by the size of the payload than by the number of active units carrying it. In heart disease, risk is more closely tied to the number of apoB particles than to the total cholesterol mass inside them. In decision making, the quality of our future choices depends less on the beauty of our beliefs than on whether those beliefs are treated as testable hypotheses. In complex systems, what matters is not the static snapshot but the interacting agents and feedback loops that produce behavior over time.

The deeper lesson is not just about biomarkers or forecasting. It is about how to think when the world is complex, layered, and easy to oversimplify.

Cholesterol, beliefs, and the tyranny of the obvious metric

Most people understand risk by looking for the obvious number. If cholesterol is high, worry. If a prediction sounds plausible, trust it. If a narrative is coherent, believe it. These instincts are natural, but they often confuse mass with mechanism.

Think about a warehouse filled with boxes. You might count the total weight to estimate how much is inside. But if the real danger comes from the number of boxes that can be opened and moved, weight alone will miss the point. Two warehouses can weigh the same and behave very differently depending on how many boxes they contain and how they are arranged. That is what apoB reveals in cardiovascular risk. The issue is not simply how much cholesterol is present, but how many atherogenic particles are circulating and capable of doing damage.

This is a powerful pattern because it generalizes. We repeatedly mistake the visible aggregate for the causal unit. We ask how much, when we should ask how many. We ask how strong the conviction is, when we should ask how it will be tested. We ask whether a story feels right, when we should ask whether it is robust under pressure.

The real enemy is not ignorance alone. It is metric drift, the tendency to optimize or rely on measurements that are easy to see, easy to explain, and often only loosely connected to what actually matters.

The world is full of comforting summaries that conceal the operative variable.

In medicine, that can mean focusing on LDL cholesterol when particle count gives a sharper view of risk. In life, it can mean focusing on productivity theater instead of actual progress. In investing, it can mean mistaking a compelling thesis for a sound process. In each case, the temptation is the same: use a proxy because it is available, then forget that it is only a proxy.

Complexity means the map is never the territory

The reason these mistakes persist is that many important systems are complex adaptive systems. They are not simple machines with one input and one output. They are living networks of interacting parts, feedback, adaptation, and surprise. A person, a market, a disease process, and even a belief system are all more like ecosystems than like clocks.

In a clock, you can inspect the gears and know the function. In an ecosystem, the same organism can be predator, prey, competitor, and signal carrier, depending on the context. That means our favorite shortcut, the one that works in simple situations, often fails in complex ones. The system changes in response to our measurement, our treatment, or our intervention.

This is why chasing a single headline metric can be dangerous. A company may improve quarterly earnings by starving the future. A patient may improve one lab number while the underlying risk remains. A forecaster may feel accurate because the narrative is tidy, while the real world is moving in a different direction.

Complexity forces a humbler stance: the model is not the truth, only a working approximation. The useful question is not, “Do I have a decisive story?” It is, “Which variables are actually causal, which are merely correlated, and which are being distorted by feedback?”

That shift changes everything. Once you think in complex systems, you stop worshiping single indicators and start asking about structure. What are the units? What is the interaction pattern? What feedback loops amplify error? Where does the system hide its real risk?

This is exactly why the number of apoB particles matters. It is not just another lab value. It is a clue that the system is built from particles that can enter arterial walls and initiate damage. It is closer to the causal action than the total amount of cholesterol riding along inside them.

The same logic applies to judgment. A belief is not valuable because it is cherished. It is valuable if it survives contact with evidence.

Beliefs are not treasures, they are instruments

One of the most liberating ideas in decision making is this: beliefs are hypotheses to be tested, not treasures to be protected.

That sentence should be etched into every investor's notebook, every clinician's mental model, and every intelligent person's operating system. Why? Because the main failure mode of smart people is not lack of intelligence. It is overattachment to a view of the world that once seemed useful and now serves the ego more than the truth.

When a belief becomes a treasure, the question changes. It is no longer, “Is this true?” It becomes, “How do I defend it?” That subtle shift is disastrous in complex environments, because the world does not care about our attachment. Reality keeps updating while we keep narrating.

Imagine a sailor who treats his map like a sacred object. He polishes it, quotes it, and refuses to mark it up. But the sea is changing. The route is shifting. Winds and currents are active forces, not passive scenery. A good map is not a relic. It is an instrument that must be revised as new information arrives.

That is what testable beliefs do. They keep us oriented without trapping us. They invite error correction. They make room for surprise. They help us distinguish between confidence and calibration.

This matters because confidence feels productive, but calibration is what preserves us from disaster. You can be brilliant and wrong, articulate and wrong, certain and wrong. The people who navigate complex systems best are not the ones who never err. They are the ones who notice error quickly, revise cleanly, and keep their identity separate from their thesis.

That is why the healthiest intellectual posture is neither cynicism nor blind conviction. It is disciplined provisionality. Hold your view firmly enough to act, lightly enough to update.

The particle principle: count the agents, not just the totals

There is a useful mental model that connects cholesterol, crowds, markets, and beliefs: call it the particle principle.

The particle principle says that in many systems, what matters most is not the aggregate quantity but the number of active agents capable of producing an effect. A few large units can behave very differently from many small ones. The total may be identical, but the dynamics are not.

In cardiovascular risk, apoB particles are the agents. They are the individual vehicles that can enter arterial walls. Measuring cholesterol mass alone is like estimating traffic danger by counting the total weight of cars on the road, while ignoring how many cars are actually moving through a dangerous intersection.

In investing, the particle principle shows up in the difference between a compelling narrative and a high quality distribution of evidence. A story with one powerful example can dominate attention, but if the underlying data points are weak or sparse, the story is fragile. The number of independent confirmations matters more than the emotional force of a single anecdote.

In public opinion, a small number of highly connected individuals can shape the apparent mood of a crowd. In complexity science, local interactions can create global patterns that no central planner intended. In each case, the system is governed by the behavior of units and their relationships, not by a smooth average that hides the interesting parts.

This is why averages can deceive. Averages are not lies, but they are often amputations. They strip away the structure that generates the outcome. If the hazard comes from distribution tails, from clustering, from contagion, or from particle count, then the average gives a false sense of control.

When a system is dynamic, the smallest meaningful question is often not, “What is the average?” but, “What is the active structure?”

That is the bridge between apoB and decision making. Both teach us to look past the surface summary and identify the causal architecture beneath it.

What to do differently when the world is complex

The practical challenge is not simply to admire this insight. It is to use it.

Start by asking a better question whenever you face a metric, a forecast, or a strong opinion:

What is the causal unit here?

For a lipid panel, that may mean asking whether particle number gives a better estimate of risk than cholesterol mass. For a business dashboard, it may mean asking whether revenue is being driven by durable demand or by short term incentives. For your own thinking, it may mean asking whether your certainty reflects understanding or just repetition.

Then ask:

What is this number hiding?

A single metric can conceal composition, distribution, feedback loops, and time lag. A person can look healthy while accumulating silent risk. A strategy can look successful while accumulating fragility. A belief can look rational while becoming emotionally protected from disconfirmation.

Next, adopt a bias toward unit based thinking. Instead of asking whether something is big, ask how many meaningful entities compose it and what those entities can do. Instead of asking whether a market is “bullish,” ask how many buyers are still forced, how many are patient, and how many are merely following momentum. Instead of asking whether your plan is “good,” ask whether its assumptions are independently testable.

Finally, build a habit of pre commitment to revision. If beliefs are hypotheses, then you should know in advance what evidence would change your mind. If a lab marker is a proxy, then know what you would measure next if the first number is ambiguous. If a forecast matters, then track your calibration, not just your hits.

The best thinkers do not merely generate opinions. They create error revealing environments.

Key Takeaways

Prefer causal units over convenient summaries. Ask whether you are measuring the thing itself, or only a rough proxy.
Treat beliefs as testable hypotheses. A belief that cannot be updated has become an identity object, not a tool.
Think in systems, not snapshots. Complex outcomes often emerge from interactions, feedback loops, and distributions, not simple averages.
Ask what a metric hides. Every number removes information. Be explicit about what disappears when you compress reality.
Design for revision. Before you commit to a view, decide what evidence would make you change it.

The deepest question is not what do we know, but what are we counting?

The real insight connecting medicine, complexity, and forecasting is that reality rarely rewards the most convenient measure. It rewards the measure that tracks the mechanism most closely.

That is why apoB matters. Not because cholesterol stops mattering, but because particle count gets closer to the machinery of disease. That is why great decision makers value complexity. Not because complexity is trendy, but because the world does not reduce itself to one neat story. That is why wise forecasters protect their beliefs by testing them. Not because skepticism is fashionable, but because truth is not a possession, it is a process.

So the next time you face a stubborn problem, resist the first number, the first story, and the first explanation that feels complete. Ask instead: what is the active unit, what is the feedback loop, what is the system actually doing, and how would I know if I am wrong?

That is a better way to think about health, markets, and the future. More importantly, it is a better way to live in a world where the most important truths are often hidden inside the things we have not yet learned to count.