If prediction is philosophically flawed, what do you actually do instead? Part 4 of the No Crystal Ball series.
The Short Answer
In a reflexive system, you cannot reliably forecast the future, because the forecast changes the future. But you can read the present with precision. The shift is from prediction (where will the price go?) to description (what is the system doing right now?). You stop trying to predict the rain and start reading the river — its level, its speed, its direction, the debris it’s carrying. The river tells you everything the weather forecast pretends to.
Where We Left Off
The first three parts of this series made a hard claim: markets resist prediction not because we lack data or computing power, but because of what they are. Participants read the same research, run the same models, and adapt to any pattern that becomes known. The act of observing an edge erases it. Markets are reflexive — the observer is also a participant, and the measurement changes the thing measured.
That’s a bleak conclusion if you stop there. It sounds like surrender. It isn’t. It’s a redirection.
The mistake was never the math. The mistake was pointing the math at the wrong question.
The River and the Rain
Picture a river. You want to know whether to cross.
The prediction approach: forecast next week’s rainfall, model the watershed, estimate runoff, and project the river’s level three days out. Elegant. Also fragile — one wrong assumption about a storm cell two valleys over and your model is useless, and you won’t know it’s useless until you’re already in the water.
The description approach: walk to the bank. Look at the current level. Watch how fast a stick floats past. Note whether the water is rising or falling against the rocks. See what it’s carrying.
The second approach makes no claim about the future. It reads the present state of a flowing system with precision — and a flowing system’s present state already contains most of what its near future will look like, because rivers, unlike rain, have momentum.
Markets have momentum too. Not destiny — momentum. The difference matters.
What Description Actually Measures
A description-first analyst doesn’t ask “where is this going.” They ask a stack of present-tense questions, each one answerable from data available right now:
- Direction and strength. Is the trend up, down, or sideways — and how convincingly?
- Momentum. Is that trend accelerating or tiring?
- Volatility regime. Is the water getting choppier or calmer?
- Volume confirmation. Is participation backing the move, or is it a thin drift that could reverse on a breath?
- Where risk is concentrated. If this breaks, where does it break from?
Notice what’s missing: a price target. A date. A “this will happen.” Every one of these questions is verifiable now — you’re either reading the river correctly or you’re not, and you find out immediately, not in hindsight.
That immediacy is the entire advantage. A prediction can only be graded after the fact, by which point the information is worthless. A description is graded on contact with reality.
Why This Survives What Prediction Doesn’t
Description is robust to the two things that kill predictive models:
It survives regime change. A predictive model trained on a trending market quietly breaks when the market starts mean-reverting, and it keeps emitting confident numbers right up until it blows up. A description-first reader sees the regime change as it happens — that’s literally one of the things being measured — and adjusts. You’re not betting that the past pattern persists. You’re reading the present pattern, whatever it is.
It doesn’t self-destruct when published. Recall Goodhart’s Law from Part 3: once a profitable pattern becomes a target, it stops being a good pattern. Published edges get arbitraged away. But “momentum is currently strong and volume confirms it” isn’t an edge to be crowded out — it’s an observation. Everyone can know it and it stays true, because it’s a statement about the present, not a claim on the future. You can’t arbitrage away a thermometer reading.
The Honest Limit
I’ll be straight about the cost, because this series has never sold a free lunch.
Description doesn’t tell you what to do. It tells you what is. The decision — cross the river or don’t — is still yours. A description-first tool hands you a clear, sourced, present-tense read of the system and then gets out of the way. It refuses to pretend it knows the future, which means it also refuses to make the decision for you.
Some people hate that. They want the oracle. They want the price target and the date, and they’ll pay more for confident nonsense than for honest uncertainty. That’s a real market — the prophecy market — and it’s enormous. We’re not in it.
We’re in the clarity business. Read the river. Decide for yourself. Cross with your eyes open.
Key Takeaways
- Prediction asks where the system is going; description asks what it’s doing now — and only the second question is honestly answerable in a reflexive system.
- Flowing systems have momentum — the present state of a river contains most of its near future, in a way the weather never will.
- Description survives regime change — because detecting the regime is part of what it measures.
- Description doesn’t self-destruct when published — an observation isn’t an edge to be arbitraged away.
- The cost is honesty — it won’t make the decision for you, and some people will always prefer the oracle.
What’s Next
In Part 5: “Where the Electrons Don’t Read the Papers,” I follow this logic somewhere it took me by surprise.
If prediction fails in reflexive systems — systems whose participants adapt to being observed — then the obvious question is: is there a system that doesn’t adapt? A system where the observer problem simply doesn’t apply, where you can predict, honestly and well, because the thing you’re studying isn’t reading your work?
There is. And chasing that question is how a markets guy ended up in a physics lab.
[Read Part 5 →]
Frequently Asked Questions
What does description-first mean in trading?
A description-first approach measures the current state of a market — trend direction, momentum, volatility, volume — instead of forecasting future prices. It treats the market as a flowing system to be read in the present rather than a target to be predicted. The output is verifiable immediately and doesn’t depend on patterns persisting into the future.
Why is reading current market conditions better than predicting?
Reading current conditions is verifiable right now, robust to regime changes, and doesn’t self-destruct when widely known. Prediction, by contrast, can only be judged in hindsight, breaks when market conditions shift, and gets arbitraged away once published. In a reflexive system where observers are also participants, description is the more honest and durable approach.
Does description-first analysis tell you when to buy or sell?
No — and that’s deliberate. It gives you a clear, sourced read of the present state of the market and leaves the decision to you. It refuses to pretend it can predict the future, which means it also won’t make the trade for you. The clarity is the product; the decision stays yours.
MarketCrystal provides AI-powered trend analysis for cryptocurrency markets. We don’t predict the future — we help you see the present clearly. [Learn more →]