The K Curve: Where Noise Gives Way to Structure

Two truths that only look like a contradiction

Two things are true about markets at the same time, and they sound like they cannot both be. The first is that the next move in a broad index is close to random price action. The second is that long-run outcomes contain structure that short-term price changes do not. Most people pick one and dismiss the other. The work that matters lives in holding both.

They reconcile the moment you separate them by time. In the short run, new information dominates. Over long horizons, the temporary part of every move has more time to fade, and cumulative outcomes track a slower structure more closely. The randomness is real up close. The pull toward the law of averages is real from far away. Time is what tells them apart.

This piece defines that slower structure, which I call the K curve. It explains what the curve measures, where the idea comes from, and why our ability to say anything useful about a market changes continuously with the horizon we care about. It is the main entrance to how OpenK thinks.

Why the short run is close to random

The random walk is not just a claim that prices are meaningless. It is a claim about information. Tomorrow's move depends on information that does not exist yet, or that exists but has not yet been interpreted in the way it will be tomorrow. By the time new information is widely known and understood, much of its effect is already reflected in the price.

Even knowing the news in advance may not be enough. A weak economic report can cause the market to rise because investors expect interest rates to fall. Strong earnings can cause a stock to decline because the result failed to clear an even higher expectation. The reaction depends not only on the information itself, but on what was already priced, how investors were positioned, and which interpretation wins.

So when I say the short run is close to random, I mean it in the strong sense. The problem is not that the tools are immature. The thing being predicted is dominated by the unknowable. Any honest framework starts by conceding how much of the short run is out of reach and is then built to work anyway.

Irrationality does not hand you the short run

It is tempting to think that if markets overreact, the overreaction is easy money. It is not. Irrational reactions are themselves unpredictable. Knowing the market will sometimes overweight a fresh headline tells you nothing about which headline, in which direction, by how much, or for how long.

Recognizing that a crowd is fearful is not the same as knowing when the fear peaks. A market can look obviously expensive and keep rising for years. It can look washed out and still fall much further. Short-term mispricing can be real and still be untradeable in advance because the timing and size of the correction are set by an interpretation shift that no one schedules.

This is why turning points look much cleaner in retrospect than they ever did in real time. Once a trend breaks, everyone can identify the leverage, the speculation, the stretched assumptions, and the warning signs. Before it breaks, those same conditions can persist far longer than expected. Being right that the market is wrong is not enough. You also have to know when it will matter, or be able to survive until it does.

A moving reference, not a fixed average

The K curve is a moving anchor. Rather than choosing one look-back window and treating it as the correct one, it draws information from many time scales at once. The result is a slower reference that reflects where the market has been, how its long-run path is changing, and what kind of return that path currently implies.

In any dataset, the average gives you a reference point. If you knew what the market's average return would be in the future, you would know whether the current path was running above or below it. The K curve is my attempt to estimate that future average as it moves.

It is not a fixed average. It is not a fair-value line, a target price, or a deterministic forecast, and it is not a moving average, which is only a smoothed summary of past prices. The distinction that matters most is between a fixed reference and a moving one. A fixed reference assumes some constant level, or constant rate, that prices eventually snap back to. The K curve assumes the opposite: the anchor itself is alive.

It bends as the market and the economy change around it. That is important because the long-run return of a market is not fixed. Productivity changes, valuations change, financial conditions change, and the productive structure of the economy changes. A useful anchor has to move with them.

A market operating on many clocks

There is no single correct time scale for understanding a market. A trader, a pension fund, a corporate treasurer, and an individual investor can look at the same price and see entirely different things. Their decisions are made across different horizons, but they all meet in the same market.

Most models choose one clock in advance. They look at a particular trend, return period, or moving average and assume that window contains the relevant information. The K curve does not. It combines the market's behavior across many horizons into one moving anchor. No single horizon tells the whole story.

How the anchor becomes useful

The K curve takes the market's historical return structure and projects that anchor forward. Current prices can then be viewed relative to the return path implied by the market's own history.

That does not produce one inevitable future. It creates a reference against which future outcomes can be studied. When price moves far away from the anchor, the range and probability of future returns may change. When price remains close to it, the market may behave differently.

The K curve is therefore not the prediction by itself. It is the anchor that makes prediction possible.

Where the K curve comes from

Good ideas rarely arrive from nowhere, and this one has a clear lineage. The K curve's closest relative in the published literature is the Heterogeneous Autoregressive model of Corsi, which formalized the idea that markets should be studied across multiple time horizons. Corsi used that principle to model volatility. The K curve applies it to a different question: estimating the market's moving anchor return.

Credibility in this work comes from being honest about what is borrowed and what is new. The multi-horizon idea is borrowed and battle-tested. What is new is the turn from that idea into a moving reference, and the role that reference plays inside a forward distribution.

Why most information is temporary

The claim that markets are more structured over long horizons rests on a claim about information: at the level of a broad index, the overwhelming majority of daily information is temporarily impactful on price. It moves expectations, liquidity, or fear without permanently changing what the economy can produce.

That is a high bar, and it needs to be. For information to leave a lasting mark on the index, it has to alter the productive structure itself: the stock of capital, the direction of technology, the capacity to generate output over time. Most news does not do that. It moves sentiment and financing conditions, and those fade as the next quarter and the next cycle arrive.

Markets and the economy still operate as a two-way loop, which is why temporary shocks can create real consequences without permanently changing the slower structure beneath them.

Predictability is a function of horizon

If most information is temporary and the reference moves, then predictability is not a fixed property of markets. It changes continuously with the horizon you care about. Near the front, uncertainty is mostly about unknown information. Far out, it is mostly about whether the structure holds. Those are two different problems, and conflating them is a shortsighted mistake.