#### Are Financial Markets Governed by Stochastic or Chaotic Processes?

In finance, markets are often modeled using stochastic processes—random price movements that we attempt to predict using probabilities. But this view might be too simplistic. Could financial markets actually behave more like chaotic systems, where small initial changes trigger large, unpredictable shifts? Moreover, could the **noise** that we see in day-to-day price data hold deeper, fractal patterns that reveal underlying trends?

In Tim Palmer’s excellent book *The Primacy of Doubt*, he explores a fascinating concept called **stochastic rounding**. As shown in **Figure 1** below drawn from Palmer’s work, when rounding values deterministically (middle column), significant information is lost, and we can no longer make sense of the original data. However, when noise is introduced through stochastic rounding (right-hand column), the information is not only preserved but amplified (Figure 1). This shows that **noise can enhance signals** rather than mask them, offering a more accurate representation of the data. In markets, what might initially seem like chaotic noise could hold crucial information that amplifies price signals over time.

**Figure 1: Stochastic Rounding: On the left are percentage figures shown in different shades of grey. In the middle, the shade of grey is rounded to one bit (black or white) in a systematic way. On the right, the shade of grey is also rounded to one bit (blackor white) but in a more stochastic way, i.e., using random noise. There is clearly more information on the right-hand side than in the middle.**

Much like Palmer’s rounding example, financial markets may exhibit a **fractal structure**, where patterns repeat across different scales. **Fractals** are self-similar; they look the same no matter the scale of observation. In financial markets, short-term price fluctuations might mirror larger patterns that appear over time, creating an intricate web of price movements that are not random but follow a deeper order.

Palmer discusses how reducing the precision in weather simulations (from 64 bits to 16 bits) appears to degrade the quality of the data, as shown in **Figure 2**. However, when stochastic rounding is applied (Figure 2, bottom), the simulation retains its accuracy. This principle can apply to market data, where introducing noise might help stabilize chaotic movements, allowing underlying **fractal structures** to emerge. As more data is collated over time, these patterns become clearer, revealing price trends that were hidden within what we originally perceived as noise.

**Figure 2: Top: Simulation of fluid turbulence where the individual fluid variables in the simulation are represented by 64-bit numbers. This representation is the default in scientific computation but uses a lot of computer resources. Middle: where the number of binary digits representing the individual fluid variables in the simulation has been reduced from 64 to 16. The turbulent whirls in the simulation are degraded in realism compared with the top figure. Bottom: the same simulation but where the stochastic rounding technique has been used when truncating to 16 bits. Now the simulation is virtually identical to the top figure. Adding noise has made the simulation more accurate. From Paxton et al. (2022).**

**Chaos and Bubbles: Noise as a Stabilizer**

During periods of market crises or bubbles, financial systems often exhibit chaotic behavior, where small, seemingly insignificant events (such as a sudden interest rate change) spiral into much larger market-wide movements. Could this chaotic behavior be linked to the **fractal geometry** underlying market fluctuations?

In **Figure 3**, Palmer introduces the concept of **simulated annealing**, where noise is applied to escape local peaks in a complex system. By introducing randomness (using a pseudo-random number generator, PRNG), the system avoids getting stuck at local maximums and continues to search for the global optimum. This process, which can be analogized to market behavior, suggests that **noise can help traders avoid being misled by false signals** or premature conclusions about trends, offering a more complete picture of market dynamics.

**Figure 3: Simulated annealing is an algorithm which uses noise to find the highest point of a complicated set of peaks and troughs. Here the algorithm is shown progressing to the highest peak in nine steps.**

**Noise: Friend or Foe?**

In markets, the role of **noise according to Palmer **is often misunderstood. Rather than dismissing it as random, Palmer’s research reveals that noise can actually stabilize nonlinear systems. In **Figure 4**, noise is applied to the chaotic **Lorenz model**, resulting in a system that becomes more predictable, with stronger regime structures. The noise actually amplifies the signal by exaggerating the pattern of transitions between regimes, making the system more comprehensible. Similarly, in financial markets, what appears as noise at first might in fact be a crucial ingredient in amplifying signals within the chaos, enabling us to better predict market behavior.

**Figure 4: Top: a time series of the X-variable of the standard Lorenz model. Bottom: a time series of the X-variable of the Lorenz model where noise has been added to the equations.**

Markets, much like weather systems or other nonlinear systems, are not purely random. Instead, they follow patterns that may only be visible when enough data is collected and analyzed at the right scale. As in Palmer’s **simulated annealing** example, where noise helps the algorithm find the highest point of a curve, noise in financial markets could assist in identifying the true trends and avoid false peaks—especially during periods of heightened volatility or uncertainty.

**Fractals** provide a framework for understanding how price data, viewed as noise at first, actually holds a repeating signature that becomes more apparent as more data is gathered. These fractal patterns provide insights into market dynamics that are deeper than traditional models of randomness would suggest. Just as noise in stochastic rounding revealed lost information in weather modeling, noise in financial markets may reveal **amplified signals** that could lead to more accurate decision-making.

## The Role of Gaussian Noise in Enhancing Market Signals

Noise in financial markets is often seen as a distortion, but in certain contexts, like **image processing** with fractals, noise can actually enhance underlying structures. For example, in **fractal convolution algorithms**, adding **Gaussian noise** to an image helps improve the visibility of faint structures (**Figure 4**). The added noise functions like a form of dithering, amplifying weak signals that might otherwise remain hidden.

**Figure 4: Effects of added Gaussian noise in the performance of the fractal convolution algorithm. The upper images are the original picture ×10 without interpolation and the result of the convolution with an XOR (9 × 9) pixel fractal, respectively. The lower images are the same original image with Gaussian noise and the result of the convolution of this figure with the same fractal. We can see that a certain amount of dithering can contribute to a better visualization of faint structures in the resized image. Source: https://www.researchgate.net/figure/Effects-of-added-Gaussian-noise-in-the-performance-of-the-fractal-convolution-algorithm_fig1_231016940**

This concept can be applied to market data as well. Just as noise enhances faint structures in images, noise in financial data—when understood properly—might help amplify **subtle market trends** or signals embedded within complex price movements. Much like in the image processing world, a certain amount of noise could actually **contribute to a clearer understanding** of market dynamics, revealing patterns that might have been missed under traditional, noise-filtering methods.

In both cases, whether in **fractal image processing** or financial data analysis, noise should not always be viewed as detrimental. It has the potential to reveal **deeper, underlying structures**, whether they be faint patterns in a visual image or hidden trends in market price action.

**Rethinking Markets through Chaos and Noise**

Perhaps it’s time to reconsider how we understand financial markets. Instead of viewing them as purely stochastic or random systems, we could draw insights from other **complex adaptive systems**—like weather patterns, ecosystems, or even neural networks. **Chaos theory** might offer a better framework for understanding market behavior, where small, seemingly insignificant events can amplify and drive large-scale movements.

Our traditional approach of trying to eliminate or ignore noise might be misguided. As Tim Palmer’s work shows, **noise** doesn’t always obscure information—it can actually **reveal** underlying structures, enhance signals, and stabilize chaotic systems. Rather than treating noise as a problem to be filtered out, we could reframe it as a critical component of market analysis. What we previously perceived as randomness or “junk data” might hold the key to unlocking deeper market patterns, such as the fractal signatures found in price fluctuations.

In this light, **randomness** may not be as random as we thought. It could be the **elephant in the room**, shaping market dynamics in ways that traditional models have overlooked. Markets are complex, adaptive, and constantly evolving, and our understanding of them must evolve too. By embracing chaos and noise as integral parts of market behavior, we may gain new perspectives on price movements, volatility, and trends—leading to a more comprehensive and accurate understanding of financial systems.

In the end, financial markets may not be chaotic or stochastic, but perhaps a mix of both, where chaos gives birth to new patterns, and noise amplifies the signals we need to see them.

## 2 Comments. Leave new

[…] a previous post, we explored whether financial markets are governed by stochastic or chaotic processes. We […]

[…] and the debate between stochastic and chaotic processes in determining market behaviour (“Are Financial Markets Governed by Stochastic or Chaotic Processes?“). These posts provided a framework for understanding markets as complex and unpredictable […]