Scientists popularised the notion of the “Butterfly Effect” as a result of a question Edward Lorenz posed: “Could a butterfly flapping its wings in Brazil trigger a tornado in Texas?”. This theory supposes that even tiny actions can lead to momentous outcomes.
The Butterfly Effect is rooted in Chaos Theory, which elucidates how a seemingly insignificant change to one state of a non-linear deterministic system can have sprawling ramifications on subsequent states.
The ‘Black Monday’ catastrophe in Hong Kong, which witnessed the stock market index crashing and losses mounting precipitously, highlights how widespread the effects of economic instability can be. The fallout was experienced worldwide prior to identifying the root cause of the crash.
The global financial crisis in 2007 was set off by the bankruptcy of a minor segment in the US mortgage market. To resuscitate the world economy, bailouts, government support and various other forms of aid were necessary.
The financial industry is known for its intricate nature, given its association with economic and investor outlook. One destabilising factor can create a notable ripple effect. What initiates such market volatility and how can scientific and engineering innovations be leveraged to diminish its impact?
Unsteadiness in the Worldwide Financial System
During a recent academic lecture, Andrew Haldane, the Executive Director of Financial Stability at the Bank of England, pointed out that the financial system has grown more complex while experiencing a drop in diversification. What consequences might this entail?
Financial systems are comparable to edifices in that the sturdier they are, the more robust the foundations must be to manage the burden. The more complicated the system, the more extensive and intricate the underlying framework should be (with columns and beams distributing the weight across the structure).
There has been a proposal that a building could sustain itself with fewer foundations if the weight was distributed uniformly using a framework. Nevertheless, if any foundation were to fail, the consequences would be felt throughout the building and potentially bring about its downfall.
A financial system with lower diversification offers fewer safeguards to counter unexpected fluctuations and shocks. It is crucial to guarantee that the framework is soundly constructed and does not rely on unstable foundations.
Working Mechanism of Random Forests:
Before comprehending the forest as a whole, we must examine the individual trees it comprises of. Decision Trees are an effective and popular Machine Learning technique utilised for classification tasks. It is depicted in the shape of a flowchart, where every node indicates a solitary attribute test.
It is not as complicated as it may sound initially. Decision trees are so prevalent that we frequently use them without even realising.
Constructing a decision tree can be beneficial when selecting a taco spot without having to go out of your way. Begin by making a list of all the nearby food establishments you know of and segregate them into two categories: those that provide tacos and those that do not.
Next, we arranged the taco-serving establishments in order of their proximity to us, classifying them as either ‘close’ or ‘distant’. We then separated them further into categories of ‘economical’ and ‘beyond our budget’, yielding a list of reasonably-priced taco joints nearby.
Committees are crucial as even the most well-informed individual can err. The presence of several specialists can serve as a counterbalance, ensuring that the team is led towards the correct decision.
Random Forests are a machine learning method which leverage several decision trees to produce a prediction. This simulates a voting system, where the outcome of the model is the forecast with the highest number of votes. Essentially, Random Forests are virtual committees that make decisions founded on the combined weight of opinions.
For Random Forest to function efficiently, it is imperative that the predictions of each tree are free of the predictions of the other trees. This can be explained in simple terms as each tree delivering a distinctive and impartial viewpoint.
This algorithm is an advanced application for tackling intricate problems that may include an infinite number of data points, and for generating predictions on highly unpredictable systems – such as the financial challenges we often confront.
Random Forests in Finance
Random Forests have showcased superior efficiency in various applications, encompassing forecasting stock prices, performing qualitative stock analysis, as well as determining option pricing and credit spreads. It is crucial to keep these two factors in mind.
Traditional predictive techniques employ linear regression; an adept method for data analysis when the correlation being studied is linear (i.e. a change in Variable A will result in a corresponding change in Variable B).
Studies indicate that height and weight are linearly correlated; however, this is not always true. In general, individuals with greater height tend to have a higher weight, but it is feasible for someone’s weight to increase without a paralleling change in height.
It can be inferred that the usefulness of a linear regression model for predicting height from weight is limited. The same principle holds true for stock prices, where the impact of specific variables becomes insignificant beyond a certain threshold, and the model’s predictive power reaches a plateau.
Random Forests advocate for researchers to contemplate categorization techniques when investigating their research inquiries. By formulating questions like “Will value X alter?” instead of “If X increases, how much will Y increase?”, the same investigation can be approached in a more intuitive manner. This change in viewpoint can often yield a significant impact on our comprehension of the problem.
Given adequate data, Random Forests can offer an understanding of whether a minor shift in the local market may have significant repercussions on the global economy. Furthermore, thanks to the assistance of the Internet of Things, Artificial Intelligence, Cloud Computing, and Data Mining, examining financial data has become more effortless than ever before.
It is broadly acknowledged that Random Forests are notably vulnerable to even the slightest biases; nevertheless, the trained model’s quality depends solely on the quality of the data supplied. If faulty data is fed into the model, the resulting outcomes will be untrustworthy.
While Random Forests may not necessarily lead to significant disruptions in the financial sector, they remain a potent resource that can be leveraged to tackle multiple challenges and augment our ability to predict market trends.