Engineering Proposal: Signal Processing

Yasin Alkaifi 

CUNY City College

 ENGL 21007

Professor Julia Brown

05/03/2023

Contents:

Summary ………………………………………………………………………………………… 3

Introduction ……………………………………………………………………………………… 4 Description………………………………………………………………………………………. 5 Conclusion………………………………………………………………………………………. 9 

References………………………………………………………………………………………. 10

Summary of the proposal: 

Signal processing, a technique that analyzes time-varying electrical signals conveying information, is widely applicable across diverse industries such as medicine, telecommunications, the oil sector, and more. Within this domain, linear and nonlinear systems are studied, with a majority of cases involving nonlinear systems, which are characterized by unpredictable values. Signal processing can be categorized into two main types: analog signal processing (ASP) and digital signal processing (DSP). While both types possess their unique advantages, contemporary technology has exhibited a preference for DSP. 

This proposal aims to delve deeper into the adaptability of DSP in the context of various cryptocurrencies, which display behavior similar to that of conventional stocks. As these digital assets constantly exhibit fluctuating values that can be challenging for humans to process and predict, DSP emerges as a valuable tool for analyzing and forecasting cryptocurrency values. The Fast Fourier Transformation algorithm can be applied to these analyses, enabling users to generate profit from their cryptocurrency investments. 

Integrating DSP into the crypto and stock markets is a relatively straightforward process that does not require advanced experience or qualifications, making it an accessible tool for a wide range of users. Additionally, the implementation costs are minimal, as DSP is an online open-source resource that can be accessed and utilized at no charge. This cost-effective nature of DSP allows for broader adoption and implementation across the crypto and stock market sectors, potentially leading to more informed investment decisions and better overall market performance.

Introduction:

In today’s world, technology plays a significant role in enhancing our daily lives by offering convenience and efficiency, such as machines that prepare our morning coffee. Beyond these visible applications, technology also operates behind the scenes in the form of software programs, which are designed to perform complex, time-consuming tasks that may be nearly impossible for humans to accomplish. In this proposal, I aim to harness modern technology and employ specific algorithms to predict cryptocurrency and stock market trends by analyzing signals. 

Predicting market fluctuations is crucial, as accurate forecasts can potentially lead to substantial profits within a matter of weeks. Signal processing, which involves using and manipulating signals like sound and images to extract information, has the potential to significantly impact a person’s financial standing in a short period. In my project, I will specifically utilize digital signal processing (DSP) to work with algorithms that predict future trends in cryptocurrency and stock market graphs. 

While it is impossible to achieve 100% accuracy in these predictions, DSP currently offers the most precise technology available for this purpose. It has already been successfully implemented in various applications, such as predicting tidal waves and earthquakes. One of the key advantages of using DSP is its cost-effectiveness, as the algorithms employed are based on open-source software, making them freely accessible to the public. I will implement this software to put DSP technology into action. 

As I am well-versed in DSP and its implementation in the stock and crypto markets, there will be no labor costs involved. However, hiring a professional to execute this task may range from $20,000 to $35,000, depending on the software engineer’s experience. Despite the seemingly high cost, the investment is worthwhile, as implementing DSP in the crypto and stock markets has the potential to generate millions, if not billions, in revenue. 

Description: 

Signal processing offers a multitude of techniques for analyzing time-varying signals, which can be particularly useful in predicting cryptocurrency price movements. One such technique is Fourier Analysis, which involves representing mathematical functions as sums of simpler sinusoidal or exponential functions. This approach simplifies the analysis of complex functions, especially when exact analytical solutions are not available, by approximating a solution using these more manageable components. 

There are two primary methods for expressing functions as sums of sinusoidal or exponential terms: 1) Fourier Series, suitable for periodic functions, and 2) Fourier Transform, more appropriate for phenomena that are not necessarily periodic. In my solution, I will apply the Fourier Transform because, while cryptocurrency prices may not be inherently periodic, I assume they display well-behaved patterns. 

The Fourier Transform algorithm works by converting a time-based function into a frequency-based function. This conversion allows us to identify the amplitude at each frequency within the original time-based function, which is particularly valuable in deciphering the most significant terms affecting the function. By focusing on these terms and filtering out the noise, I can gain a clearer understanding of the underlying patterns within cryptocurrency prices, leading to more accurate predictions. 

In practical terms, the Fast Fourier Transform (FFT) algorithm is used to handle large volumes of data and perform the transform on our behalf using a computer. This computational approach streamlines the analysis process, making it more efficient and precise. As a result, investors can make better-informed decisions based on the insights gained from this advanced analysis of cryptocurrency price movements. 

Furthermore, the Fourier Transform can be combined with other signal processing techniques and machine learning algorithms to further enhance the accuracy and reliability of the predictions. By incorporating additional layers of analysis, I can account for various factors that influence price movements, such as market sentiment, news events, and macroeconomic indicators. This comprehensive approach to cryptocurrency price prediction will provide investors with a more robust understanding of the market dynamics and trends, ultimately leading to more successful investment strategies. 

While reducing noise in cryptocurrency prices is helpful, it alone does not offer significant insights for price prediction. To address this, my secondary solution involves applying the Fourier Transform to other relevant metrics, such as the U.S. Gross Domestic Product (GDP) over the past decade, and comparing these metrics with crypto prices to identify potential correlations. I would iterate through different metrics until a suitable pairing is found, enabling us to exploit market inefficiencies. 

In fact, combining two or more metrics after applying the Fourier Transform can increase the confidence in our predictions. However, a potential drawback of this method is that cryptocurrency prices are not solely influenced by physical phenomena like weather patterns but are also affected by human emotions, unforeseen events, and public trends, making them highly unpredictable. To mitigate this, I recommend that an actuarial team analyze our models and develop a risk management strategy to determine the appropriate level of funds to be deployed at any given time, minimizing potential losses. 

It is crucial to note that the Fourier Transform is not a novel solution for this problem, as it has been and continues to be used in trading strategies. my advantage lies in how I apply it, making it essential to maintain the proprietary and confidential nature of our approach.

The total net worth of the stock market is 95 trillion dollars, and accurately forecasting stock prices can lead to substantial profits. I have collected recent stock market data from the internet and analyzed it using DSP, with Python programming language for implementation. By representing stocks as signals, I was able to predict stock prices. As seen in Figure 1, our prediction (green line) closely aligns with the actual stock market performance (orange line). 

Despite its effectiveness, DSP is not infallible and can pose risks when investing large sums of money at once, potentially confusing the algorithm and lowering its prediction accuracy. The DSP stock predictor should be used responsibly and efficiently for optimal results. Moreover, having multiple users may disrupt predictions, so maintaining the confidentiality of the method is crucial to its effectiveness. 

By combining Fourier Transform with other relevant metrics and risk management strategies, our approach to cryptocurrency and stock price prediction offers a more comprehensive and reliable solution for investors, enabling them to make well-informed decisions and maximize their profits.

Figure 1

Figure 2

Conclusion:

In conclusion, our data-driven, analytical approach to predicting cryptocurrency prices helps me minimize and, in some cases, eliminate the human bias associated with intuitive decision-making. I strongly advocate for the use of signal processing to support my trading decisions, as it is versatile and adaptable to various market conditions, working effectively with both linear and nonlinear systems. I specifically recommend the Fourier Transform for its ability to filter out noise, enabling me to extract more meaningful information and identify patterns within cryptocurrency prices. Daily fluctuations in these prices make traditional pattern recognition challenging, but the Fourier Transform offers a powerful solution to this issue. Furthermore, my proposed algorithm and strategy can be implemented using current computer systems without incurring additional operational expenses, thanks to their existing processing capabilities. This cost-effective aspect of my solution is a highly desirable feature in engineering and financial applications. By employing signal processing, particularly the Fourier Transform, my approach provides a more reliable and efficient way to predict cryptocurrency price movements, empowering me to make better-informed decisions in the ever-evolving and often unpredictable world of digital assets.

References

Binnoy, V. (1970, January 01). Fourier transforms intuitively explained with examples & analogies. Retrieved May 4, 2023, from http://visualizingmathsandphysics.blogspot.com/2015/06/fourier-transforms-intuitively.html

Flis, B. (2022, January 11). Fourier transform on Bitcoin prices. Retrieved May 4, 2023, from https://blog.devgenius.io/fourier-transform-on-bitcoin-prices-aaae397ef28

Jelinek, S. (2019, January 01). Forecasting cryptocurrency time series using fuzzy transform, fourier transform and fuzzy inference system. Retrieved May 4, 2023, from https://www.academia.edu/66080489/Forecasting_Cryptocurrency_Time_Series_Using_Fuzzy_Transform_Fourier_Transform_and_Fuzzy_Inference_System

Rodriguez, T. (2018, November 03). Bitcoin case study: Applying basic digital signal processing into financial data. Retrieved May 4, 2023, from https://medium.com/drill/btc-case-study-applying-basic-digital-signal-processing-into-financial-data-ec34cd47c77b

Wei, W., Urquhart, A., Urom, C., Tiwari, A., Symitsi, E., Selgin, G., . . . Blume, L. (2020, December 05). Returns and volume: Frequency connectedness in cryptocurrency markets. Retrieved May 4, 2023, from https://www.sciencedirect.com/science/article/abs/pii/S0264999320312499