THE GOD OF ODDS CALCULATIONS

No part of this book maybe transmitted in any form of by any means electronical, mechanical, photocopying, recording, or otherwise the prior written permission of the Author. Table of Contents Dedication Acknowledgement Preface Introduction CHAPTER ONE DansilverRDS Theory CHAPTER TWO Common Types of Betting Algorithm CHAPTER THREE Prefessor, Gambling Rivalry and Voodoo Pastor CHAPTER FOUR Professor Family War at Court Online CHAPTER FIVE Y,Y Result or Dansilver Algorithm CHAPTER SIX Professor's Vision Concerning His Country Nigeria And 2027 General Election DEDICATION This book is dedicated to sport bettors and investors. ACKNOWLEDGEMENT This book would not be complete without first acknowledging The Holy Spirit, the King of Kings, for divine inspiration, wisdom, and guidance in building this forecasting model and for the successful conversion of this model into a book. From the days of trials, losses, and errors to the final breakthrough, His grace kept me going. Secondly, I owe a deep debt of gratitude to my family members for their unwavering support during the difficult time of building this forecasting model to this final stage. To a mother without compromise, Mrs. Nwayieze Oduah - thank you for your prayers, discipline, and love. To another silent bettor and brother, Mr. Chibueze Obasi, and Mr. Chidiebere Obasi aka Eb Jackson + OJ Comedy, thank you for standing by me when the model failed and when it succeeded. I am very grateful to an amiable mother, Mrs. Ugochi Ngozi, and the entire family members of the Ashionu family for their invisible yet powerful roles during the time of building this forecasting model. Your encouragement behind the scenes meant everything. I also want to appreciate other engineers and friends I may not mention by name for their support and role in my life and in this journey. You know yourselves, and I thank you. Moreover, my casino friends and fans all over Owerri city - this journey is ours together. Your conversations, challenges, and belief in the system pushed me to refine it. Finally, this work would not be complete without expressing my profound gratitude to many barristers and lawyers whose quotes, words, and articles made this book great and legally grounded. Number one; Confidence Aribibia, Ogam Solicitors, Emmanuella Chinenye Ogwale, and many more barristers whose insights I cannot forget. Thank you all. Introduction CHAPTER ONE DANSILVER-RDS THEORY DansilverRDS theory calculate and estimate amount of money wager on each single bet with Return On Investment (ROI). This formula helps to determine strategics of profits. Formula for DANSILVER THEORY BS/O-M SAMPLES OF CAPITAL STAKE LEVELS CAPITAL POTENTIAL WINNING PROFIT MARGIN $7,000. $10,000. $3,000. -100 $11,000. $16,000. $5,000. -164 $20,000. $30,000. $9,000. -300 $70,000. $100,000. $30,000. -1000 $100,000. $150,000. $40,000. -150 $200,000. $300,000. $90,000 -300 $300,000 $450,000. $140,000 -450 $500,000. $740,000. $240,000. -800 $1,000,000. $1,400,000. $1,400,000. -1,400 40%/100 × Capital= Stake Profit COMMON TYPES OF BETTING ALGORITHMS Every algorithm requires three things to be successful: data, probability, and mathematics. Here are a few types of algorithms listed below that are commonly used in sports forecasting. 1. Reinforcement Learning: These are adaptive models that re-evaluate strategies over time based on forecasting outcomes of data. The model learns from its mistakes and improves. Most statisticians rarely use it because it can easily slip off a small mistake in reward calculation can cause a big error in prediction. It is powerful but unstable if not handled well. 2. Monte Carlo Simulation: This type of algorithm estimates probability distributions for game outcomes through several simulated scenarios, often thousands or millions of them. It asks "what if we played this game 10,000 times?" Most statisticians use this algorithm for forecasting sports outcomes because it helps to see all possible outcomes, not just one. 3. Logistic Regression: This type of algorithm is used to estimate probability distributions for game outcomes, especially win or lose scenarios. It helps statistician model creators to determine the win, lose, or draw probabilities of any prediction. In my first year of creating a model, I used this algorithm but it wasn't successful. After researching and studying regression analysis and correlation analysis deeply, I had to change my algorithm, and the result was excellent. Logistic regression is good for simple binary outcomes, but not for complex multi-outcome forecasting. 4. Neural Networks:This type of statistical model is commonly used to analyze complex relationships between multiple variables and large datasets. It works like a human brain, with layers of nodes that learn patterns. It is very powerful for big data, player tracking data, and in-play betting models, but it requires a lot of data and computing power and can easily overfit if not tuned properly. NOTE: sportbook also use advanced analytics which increases market efficiency, but no algorithm can guarantee 100% accuracy or eliminate gambling risk completely. The accuracy of every model depends on the quality of data and the type of model used, although some models suffer from overfitting due to incomplete or inaccurate data. A good algorithm with bad data will still give you bad predictions. WHAT IS OVERFITING? According to Sigma Computing, overfitting in forecasting models is defined as "when a model learns the noise or random fluctuations in the training data rather than the underlying patterns, resulting in poor performance on new and unseen data in a statistical model." It can also be defined as a modelling error that occurs when a function corresponds too closely to a particular set of data. It is not learning, it is memorizing. Furthermore, it is the stage when the model learns the random fluctuations and noise in the data without following the true underlying patterns. This can result in inaccurate predictions and serious misallocation of resources. This leads to a situation where the model performs exceptionally well on the training data but performs poorly on new and unrecorded data. SigmaComputing.com elaborates that "overfitting occurs when an algorithm fits too closely to its training data, resulting in a model that can't make accurate predictions or conclusions." The model performs well on known data but struggles when faced with anything new. It can predict perfectly in-sample while completely missing the point when applied elsewhere. It is like a student who memorizes the answers to past questions but fails the real exam because the questions were rephrased. We have three types of fitting that affect model building: 1. Overfitting: The model is too complex. It learns too much, including noise. High accuracy on training data, low accuracy on test data. 2. Underfitting: The model is too simple. It learns too little and misses the underlying pattern. Low accuracy on both training and test data. 3. Appropriate Fitting (Good Fit): The balanced model. It learns the true pattern, ignores the noise, and performs well on both training and new data. In short, overfitting is when your model knows the training data too well to be useful in the real world. FACTORS THAT CONTRIBUTE TO OVERFITTING Generally, overfitting performs well on training data but fails to follow general patterns when faced with new data. Overfitting is primarily caused by excessive model complexity, insufficient or poorly processed data, and having many features but a small number of examples. 1. Error in Predictions: Prediction becomes less accurate, which results in prediction mistakes for unseen or unrecorded data. The model is confident, but wrong. It has memorized the past instead of learning how to predict the future. 2. Complication in Model Loading: When you load inaccurate data into the model, the model learns the errors caused by mistakes when imputing the data. This complication makes the model hard to understand and even harder to fix. It is not true that every overfitting case is a result of insufficient data; an error in input can also cause an overfitted model. Whenever the model or dataset contains an error, most statisticians call it noisy data because it slips away from the designed pattern. 3. Irrelevant Features: Irrelevant features that don't follow the pattern increase the noisy dataset. They enhance complexity in the model. Let's say you want to build a detection model that identifies counterfeit currency, but you train the model with irrelevant information like images of old currency notes that are still accepted. Such a model will perform poorly because it cannot generate accurate predictions. Your model should contain only images of all real currencies currently in circulation and images of all counterfeit currencies when loading the data. If you add irrelevant features, the model fails to predict outcomes on new or unrecorded data. HOW TO PREVENT OVERFITTING 1. Data Volume Increase: Increasing data volume is the most common method to prevent overfitting because it retains the most common features and simplifies the model, such as image retention, number tracking, and much more. Increasing data volume prevents variance from rising in your model to the point where it no longer accurately identifies patterns and trends in new data. 2. Pause Input Data: When data inputs contain errors, the best strategy is to pause loading input data, especially when validation performance stops improving. Check and recheck your dataset before you continue loading data. If you don't pause, the model will cease to capture the core patterns and nuances of the data. 3. Training vs. Final Testing: Get two separate spreadsheets for training and testing. To know the estimated performance level, split your dataset into two - one for training and one for cross-validation testing. Cross-check that the model learns from the training sheet and then test its knowledge on the final testing sheet. 4. Availability of Excess Data: The more data you have, the easier it is to load inputs into the model and filter out irrelevant features. In my first model, I tagged these irrelevant features as "Fake Gold, Iron and Chameleon" because they are noisy data that change color and deceive the model. Accurate data always follows a pattern when inputted into a database. So, sufficient data is the secret to preventing overfitting because it helps you quickly signal wrong input data. 5. Complex Model: A complex model will require careful tuning, while a simple model is less likely to overfit. My first forecasting model, which was a regression analysis model, was overfitting which caused me to keep losing money. It was a simple model that was too complex for the underlying data structure. I switched from a regression analysis model to a multinomial logit model after several trials and errors. It helped me to reduce my overfitting problem. The right model reduces overfitting while the wrong model increases the chance of overfitting. The best model creators make research on the type of model that is compatible with their project. 6. Data Augmentation:This increases the amount of data by slightly changing previously existing data. Image Data Augmentation like rotation, scaling, flipping, and cropping are used in deep learning computer models. For instance, assuming you are building a model that detects people who wear tattoos, your model must retain the image of temporary, permanent, even homemade and local tattoos. If there is overfitting, it may detect temporary and permanent tattoo bearers but be unable to detect local tattoo bearers. 7. Text Data Augmentation: Paraphrasing, synonym replacement, and contextual embeddings can enhance natural language processing tasks. A model that embraces a broader set of features and variability can better approximate the true underlying distribution in the data. 8. Remove Irrelevant and Add Relevant Features: This sounds like addition and subtraction of irrelevant features to track input data in a stored model. This is a manual process in training a model to improve its generalizability. Irrelevant features can add noise and make the model overly sensitive, especially to the training data. Remove variables and characteristics that don't contribute to the prediction pattern. In my model, those variables are ranked in their level with alphabets to indicate their values from A to E, such as A as Master, B as Professional, C as Expert, D as Novice, and E as Try Your Luck. WHAT IS UNDERFITTING? There are many definitions of underfitting, and it has a synonymous relationship to overfitting, but