Validate Prediction Market EV System

**Seeking a Software Lead and a Statistical Lead to complete and validate an existing prediction-market system — total budget USD 800** We are looking for two complementary specialists, or a small firm providing two named leads: 1. **Software Lead:** responsible for software completion, correctness and operational reliability. 2. **Statistical / Quantitative Lead:** responsible for the statistical methods, probability calibration and expected-value calculations implemented in the software. The two specialists must collaborate and critically review each other’s work. Individual applications for either role and joint proposals are welcome. **GOAL** The system forecasts YouTube video view counts and evaluates prediction-market contracts that settle according to view-count ranges at different horizons. Our objective is to identify executable betting opportunities with positive net expected value and maximise that value within agreed capital and risk limits. We want a defensible calculation showing whether expected proceeds exceed the amount staked plus all relevant costs. Before betting, we want to understand the estimated advantage, its uncertainty, its assumptions and the circumstances in which the system should abstain. We seek measurable evidence of an advantage, not a guarantee of future profits. **PROJECT STATUS — NOT A BUILD FROM SCRATCH** The software is being completed. A substantial codebase, data-collection infrastructure, historical datasets, forecasting models, betting strategies, tests and documentation already exist. Our provisional estimates for initial scoping are: - **Software: approximately 80% implemented.** The principal modules and workflows exist. Remaining work includes reviewing correctness, completing integrations, fixing defects and validating the full workflow. - **Statistical components: approximately 60% implemented.** Forecasting, uncertainty estimation, range probabilities and expected-value calculations exist. Remaining work includes auditing these calculations, improving calibration and establishing credible evidence of positive net expected value. These percentages estimate implementation progress, not proven correctness or profitability. Both leads must confirm or revise them after inspecting the project. We will provide the code, available data, documentation and a structured roadmap of prompts and implementation tasks. You may revise this roadmap, remove unsupported branches and propose alternatives where evidence supports an improvement in net expected value. **SYSTEM MODULES** 1. Data collection. 2. Data preparation and optimisation. 3. Data analysis and probabilistic forecasting. 4. Betting strategy and execution. Several models and strategies already exist. These must be compared consistently, and additional strategies may be proposed. Every material software or statistical decision should support a correct calculation or reliable realisation of net expected value. **SOFTWARE LEAD — RESPONSIBILITIES** - Audit the implementation and completion roadmap. - Complete outstanding work and correct defects. - Verify data integrity, timestamps and matching between videos, markets, settlement horizons, ranges and YES/NO contracts. - Ensure backtests use only information available at each historical decision time. - Implement the agreed statistical methods faithfully. - Verify executable prices, order-book depth, liquidity, position sizing, execution behaviour and risk controls. - Make calculations and decisions traceable and reproducible. - Deliver appropriate automated tests, documentation and operating instructions. **STATISTICAL LEAD — RESPONSIBILITIES** - Audit forecast bias, uncertainty estimates and range probabilities. - Assess and improve calibration, potentially using quantile regression, conformal prediction and probability recalibration. - Evaluate reliability, interval coverage, Brier score and log-loss against meaningful baselines. - Use confidence intervals, hypothesis tests and p-values, Bayesian inference or other justified methods to quantify evidence and uncertainty. - Account for dependence between observations, contracts and horizons. Frequent observations from one video must not be treated as independent events. - Address overfitting, multiple strategy comparisons, selection bias and repeated examination of accumulating results. - Quantify how probability uncertainty affects estimated net expected value. - Recommend which models, branches and strategies to retain, revise, disable or reject. - Define when evidence is insufficient and the system should not bet. Methods must be justified by the data and the decision being made. We do not require every listed technique to be used mechanically. **MAIN HYPOTHESIS AND VALIDATION** The principal hypothesis is that the selected strategy has positive expected net profit: expected proceeds exceed the stake and all relevant costs. We expect a coherent theoretical calculation and credible empirical validation. Historical profitability alone is insufficient to establish a future advantage. Before final evaluation, both leads must agree in writing on: - Hypotheses, metrics, baselines and acceptance thresholds. - Data eligibility, independent evaluation units and required sample size. - Training, calibration, strategy selection and untouched evaluation procedures. - Treatment of fees, spread, slippage, liquidity, latency and unfilled or partially filled orders. - Capital limits, exposure across related positions, stopping rules and no-bet conditions. Backtesting should use chronological out-of-sample or walk-forward evaluation. Results must show uncertainty, sensitivity to assumptions and concentration of returns across videos. The analysis must distinguish estimated future expected value from realised historical profit and explain the assumptions required for the advantage to persist. **RECIPROCAL VALIDATION** We specifically seek two people capable of challenging and validating each other’s work within