Financial Analyst · FP&A · Equities & Options · Crypto · Tokenization · AI
Financial analyst focused on forecasting, budgeting, and variance analysis. Active in equities and options trading, with a strong interest in crypto, asset tokenization, and AI-driven financial tools. I build open source Python tools for finance work that has to tie out: OpenVariance for variance analysis and forecasting, and Form4Clean for SEC insider trading data.
| Area | What that looks like |
|---|---|
| FP&A | Budgeting, rolling forecasts, budget-vs-actual variance analysis, driver-based models |
| Equities & options | Fundamental and technical research, options pricing and Greeks, volatility, strategy backtesting |
| Crypto | Market structure, on-chain data, exchange APIs, DeFi mechanics |
| Tokenization | Tokenized stocks, bonds, stablecoins, and real-world assets (RWAs) |
| AI | LLM agents and MCP tools for financial analysis, forecasting, and research automation |
| Project | What it does |
|---|---|
| OpenVariance | Open source FP&A variance analysis. Explains why the numbers moved with driver bridges that reconcile to the cent, builds linked three-statement forecasts, pulls public-company data from SEC EDGAR, writes formula-driven Excel, and runs as an MCP server so AI agents can use it. Python. |
| Form4Clean | Clean, research-ready SEC Form 4 insider trading data. Downloads filings from EDGAR, classifies every transaction code, reconciles amendments without double-counting, runs 39 data quality checks, and gives point-in-time views with no look-ahead bias. Library, CLI, exports and a local dashboard. Python. |
| panther-tokenization-onboarding | Educational portfolio simulator for tokenized assets. Mean-variance allocation, a glide path with correlated stock and bond returns, and live checks of which issuer and chain can fill each order at its real size. TypeScript. |
| Panther Terminal | Market intelligence platform for tokenized securities and real-world assets. |
OpenVariance and Form4Clean are MIT licensed. Both run locally, use only free public data, and keep money math exact. The tokenization simulator is public but not licensed for reuse.
32 pull requests merged into 22 public repositories outside my account, as of October 22, 2026 (search). Mostly correctness fixes in the Python finance and crypto libraries I use.
| Project | Fix | What was wrong |
|---|---|---|
| yfinance | Make download() thread-safe |
Concurrent calls shared module-level result dicts, so one thread could overwrite another's data and return a corrupted DataFrame. Added a lock and a two-thread test. |
| vectorbt | Record active drawdowns when a series ends in NaN | The numba kernel only saved a running drawdown on the last row, and skipped that check on NaN, so any asset whose data ended early reported no drawdown at all. |
| viem | Keep an empty version in the EIP-712 domain |
A truthiness check dropped version: "" from the type hash, so signatures for contracts like OpenZeppelin's EIP712("Name", "") failed to verify. Result now matches ethers v6. |
| solana-py | Decode every signer in M-of-N multisigs | The decoders read the last m accounts as signers, but m is the signature threshold, so an M-of-N multisig with M < N lost its first N - M signers on round trip. |
| ffn | Ignore leading missing prices in monthly and yearly stats | An asset that started trading later got empty months before its first price, which zeroed its first month, skewed YTD, and inflated the win-rate denominators. Single-series and grouped stats disagreed on the same data. |
Others: a UK bank holiday calendar fix in FinancePy, ccxt tickers missing bid and ask, a news cache in yfinance that ignored its arguments, and smaller fixes in aiohttp, bokeh, CPython, cryptofeed, FinanceToolkit, nostr-tools and skfolio.
- Building OpenVariance and Form4Clean
- Contributing fixes to the Python finance and crypto libraries I use, including yfinance, FinancePy, ffn, bt and solana-py
- Researching tokenized securities and RWA market structure
Open to conversations about FP&A, markets, and on-chain finance. Say hi at contact@wavebyrd.com.