Expert Quant to Take an Intraday Options Strategy to Net Profitability (Nautilus, Optuna)

We are an algo trading firm. Please read this carefully before applying, because it will save us both time. We are not an individual and we are not a client shopping for basic technical help. We already have data scientists on staff. We already have algo traders on staff. We already have all the coding firepower we need in house. So if your pitch is that you can write Python, that you are a data scientist, or that you can build a backtest, we are not the right fit. If that is your USP, please do not apply. We are looking for one thing: a proven expert who has personally been there and done it. Someone who has built systematic strategies, taken them to real profitability, and can lift our work to the next analytical level. To be precise about the depth we mean: we are not looking for someone to run Optuna and hand back a marginally better number. We already do that. We want someone who goes into the strategy itself, the entry and exit conditions, the gating logic, and the statistical behavior of every signal, and brings genuine analytical depth to redesign and improve it. Parameter search alone is not the job. THE PROJECT We have a put-only intraday options strategy on US equities, fully built in Python on an event driven backtest engine in the NautilusTrader style, with an Optuna optimization pipeline. It shows a real edge in a clean backtest. We need it taken to net profitability in the real world, and proven out of sample. This project is your test. We run a busy pipeline with 30 to 50 new clients a month, so there is a long runway of ongoing work for the right person. But we do not hand that over on a promise. You prove yourself on this one strategy first. THE HONEST STATE The strategy is profitable in an idealized backtest. That backtest is frictionless, no commissions and no slippage, and it models the underlying move rather than the real option premium. Your job is to close the gap to live trading. We do not want someone who waves a frictionless equity curve and calls it done. We want someone who knows exactly why frictionless results lie and who has closed that gap before. THE STATISTICAL TARGET Success means a configuration that hits all of this on out of sample data, with realistic commission and slippage modeling on the actual traded vehicle: - Net profit factor of 1.3 or better after costs, with 1.5 plus as the goal - Positive net expectancy per trade after costs - Annualized Sharpe of 1.5 or better, net - Maximum drawdown of 12 percent or less - Robust across at least three walk forward out of sample windows, not carried by a few symbols or days - At least 300 out of sample trades for statistical significance - Zero look ahead bias, fully reproducible We independently re verify every number on data you have not seen. If it does not reproduce on our side, it does not count. YOUR HONEST READ ON THE UPSIDE Early on, looking closely at this strategy, we want your honest expert assessment of what real returns it can realistically reach. We are after real returns, not inflated ones. If your honest view is more conservative than the target above, we want to hear that, with the reasoning. A grounded verdict is worth far more to us than optimistic noise. WHAT WE PROVIDE - A private GitHub repository with the full codebase: event engine, Optuna runner, indicators, scoring and lane logic, and exit engine - A two year, minute level dataset across roughly 230 US equities - Current backtest results plus a detailed per trade analytics workbook that exposes every gate value, score, and indicator reading at the moment each trade fired - Clear success metrics, so there is no ambiguity about what winning looks like REQUIRED SKILLS - Hands on NautilusTrader, or a directly comparable event driven backtest and live framework - Optuna or equivalent, used correctly with walk forward validation - Strong Python, pandas, and numpy - Real experience modeling commissions and slippage for intraday US equities and options - Market microstructure and options mechanics - A disciplined defense against overfitting, curve fitting, and look ahead bias Nice to have: IBKR or live execution, options Greeks and volatility modeling, market regime analysis. PROOF OF WORK You must show real strategies you have built, with their actual statistics. For at least one strategy we want to see the profit factor, the Sharpe, the win rate, the maximum drawdown, the out of sample period, and whether it traded live, on paper, or only in backtest. Show us the artifacts: equity curves, tearsheets, walk forward reports, a GitHub history, or a verifiable track record. Redacted is fine. Personal projects are fine. Here is a hard filter, and we mean it. If your only answer is that everything is under NDA and you can show nothing at all, this role is not for you. Every serious practitioner can demonstrate their method and results in some form. Proposals that lead with an NDA as the reason to show nothing will not be considered. PLEASE ANSWER IN YOUR PROPOSAL 1. Describe one strategy you took to profitability. What were the net metrics after costs, and how did you confirm it was not overfit? 2. How do you model commissions and slippage for intraday options, and how much of a frictionless edge typically survives? 3. A strategy shows a profit factor of 1.68 in a frictionless backtest. What are the first three things you check before trusting it? 4. How do you set up an Optuna walk forward so the chosen configuration is not curve fit to the validation window? ENGAGEMENT This is an hourly engagement, but we are result driven and very practical about it. We are not interested in paying for hours that do not move the strategy forward. The real reward here is for results. Take this strategy past the finish line and the pay reflects that, along with a long runway of further work. We start with a short paid diagnostic so both sides can confirm fit, including your honest read on the achievable upside. Generic copy and paste proposals will be ignored. Tell us in plain terms how you would attack the gap between backtest and live on this specific strategy.