Three Pre-Registered Bets: A One-Shot Out-of-Sample Test of Retail Strategy Candidates

A retail investor with public data and a 300,000 EUR budget can try hundreds of strategy variants cheaply; the hard part is knowing what, if anything, survives. This paper documents one complete pass through such a process, in enough detail to rebuild it. At least 51 distinct strategy specifications (98 logged backtest runs, plus 48 unlogged grid runs and early tuning that was not logged) were evaluated on data up to 1 March 2026, and three "bets" were frozen: a trend-following risk-parity ETF portfolio (A), a 50/50 mix of A with the S&P 500 (B), and a 70/30 mix of A with a low-volatility, high-quality sleeve of 20 US stocks (C). The frozen engine source was hashed, and each bet was run exactly once on 165 trading days of unseen data (2 March to 23 October 2026), with point-in-time signals, next-open execution and a full transaction-cost model. The annualised Sharpe ratios were 0.08 (A), 1.18 (B) and 0.43 (C), against 1.68 for the S&P 500 in euros; maximum drawdowns were between -4.0% and -5.2%, against -6.6% for the index. With a standard error between 1.2 and 1.6 on a Sharpe ratio estimated from 165 days, none of this is distinguishable from zero, from the validation values, or from the benchmark. The Deflated Sharpe Ratio of the validation results is between 0.001 and 0.043 under every assumption about the number of trials considered here. The result is a null result in the strict sense: the one-shot test was too short to inform the question it was designed to answer. The value lies in the protocol, in the appendices that specify all three bets, the simulator and the data so that they can be rebuilt, in the list of pitfalls found before the freeze, and in a quantification of how much out-of-sample time such a test needs.Read next: All candidates were built with the protocol for leak-resistant point-in-time backtesting. The workflow behind the test is described in running an autonomous AI agent on trading research. Taking a winner live requires the requirements for an autonomous order engine.This preprint is part of an eight-paper series on retail equity research from public data by Christopher Helm, covering German and Nordic small caps, backtesting methodology, data sources and AI-assisted research. All papers in the series:How Wise Is the Crowd? Timing and Selection Skill of Retail Social-Trading PortfoliosA Leak-Resistant Point-in-Time Backtesting Protocol and Engine for Retail Equity StrategiesWhat Did Not Work: Negative Results on Event-Driven Signals in German and Nordic Small CapsFree, Point-in-Time Public Data for German and Nordic Small-Cap Research: Sources, Coverage and PitfallsRequirements for an Autonomous Retail Order Engine Running Unattended for a DecadeRunning an Autonomous AI Agent on Quantitative Trading Research: Protocol, Supervision and Failure ModesFrom Pre-Announcement to Verified Facts: Design of a Primary-Source Report-Analysis Funnel for Small-Cap EquitiesChristopher Helm is an entrepreneur, consultant, and researcher. More research and articles on his blog.