Before a model is offered to subscribers it is evaluated on more than a decade of historical data, under rules designed to prevent the most common backtesting errors. This page describes that process and shows the results of the stress tests and simulations.
Every fundamental figure is used only from the date its SEC filing became public. If a company restated results later, the backtest still sees what an investor saw at the time. Nothing in a selection decision uses information from after the decision date.
The universe is limited to US listed companies with at least $1B market value and $5M average daily volume, so every position could genuinely have been traded at the tested size. Financials and other hard to model sectors are excluded.
The models follow a fixed schedule and a fixed ruleset. There is no discretionary override in the backtest, and the same code that produced the historical results produces the live signals.
Revenue growth and stability are verified at every rebalance from the filing record, including a check against each company's own revenue history that catches restatements and divestitures a simple growth figure would miss.
Every put option in the hedged versions is priced with a standard option model, including volatility skew, at the moment it would have been bought. Premiums are paid in full in the results; payoffs occur only when the market actually fell through the strikes.
Backtested results are not live trading: they benefit from hindsight in research design, option prices are model based rather than quoted, and transaction costs beyond the bid ask of liquid names are not modelled. Roughly one year of the record is live operation.
How each version behaved through the three sharpest declines in the period, measured peak to trough. The S&P 500 column is the same window for comparison.
| Episode | Vortex | Vortex Shield | Diversified Shield | S&P 500 |
|---|
Peak to trough returns over each window, from the same equity curves shown on the performance page. In fast crashes the books initially fall with the market; the hedge's convex payoff shows up in deeper and longer declines, most clearly the 2022 bear market, and in the calendar year outcomes on the performance page (Vortex Shield: 2020 +64%, 2022 +16%).
A block bootstrap simulation: one year paths are assembled from randomly drawn 20 day blocks of each strategy's own daily returns, preserving short term clustering. The table shows the distribution of 5,000 simulated years.
| Strategy | Median year | 5th percentile | 95th percentile | Chance of a losing year | Median max drawdown |
|---|
Simulation resamples the historical return distribution; it cannot produce market conditions the sample never contained, and it is not a forecast. It shows how sensitive each strategy's outcome is to the ordering of returns.
Every calendar year, every version, against the S&P 500, with drawdowns and worst years disclosed in full.
Review the performance page