Starting Index
transparent model intelligence
Account
Evidence

S&P 500 momentum strategy backtest and validation.

The live model family is published forward from its public inception. This page shows historical research backtest: point-in-time constituent membership, SPY comparison, validation checks, and the annual return view used by the research workflow to inspect behavior.

The three curves assume zero interest while defensive, with modeled trading costs retained. The validation figures use the same cash assumption, recalculated from saved paths.

Start Free AccessSee Method
Backtest WindowMay 1, 1996-Jun 24, 2026

Research data only. Subscriber allocations are published separately. The live platform record and profile record are tracked forward from publication records.

Research Equity Curve
Core, Balanced, and historical Growth versus SPY. Annual bars follow the selected model
Core
Balanced
Growth
SPY
$10.0K$20.0K$50.0K$100K$200K$500K$1.0M$2.0M$5.0M$10M200020052010201520202025Growth of $10,000 ยท log scaleCoreBalancedGrowthAnnual return ยท Core+239%-239%1997 Core +21.4%, excess -9.7%1997 SPY +31.1%, excess -9.7%1998 Core +35.2%, excess +8.8%1998 SPY +26.4%, excess +8.8%1999 Core +78.0%, excess +58.6%1999 SPY +19.4%, excess +58.6%2000 Core +67.3%, excess +77.1%2000 SPY -9.8%, excess +77.1%2001 Core +0.0%, excess +11.3%2001 SPY -11.3%, excess +11.3%2002 Core +0.0%, excess +23.6%2002 SPY -23.6%, excess +23.6%2003 Core +57.9%, excess +35.7%2003 SPY +22.2%, excess +35.7%2004 Core +26.7%, excess +18.0%2004 SPY +8.7%, excess +18.0%2005 Core +11.2%, excess +7.7%2005 SPY +3.5%, excess +7.7%2006 Core -2.4%, excess -14.2%2006 SPY +11.8%, excess -14.2%2007 Core +17.6%, excess +14.1%2007 SPY +3.4%, excess +14.1%2008 Core -14.7%, excess +23.1%2008 SPY -37.7%, excess +23.1%2009 Core +36.2%, excess +16.3%2009 SPY +19.9%, excess +16.3%2010 Core +12.1%, excess +1.1%2010 SPY +11.0%, excess +1.1%2011 Core -6.8%, excess -5.6%2011 SPY -1.2%, excess -5.6%2012 Core +27.3%, excess +15.6%2012 SPY +11.7%, excess +15.6%2013 Core +46.7%, excess +20.3%2013 SPY +26.4%, excess +20.3%2014 Core +19.6%, excess +7.3%2014 SPY +12.4%, excess +7.3%2015 Core -1.6%, excess -0.8%2015 SPY -0.8%, excess -0.8%2016 Core +18.1%, excess +6.9%2016 SPY +11.2%, excess +6.9%2017 Core +27.7%, excess +9.3%2017 SPY +18.5%, excess +9.3%2018 Core -17.0%, excess -10.0%2018 SPY -7.0%, excess -10.0%2019 Core +24.0%, excess -4.7%2019 SPY +28.7%, excess -4.7%2020 Core +17.3%, excess +2.2%2020 SPY +15.1%, excess +2.2%2021 Core +23.3%, excess -5.5%2021 SPY +28.8%, excess -5.5%2022 Core +9.8%, excess +29.7%2022 SPY -19.9%, excess +29.7%2023 Core +22.7%, excess -2.1%2023 SPY +24.8%, excess -2.1%2024 Core +25.3%, excess +1.3%2024 SPY +24.0%, excess +1.3%2025 Core +45.3%, excess +28.7%2025 SPY +16.6%, excess +28.7%2026 Core +116.6%, excess +109.3%2026 SPY +7.3%, excess +109.3%
Validation
We checked how historical Core results varied across model settings, time periods, and difficult markets.
Parameter Tests99

Saved setting variations recalculated with zero defensive interest and recorded stock costs.

Time-Window Checks594

Six saved time windows per parameter path; no new signals or training are generated.

Stress Windows495

Five saved difficult-market windows per parameter path, using the same cash assumption.

Parameter Median19.82%

The middle return across saved setting variations with zero-interest cash.

Time-Window Median18.89%

The middle return across the six saved time windows with zero-interest cash.

Stress Median-5.64%

Middle hard-market result versus SPY at -31.54%. The stock paths and modeled costs are retained; defensive interest is zero.

Summary Read
Core quick read from the zero-interest cash research curve
Return Engine22.42% CAGR

Growth multiple: Core 445.4x ยท SPY 11.2x.

Risk Shape19.30%

Difference between SPY's worst drawdown and Core's worst drawdown. Positive is better.

Sharpe Ratio0.91

Core's daily return / sample volatility, annualized over 252 sessions with zero risk-free rate.

Win Rate72%

Core outperformed SPY in 21 of 29 full calendar years (1997-2025), measured first to last recorded session.

The retained run records point-in-time S&P 500 membership. 17.6% of days were handled defensively, with zero interest in the curves above.

Research Standard
What the backtest is designed to protect against
  • Point-in-time S&P 500 membership, not today's constituents projected backward.
  • Decisions use data available at the rebalance date.
  • Delisted and replaced symbols are included when historical data is available.
  • SPY is used as the broad-market benchmark for the same dates.
  • The ranking formula is proprietary. The testing standard is intentionally visible.
  • Annual bars compare the first and last recorded observation within each year; 1996 and 2026 are partial.
  • Validation figures are recalculated from saved historical model paths with zero defensive interest. Time windows are slices of those paths; these checks do not generate new signals or perform independent training and testing.
Known Limits
The parts a serious investor should still care about
  • Backtests are historical research, not live account results.
  • Active membership-price coverage is 93.32%; the run's coverage check failed.
  • Drawdowns are real and can be emotionally difficult even when long-term returns are strong.
  • Future market structure can differ from the research period.
  • Live publication and audit records are the source of truth once a model period has been published.
  • Historical Growth uses rank-linear weights; the currently inspected Growth preset uses equal weights.
  • Original source and input provenance, stock and SPY dividend treatment, and backtest/publication signal-rule differences remain under review.

Research evidence is provided for due diligence and context. It is not investment advice and does not guarantee future results.