The Golden Ratio Dual Gate backtest begins in October 1998 and compounds at 16.2% a year through mid-November 2026. One date pair produces one attractive number.
An investor can start in any month. We replayed every eligible 5-year and 10-year start from the production backtest series, then ran the same windows with monthly contributions. The strategy did well across this history, while the spread between starts was large enough to change the experience completely.
The rules stayed frozen
The strategy checks its 2 gates daily. It takes 50% UPRO plus 10% each in SPMO, VBR, DBMF, GLD and TLT while both SPY and a TIPS canary are above their 200-day averages. Each gate has a band against churn: SPY switches on 1% above its average and off 1% below it, the TIPS canary at 0.1% either side. When either gate is off, it drops UPRO and holds 20% in each of the 5 diversifiers.
We preserved those rules and used the same NAV series the strategy page reports, so every figure below reconciles with what a subscriber sees there. The price builder uses real ETFs where available and documented proxy chains before fund inception. Managed futures used to set the start date; their history now reaches back to 1987. Since November 2026 the TIPS canary reaches back too, through a real TIPS mutual fund (PIMCO Real Return, PRTNX) placed before VIPSX, so the backtest starts in October 1998 instead of April 2008. Every figure below is recomputed on that longer history.
Every 5-year start
There were 277 completed monthly starts. The strategy's 5-year CAGR ranged from 0.2% to 32.9%, with a median of 17.5%. It beat SPY in 97.8% of lump-sum windows. The 6 it lost all started between December 2008 and April 2010, right at the bottom, when SPY's rebound was hard to beat. The weakest, from February 2004, made 0.2% a year while SPY lost 6.6% a year.
Monthly contributions changed the comparison only slightly. Golden Ratio beat SPY in 94.9% of 5-year DCA windows. A saver who began near a difficult switch sequence could still trail a plain index for the full 5 years.
| Horizon | Starts | CAGR range | Median CAGR | Beat SPY, lump | Beat SPY, DCA |
|---|---|---|---|---|---|
| 5 years | 277 | 0.2% to 32.9% | 17.5% | 97.8% | 94.9% |
| 10 years | 217 | 2.9% to 26.6% | 13.7% | 100.0% | 100.0% |

The 10-year result looks steadier against the index. Every completed start beat SPY for both lump sum and monthly DCA. The weakest strategy window, from June 1999, compounded at only 2.9% a year, still ahead of SPY's -2.2% over the same decade.
Those 217 starts overlap heavily. Neighboring starts share almost the entire decade, and the whole sample comes from one 28-year era. The 100% figure describes this historical record rather than 217 independent trials.
All rolling figures below are computed from the same daily series the strategy page reports, sampled at month-end, so they reconcile with the page up to tiny convention differences.
The drawdown view
The full daily backtest reached a -37.3% max drawdown. It came in 2020, from February 19 to March 18. Rolling 5-year month-end drawdowns ranged from -30.5% to -5.3%. The difference comes partly from sampling month-end NAV inside each rolling table, which can miss an intramonth low, and partly from start dates that sidestep the bad stretches entirely.
Start date also decides whether an investor meets the 2000 to 2002 bear market or the 2008 crash immediately, enters after them, or begins near the 2020 and 2022 switches. A headline drawdown taken from the full history tells the worst point. It says little about when a specific investor encountered it.
A favorable result with a short clock
The rolling table supports the strategy across the available history. It also puts the 16.2% headline in context. A real 5-year investor could have earned 0.2% or 32.9% a year from the same unchanged rules, and the difference between those two lives is nothing but the calendar.
The remaining uncertainty is time. The test now begins before the dot-com peak and ends during a strong US equity and gold period. Several sleeves are reconstructed before their funds launched: managed futures before 2019 rest on the BTOP50 index, UPRO before mid-2009 is a simulated 3x S&P 500, and the TIPS canary before mid-2000 is PRTNX rather than VIPSX. Another inflation cycle, a long managed-futures slump or a different equity leadership regime can widen the range beyond anything in this table.
The full rules and current signal are on Golden Ratio Dual Gate. The original construction, including the 2013 gate failure, is covered in the related article.
Related: The Golden Ratio Dual Gate, In Full.