Pro starting options and benchmark results
See how starting from cash, copying a running account, and matching research targets compared with SPY and QQQ.
Pro hot-start comparison · Historical allocation backtest · Year-by-year returns
Starting Pro under the newsletter rules
What does a hot start match?
An exact copy starts caught up. Copy the running account’s positions, weights and cash at the same time, then follow the same orders at the same prices and costs. That gives you the same portfolio path from that point. We verified this on all 6,158 eligible start dates in the archive.
- Cash start. Start empty and follow the new entries and exits.
- Exact copy. Inherit the already-running entry-and-exit account at the joining close. This test gives the same results as the running account over each window.
- Target match. Attempt a one-time copy of the dated research targets at the next open, then follow the same entries and exits. This is the closest replay to the optional Sheets preview. Target weights can differ from the sizes in an already-running account.
| Benchmark | Cash start | Exact copy | Target match |
|---|---|---|---|
| SPY · 6 months | 40% | 51% | 54% |
| SPY · 1 year | 45% | 51% | 53% |
| SPY · 2 years | 46% | 51% | 54% |
| QQQ · 6 months | 37% | 42% | 46% |
| QQQ · 1 year | 40% | 45% | 46% |
| QQQ · 2 years | 41% | 45% | 45% |
In the six-month target-match replay, the median return rose from 2.2% to 5.0% compared with cash starts. The median decline from a peak deepened from 3.1% to 6.6%.
See how to preview Match current allocations in Pro Sheets → · Download comparison data (CSV)
Hypothetical research. Each row uses the same 5,784 overlapping daily starts from January 2001 to December 2023, with outcomes through 2025. The cash and target-match paths start with a modeled $100,000 and skip unsupported or over-limit orders. The exact-copy test carries forward the running account at the matched closing marks and excludes the cost of buying the initial basket. Subsequent orders model 0.06% per fill and 3% annual short carry. The current Sheets apply their own eligibility and account checks. Research history is partly reconstructed and historical short availability is unverified. Subscription fees, taxes and additional market impact are excluded.
Earlier weekly cash-start study, including Free and fund-and-cash mixes
Hypothetical MI 2.2 research on 1,200 overlapping weekly starts from January 2001 through December 2023, with outcomes through 2025. Pro uses a modeled $100,000 account and skips unsupported or over-limit orders. Modeled costs are 0.06% per fill and 3% annual short carry for Pro. Subscription fees, taxes and additional market impact are excluded.
| Measure | 6 months | 1 year | 2 years |
|---|---|---|---|
| Median total return | +2.2% | +7.3% | +20.7% |
| Ended below starting value | 25% | 16% | 10% |
| Beat fully invested SPY | 40% | 45% | 46% |
| Beat fully invested QQQ | 37% | 40% | 41% |
| Median deepest decline | -3.1% | -6.3% | -7.9% |
| Worst observed decline | -20.0% | -26.7% | -27.5% |
| Measure | 6 months | 1 year | 2 years |
|---|---|---|---|
| Median total return | +3.9% | +9.4% | +20.5% |
| Ended below starting value | 28% | 22% | 3% |
| Beat fully invested SPY | 43% | 47% | 54% |
| Beat fully invested QQQ | 36% | 36% | 40% |
| Median deepest decline | -6.0% | -8.6% | -12.1% |
| Worst observed decline | -15.7% | -16.1% | -16.8% |
Read this study’s scope and limitations · Download the weekly study summary
What if the alternative is partly invested, with cash?
Compare against the alternative you would actually have chosen. This example starts with half in the fund and half in cash earning modeled Treasury-bill-rate interest. The mix then drifts; it is not rebalanced. It is an illustration, not a claim about what a typical investor owns or a match for MI’s risk.
| Measure | 6 months | 1 year | 2 years |
|---|---|---|---|
| Beat half SPY / half cash | 63% | 71% | 75% |
| Beat half QQQ / half cash | 52% | 59% | 64% |
| Measure | 6 months | 1 year | 2 years |
|---|---|---|---|
| Beat half SPY / half cash | 46% | 54% | 68% |
| Beat half QQQ / half cash | 41% | 52% | 63% |
At two years, Free beat half SPY / half cash in 3 out of 4 starts; Pro did so in about 2 out of 3. Against fully invested QQQ, both were closer to 4 out of 10. The comparator changes the answer.
Cash interest uses lagged archived 3-month Treasury-bill rates as a yield-accrual proxy, not an actual Treasury fund return. MI replay cash earns zero interest. No extra initial fund brokerage fee is charged. These mixes do not match MI’s stock, sector, short or gross-exposure risks.
See the fully invested fund outcomes on those same dates
| Fund / measure | 6 months | 1 year | 2 years |
|---|---|---|---|
| SPY · median return | +6.2% | +13.0% | +23.4% |
| SPY · ended at a loss | 27% | 21% | 14% |
| SPY · median deepest decline | -7.5% | -10.1% | -18.6% |
| QQQ · median return | +9.0% | +16.0% | +31.5% |
| QQQ · ended at a loss | 27% | 19% | 15% |
| QQQ · median deepest decline | -10.9% | -15.0% | -17.3% |
SPY is the S&P 500 fund comparison; QQQ is the Nasdaq-100 fund comparison. Both use distribution-adjusted returns including embedded fund expenses. Fully invested funds have different exposure from a strategy building from cash. These remain useful alternatives; holding cash does not excuse every performance gap.
Historical allocation backtest
Hypothetical MI 2.2 allocation backtest. The Pro model applied ongoing target changes and recovery rules. The Pro startup comparison above uses the newsletter’s entry-and-exit rules. SPY and QQQ show historical total returns.
| Portfolio | Compounded annually | Largest decline |
|---|---|---|
| MI 2.2 Pro model | 23.12% | −22.56% |
| SPY | 7.98% | −55.19% |
| QQQ | 8.30% | −82.96% |
| MI 2.2 Free model | 13.98% | −25.40% |
Pro finished ahead of both funds in 16 of 26 years and trailed both in 3. Free finished ahead of both in 10 years. Largest decline measures the drop from a previous peak, including losses within a year.
The models already held positions and cash when the comparison began. Every row uses the December 31, 1999 close through the December 31, 2025 close. Fund returns include reinvested distributions.
Pro modeled short positions and could exceed 100% gross exposure. The models include trading and applicable short borrow, financing and recovery costs. Subscription fees, taxes and additional market impact are excluded. The research uses selected and partly reconstructed stock history, with incomplete delisting coverage and no untouched holdout.
Today’s newsletter uses an entry, hold and exit rule. The model backtest uses historical allocation rules that can produce different holdings, sizes and returns.
Pro’s year-by-year results · Free’s year-by-year results
Established-model costs, dates and limits
Hypothetical MI 2.2 established-model research. Not live-account or customer returns, and not an exact backtest of today’s newsletter or Sheets. Includes inherited modeled trading, borrowing, financing and recovery costs. Subscription fees, taxes and additional market impact are excluded. Selected and reconstructed stock history, incomplete delisted-company coverage and rules developed on examined history limit the evidence. No untouched holdout. Pro includes short exposure and can exceed 100% gross exposure. These are different risks from SPY or QQQ.
Free retains 0.06% ordinary fill costs and the archived recovery accounting. Pro retains component-level costs, 0.12% on half the sum of absolute allocation-weight changes, 3% annual modeled short carry, 5% annual financing and 0.12% recovery/rotation transfer costs. Eligible Pro cash receives modeled Treasury-bill carry; gradual-start replay cash earns zero. These inherited return curves are not a fully reconciled historical subscriber brokerage ledger. Actual short availability, fills and cash-sweep returns are unverified.
The long-run table starts in 2000, the first complete calendar year with both fund histories, and includes the dot-com decline. Starting-date choice materially affects comparisons. SPY and QQQ are distribution-adjusted fund total returns with embedded fund expenses; no earlier index bridge is used in these tables. CAGR is the compounded annual growth rate, not a return earned every year. Deepest decline is the largest daily fall from a prior peak during the stated period; it is not a future loss limit.
Download definitions and source hashes · Long-run table · Six-month, one-year and two-year table
How did the allocation backtest compare in the earlier weekly study?
This historical allocation-model comparison uses the 1,200 overlapping dates from the earlier weekly study, January 4, 2001 through December 28, 2023, with outcomes through December 26, 2025. Returns are cumulative. Each row applies its own allocation and execution rules.
| Measure | 6 months | 1 year | 2 years |
|---|---|---|---|
| Median total return | +5.1% | +11.0% | +22.5% |
| Ended below starting value | 28% | 22% | 4% |
| Beat SPY | 54% | 54% | 57% |
| Beat QQQ | 42% | 42% | 42% |
| Median deepest decline | -7.6% | -10.3% | -12.5% |
| Worst observed decline | -15.7% | -16.8% | -16.8% |
| Measure | 6 months | 1 year | 2 years |
|---|---|---|---|
| Median total return | +8.5% | +18.1% | +40.3% |
| Ended below starting value | 19% | 10% | 1% |
| Beat SPY | 68% | 74% | 92% |
| Beat QQQ | 58% | 62% | 69% |
| Median deepest decline | -9.0% | -11.2% | -14.3% |
| Worst observed decline | -19.2% | -22.6% | -22.6% |
In plain language: over two years, established Pro beat SPY about 9 in 10 times and QQQ about 7 in 10. Established Free beat SPY about 6 in 10 and QQQ about 4 in 10. A strong full-period result is not the same as winning most windows against every benchmark.
Year-by-year returns
How did the Pro allocation backtest vary by year?
Hypothetical MI 2.2 Pro established allocation model. January 4, 1982 close through December 31, 2025 close. Pro used short positions and could exceed 100% gross exposure. Before the ETFs launched the benchmarks use index history. The early Nasdaq price series excludes dividends.
| Series | Compounded annually | Middle half of years | Worst to best year |
|---|---|---|---|
| MI 2.2 Pro model | +20.06% | +5.33% to +30.47% | −7.87% to +190.10% |
| S&P 500 / SPY | +12.11% | +5.03% to +26.22% | −36.79% to +38.05% |
| Nasdaq / QQQ | +14.37% | +6.19% to +33.66% | −41.73% to +98.68% |
The high of +190.10% came in 2020. The worst calendar-year return was −7.87% in 1990. Pro’s largest peak-to-trough decline across the full study was 24.86%, including losses within a year.
See the comparison using only full ETF years
| Series | Compounded annually | Middle half of years | Worst to best year |
|---|---|---|---|
| MI 2.2 Pro model | +23.12% | +6.71% to +33.79% | −6.99% to +190.10% |
| S&P 500 / SPY | +7.98% | +1.40% to +24.09% | −36.79% to +32.31% |
| Nasdaq / QQQ | +8.30% | +2.04% to +31.35% | −41.73% to +54.86% |
Measured from the December 31, 1999 close through the December 31, 2025 close. This includes the dot-com decline. Pro beat both ETFs in 16 of 26 years, beat one in 7, and trailed both in 3.
Across all 44 year-end results, Pro beat both benchmarks in 20, beat one in 13, and trailed both in 11.
See every year, including the 11 years Pro trailed both
Years Pro trailed both. 1983, 1985, 1987, 1988, 1992, 1996, 1997, 1998, 2016, 2019, 2023.
| Year | Pro model | S&P 500 / SPY | Nasdaq / QQQ | Pro comparison |
|---|---|---|---|---|
| 1982 | +29.15% | +21.21% | +18.86% | Beat both |
| 1983 | +5.76% | +22.40% | +19.87% | Trailed both |
| 1984 | −1.79% | +6.10% | −11.31% | Beat one |
| 1985 | +18.71% | +31.60% | +31.49% | Trailed both |
| 1986 | +15.78% | +18.60% | +7.77% | Beat one |
| 1987 | +1.28% | +5.10% | +10.49% | Trailed both |
| 1988 | +9.74% | +16.57% | +13.54% | Trailed both |
| 1989 | +35.19% | +31.68% | +26.17% | Beat both |
| 1990 | −7.87% | −3.10% | −10.41% | Beat one |
| 1991 | +32.92% | +30.47% | +64.99% | Beat one |
| 1992 | −0.53% | +7.62% | +8.86% | Trailed both |
| 1993 | +15.64% | +9.62% | +10.57% | Beat both |
| 1994 | +0.86% | +0.40% | +1.50% | Beat one |
| 1995 | +45.77% | +38.05% | +42.54% | Beat both |
| 1996 | +9.54% | +22.50% | +42.54% | Trailed both |
| 1997 | +4.91% | +33.48% | +20.63% | Trailed both |
| 1998 | +19.60% | +28.69% | +85.31% | Trailed both |
| 1999 | +76.50% | +20.39% | +98.68% | Beat one |
| 2000 | +7.59% | −9.74% | −36.11% | Beat both |
| 2001 | +4.81% | −11.76% | −33.34% | Beat both |
| 2002 | −6.99% | −21.58% | −37.37% | Beat both |
| 2003 | +37.14% | +28.18% | +49.67% | Beat one |
| 2004 | +15.91% | +10.70% | +10.54% | Beat both |
| 2005 | +18.29% | +4.83% | +1.57% | Beat both |
| 2006 | +10.97% | +15.85% | +7.14% | Beat one |
| 2007 | +13.23% | +5.15% | +19.03% | Beat one |
| 2008 | +11.23% | −36.79% | −41.73% | Beat both |
| 2009 | +66.52% | +26.35% | +54.68% | Beat both |
| 2010 | +21.11% | +15.06% | +20.14% | Beat both |
| 2011 | +5.46% | +1.89% | +3.48% | Beat both |
| 2012 | +16.40% | +15.99% | +18.11% | Beat one |
| 2013 | +35.16% | +32.31% | +36.63% | Beat one |
| 2014 | +22.26% | +13.46% | +19.18% | Beat both |
| 2015 | +1.90% | +1.23% | +9.44% | Beat one |
| 2016 | +6.42% | +12.00% | +7.10% | Trailed both |
| 2017 | +38.13% | +21.71% | +32.66% | Beat both |
| 2018 | −3.80% | −4.57% | −0.13% | Beat one |
| 2019 | +28.12% | +31.22% | +38.96% | Trailed both |
| 2020 | +190.10% | +18.33% | +48.41% | Beat both |
| 2021 | +29.65% | +28.73% | +27.42% | Beat both |
| 2022 | −0.15% | −18.18% | −32.58% | Beat both |
| 2023 | +25.10% | +26.18% | +54.86% | Trailed both |
| 2024 | +44.51% | +24.89% | +25.58% | Beat both |
| 2025 | +57.87% | +17.72% | +20.77% | Beat both |
Pre-ETF history uses S&P 500 index total returns, Nasdaq Composite prices through 1985 and Nasdaq-100 prices from 1986, without Nasdaq dividends. SPY and QQQ thereafter include reinvested distributions and fund expenses. Pro includes modeled component costs, 0.12% on half the total absolute allocation-weight changes, 3% annual short carry, 5% annual financing and 0.12% recovery or rotation transfer costs when applicable. Subscription fees, taxes and additional market impact are excluded. The selected and reconstructed universe has incomplete delisting coverage and no untouched holdout. These are simulated model results. The current newsletter and accounts that start gradually or hold quantities between signals follow different portfolio paths. See the established-model risk and methodology tables.
How did Free’s annual returns vary?
Hypothetical MI 2.2 Free established-model backtest. January 4, 1982 close through December 31, 2025 close. The benchmark series use index history before the ETFs launched. The early Nasdaq price series excludes dividends.
| Series | Compounded annually | Middle half of years | Worst to best year |
|---|---|---|---|
| MI 2.2 Free model | +14.77% | +0.44% to +28.10% | −8.43% to +77.97% |
| S&P 500 / SPY | +12.11% | +5.03% to +26.22% | −36.79% to +38.05% |
| Nasdaq / QQQ | +14.37% | +6.19% to +33.66% | −41.73% to +98.68% |
Compounded annually shows how fast the balance grew. The middle half shows where the central 50% of annual returns fell. The worst-to-best range shows the extremes. The drawdown figures above measure losses from a peak, including within a year.
See the comparison using only full ETF years
| Series | Compounded annually | Middle half of years | Worst to best year |
|---|---|---|---|
| MI 2.2 Free model | +13.98% | −0.78% to +26.47% | −8.43% to +66.01% |
| S&P 500 / SPY | +7.98% | +1.40% to +24.09% | −36.79% to +32.31% |
| Nasdaq / QQQ | +8.30% | +2.04% to +31.35% | −41.73% to +54.86% |
Measured from the December 31, 1999 close through the December 31, 2025 close. This includes the dot-com decline. Free beat both ETFs in 10 of 26 years, beat one in four, and trailed both in 12.
Across all 44 year-end results, Free beat both benchmarks in 15, beat one in 10, and trailed both in 19.
See every year, including the 19 years Free trailed both
Years Free trailed both. 1983, 1985, 1987, 1992, 1996, 1997, 1998, 2003, 2004, 2006, 2010, 2011, 2012, 2013, 2015, 2016, 2018, 2019, 2023.
| Year | Free model | S&P 500 / SPY | Nasdaq / QQQ | Free comparison |
|---|---|---|---|---|
| 1982 | +29.44% | +21.21% | +18.86% | Beat both |
| 1983 | +6.76% | +22.40% | +19.87% | Trailed both |
| 1984 | −0.77% | +6.10% | −11.31% | Beat one |
| 1985 | +17.31% | +31.60% | +31.49% | Trailed both |
| 1986 | +15.40% | +18.60% | +7.77% | Beat one |
| 1987 | +2.18% | +5.10% | +10.49% | Trailed both |
| 1988 | +15.71% | +16.57% | +13.54% | Beat one |
| 1989 | +29.47% | +31.68% | +26.17% | Beat one |
| 1990 | −2.31% | −3.10% | −10.41% | Beat both |
| 1991 | +31.61% | +30.47% | +64.99% | Beat one |
| 1992 | +5.56% | +7.62% | +8.86% | Trailed both |
| 1993 | +13.53% | +9.62% | +10.57% | Beat both |
| 1994 | +5.42% | +0.40% | +1.50% | Beat both |
| 1995 | +44.00% | +38.05% | +42.54% | Beat both |
| 1996 | +4.18% | +22.50% | +42.54% | Trailed both |
| 1997 | +0.14% | +33.48% | +20.63% | Trailed both |
| 1998 | +15.67% | +28.69% | +85.31% | Trailed both |
| 1999 | +77.97% | +20.39% | +98.68% | Beat one |
| 2000 | +0.49% | −9.74% | −36.11% | Beat both |
| 2001 | −1.14% | −11.76% | −33.34% | Beat both |
| 2002 | −4.11% | −21.58% | −37.37% | Beat both |
| 2003 | +16.74% | +28.18% | +49.67% | Trailed both |
| 2004 | +8.32% | +10.70% | +10.54% | Trailed both |
| 2005 | +13.69% | +4.83% | +1.57% | Beat both |
| 2006 | −2.20% | +15.85% | +7.14% | Trailed both |
| 2007 | +13.52% | +5.15% | +19.03% | Beat one |
| 2008 | −4.33% | −36.79% | −41.73% | Beat both |
| 2009 | +37.50% | +26.35% | +54.68% | Beat one |
| 2010 | +14.50% | +15.06% | +20.14% | Trailed both |
| 2011 | −6.21% | +1.89% | +3.48% | Trailed both |
| 2012 | +11.66% | +15.99% | +18.11% | Trailed both |
| 2013 | +31.64% | +32.31% | +36.63% | Trailed both |
| 2014 | +21.55% | +13.46% | +19.18% | Beat both |
| 2015 | +0.30% | +1.23% | +9.44% | Trailed both |
| 2016 | +3.28% | +12.00% | +7.10% | Trailed both |
| 2017 | +29.88% | +21.71% | +32.66% | Beat one |
| 2018 | −8.43% | −4.57% | −0.13% | Trailed both |
| 2019 | +22.92% | +31.22% | +38.96% | Trailed both |
| 2020 | +66.01% | +18.33% | +48.41% | Beat both |
| 2021 | +27.66% | +28.73% | +27.42% | Beat one |
| 2022 | −4.92% | −18.18% | −32.58% | Beat both |
| 2023 | +21.82% | +26.18% | +54.86% | Trailed both |
| 2024 | +35.50% | +24.89% | +25.58% | Beat both |
| 2025 | +55.59% | +17.72% | +20.77% | Beat both |
The early S&P series uses index total returns. The early Nasdaq series uses Composite prices through 1985 and Nasdaq-100 prices from 1986, without dividends. After SPY and QQQ launched, the fund series include reinvested distributions and reflect fund expenses. Free can hold cash and uses selected, reconstructed stock history with no untouched holdout. The model includes 0.06% per ordinary fill and archived recovery-transfer costs. It excludes subscription fees, taxes and additional market impact. These are simulated model results. Subscriber results depend on when and how an account is invested.