Starting Free or Pro
What to expect when you start MI.
Start with the amount you choose, understand the new decisions, and measure what your own account actually does. Your first months can look different from the portfolio in the newsletter. This guide shows both the established model’s historical results and what happened in gradual-start research.
1. Similar holdings usually take months.
In this historical research, half the gradual Free starts reached eight out of ten matching names in 98 days—about three months. For Pro, it was 112 days—about four months. Around 9 in 10 starts in either tier had reached that milestone by six months.
Hypothetical MI 2.2 research · 1,200 overlapping weekly starts, January 2001–December 2023 · outcomes through December 2025 · corrected Pro source, November 15, 2026. “Matching” requires the same direction and original model entry for at least 80% of reference names, over four consecutive weekly checks, while the reference has at least 10% gross exposure. This is first attainment; matching can later break.
| Tier | 6 months | 1 year | 2 years |
|---|---|---|---|
| Free | 88% | 98% | 100% |
| Pro | 93% | 97% | 100% |
Matching names does not mean matching sizes. In Free, about 86% of starts reached the separate size milestone by six months. In Pro, only 13% did so; about 47% had reached it by two years. There was no typical Pro arrival time within the measured two years.
The size milestone requires at least 80% overlap at comparable position weights and gross exposure within 15 percentage points for four weekly checks. These are research definitions, not a claim that any account becomes permanently “steady state.”
2. A typical completed trade lasts weeks, with a long tail.
| History | Median | Mean | Completed cases |
|---|---|---|---|
| Free buy-to-sell trades | 21 days | 38 days | 3,059 |
| Pro long-position episodes | 30 days | 47 days | 4,597 |
Half the completed Free trades lasted three weeks or less; about 6 in 10 ended within 30 days. Pro’s revised long-position reconstruction has a median of about a month. A smaller number of long holds pulls the average above the median.
Pro’s corrected figures reconstruct long-position episodes from dated model targets; they are not original broker trade tickets. The Free figures pair reconstructed buys and sells. Positions still open at the endpoint are excluded. These different records describe historical duration, not an exit schedule or the exact current script.
3. Today’s holdings can be older than a typical completed trade.
A position held for six months appears in many more newsletters than one held for a week. That is why today’s holdings can look old even when many completed trades were short. A trade’s total life, its age today and its remaining life are different measurements. None tells you when a current holding must exit.
4. Making money and beating an index are different results.
In the gradual-start research, the six-month median return was +3.9% for Free and +2.2% for Pro. About 3 in 4 starts ended above their initial value; about 1 in 4 lost money. Only around 1 in 3 beat fully invested QQQ over those same six months.
The median deepest decline during those windows was 6.0% for Free and 3.1% for Pro. A positive ending return can include a painful fall along the way. Those median declines are not worst cases or loss limits.
See returns, losses and SPY/QQQ comparisons at six months, one year and two years →
Not customer returns or an exact backtest of today’s Sheets. MI costs: 0.06% per fill; Pro also models 3% annual short carry. Subscription fees, taxes and added market impact are excluded. Selected/reconstructed history and unverified historical borrowing limit the findings. Historical frequencies are not future odds.
5. Choose the comparison before you see the winner.
If the alternative was putting the same amount into SPY or QQQ on day one, keep that comparison. If you would have held part in cash, show that alternative too. There is no single “normal investor” allocation.
Over two years, Free beat an illustrative half-SPY, half-interest-earning-cash portfolio in 3 out of 4 historical starts. Corrected Pro did so in about 2 out of 3. Against fully invested QQQ, both gradual starts won about 4 out of 10. The full tables keep these alternatives together.
The cash mix begins at 50/50, then drifts without rebalancing. Cash interest is modeled from lagged Treasury-bill rates; MI replay cash earns zero. It is an illustrative alternative, not an exposure-matched portfolio or a claim about investor behavior. Asset allocation varies with circumstances: Investor.gov explains the concept.
6. A precise hot start gets you into the running portfolio.
An exact copy starts with the same holdings, weights and cash as the running account. With the same future orders, prices and costs, the accounts follow the same path from your start date.
The optional Pro Sheets preview makes a one-time match to eligible research targets. Target weights can differ from the weights in an existing entry-and-exit account. Compare the startup replays against SPY and QQQ.
The separate historical allocation backtest follows ongoing target changes and recovery rules. Its results describe that allocation policy.
The one-time target match puts money to work sooner. The daily-start study shows both the return and drawdown tradeoff. See the matched-date results.
The historical target-match replay allowed skipped orders. Today’s Sheet can block the preview or stop when it encounters an eligibility or account limit. The earlier Free study follows new additions at the issue’s stated size.
7. A smaller allocation is different from cherry-picking signals.
You can set aside a chosen portion for MI and track it separately, including its unused cash. Also track your whole portfolio. Selecting only favorite signals, skipping exits or changing participation after a strong run produces a different strategy with no established result in this study.
Account size, minimum orders and short availability can affect which Pro actions are possible. Paper practice can help you learn the workflow, but its result is not a real-money return. A later funded account has its own start date.
8. Larger positions trade more upside for more downside.
Our sizing experiments found return-versus-loss tradeoffs, not a universal best size. The optional 5.5% Free idea remains separate research; it is not a blanket instruction to enlarge every position. Use the entry percentage explicitly printed in your current issue and stay within your chosen account limits.
A green risk state is a model condition, not a safety guarantee. For manual Free sizing, the entry calculator converts your assigned amount and the issue’s entry percentage into dollars.
A useful review has more than one score.
After the first few issues, check setup, amounts and fills. At three or six months, separate cash exposure, missed actions, strategy losses and benchmark lag. At one and two years, keep the original start date and compare cumulative returns, deepest declines, fees and the effort required. These are review checkpoints, not a required waiting period or a promise that losses recover.
If you missed an issue, check what actually happened first. Never place an order twice because you did not read the email.
Study scope and limitations
Hypothetical MI 2.2 research · 1,200 overlapping weekly starts, January 2001–December 2023 · outcomes through December 2025 · corrected Pro source, November 15, 2026. Each horizon uses the same 1,200 start dates, from January 4, 2001 through December 28, 2023; the latest endpoint is December 26, 2025. The windows overlap. They are not 1,200 independent trials, subscribers or future probability estimates.
The Free study uses the frozen standard sizing. Pro uses the corrected historical core and an illustrative $100,000 account that skips unsupported or over-limit orders. Current Sheets can pause instead. Neither the gradual replay nor the one-time basket experiment establishes exact subscriber or current automation performance.
MI replay costs are 0.06% per fill and, for Pro, 3% annual modeled short carry calculated over trading sessions. Unused MI cash earns zero interest. Subscription fees, investor taxes and added market impact are excluded. Historical short availability, delivery and fills are unverified. SPY and QQQ use distribution-adjusted total returns, including embedded fund expenses, without an added initial brokerage commission.
The histories include selected and reconstructed inputs, incomplete historical universe and delisted-company coverage, and rules developed using history already examined. They are not an untouched validation sample. Pro’s corrected early source replaces the older startup figures; the research label MI 2.2 does not claim the delivered product has changed version. The older archived studies on Results remain separately labeled.
Download the summary, definitions and source hashes · Download the comparison table · View the result tables