MSFT and the price of feeling safe
The Market Wave Algorithm for Equities on the market's most trusted stock, beating buy and hold and dollar-cost averaging across three windows.
Microsoft: the stock investors own so they can stop thinking about the market. Over the last five years it made that impossible twice.
Why this case study, and why now
We have already shown what a rules-based algorithm does with volatile names and broad index ETFs. This study asks a harder question: what does it do with the stock everyone calls safe?
Microsoft is the position investors buy to stop worrying. Yet a holder over the last five years lived through two drawdowns of roughly 40%, years of dead money, and a recovery so compressed that missing one week in August 2026 meant missing most of it.
We ran the Market Wave Algorithm for Equities on MSFT across three windows. Same stock, same algorithm, three different jobs.
The whipsaw year
Strategy+33.2%
- Buy and hold
- +5.9%
- DCA
- +2.7%
The two year grind
Strategy+68.4%
- Buy and hold
- +20.1%
- DCA
- +4.7%
The full cycle
Strategy+111.5%
- Buy and hold
- +49.7%
- DCA
- +21.3%
Every run starts from the same hypothetical $100,000 in cash and zero shares. The algorithm beat both benchmarks in all three windows.
Backtested results. Past performance does not guarantee future results. For educational purposes only.
The subject
Microsoft is the second largest company in the world by most measures: Windows, Office, Azure, gaming, and now one of the two or three central positions in the AI buildout. It sits in nearly every index fund, every retirement account, every "quality" portfolio. It is the definition of a stock you are supposed to be able to ignore.
Here is what ignoring it actually felt like:
- November 2021MSFT prints an all-time high near $349 as the pandemic software trade peaks
- November 2022The rate shock drags it to roughly $213, a drawdown of about 39%
- 2023 through mid 2024The AI rally carries it back, reaching about $468 by July 2024
- August 5, 2024The global carry-trade unwind knocks it to roughly $385 in days
- July 2025A new all-time high near $555 on Azure strength
- January to June 2026A grinding slide from about $484 to a one-year low near $345, driven by a $190 billion capex guide, doubts about Copilot monetization, and FTC scrutiny of its cloud and AI bundling
- July 30 to August 3, 2026Blowout results, Azure past $100 billion in annual revenue, and the largest single-day gain in the company's history. About 25% in three trading days, back near $500
Here is the catch that matters for this study: the market's most trusted stock delivered its five-year return in a handful of violent weeks, separated by long stretches that punished anyone who looked at their statement. Safety in the name never translated into safety in the experience.
The investor's dilemma
MSFT sets up three specific failure modes.
The safety trap
Because Microsoft is "safe," almost nobody has a plan for it. No exit rules, no re-entry rules, no cash discipline. The position that needs the least attention gets none, and then a 39% drawdown arrives with no playbook.
The round-trip ride
Twice in five years, holders watched enormous gains evaporate: the 2021 high round-tripped through 2022, and the 2025 high gave back 38% by June 2026. Riding a round trip in a volatile name feels like bad luck. Riding one in Microsoft feels like betrayal, and that feeling is what forces bad decisions at lows.
The capitulation candle
The one-year low printed in late June 2026, five weeks before the largest single-day gain in company history. Whoever finally gave up in June sold almost the exact bottom and then watched the entire recovery happen in three days. Nobody times that re-entry emotionally.
Dollar-cost averaging deserves honest credit here: it removes the timing decision and it kept buying through the 2026 lows. But its structure has a cost. It buys the same slice at a euphoric top and a generational bottom, it never sells, it never takes a profit, and after nearly five years on MSFT it had realized exactly nothing. Discipline that only ever points one direction is half a system.
A rules-based algorithm restores the other half: predefined exits, banked profits, and cash raised without a committee meeting in your head.
The algorithm
The Market Wave Algorithm for Equities is a rules-based algorithm that:
Rides confirmed moves
It scales in on weakness by rule, at levels set in advance, not when a headline makes the dip feel safe.
Exits on rules, not nerves
It takes profit into strength at predefined thresholds, converting paper gains into realized cash on the way up.
Steps aside in chop
Buffers and thresholds keep it from churning when the market goes nowhere.
The algorithm is configurable, and this study uses that on purpose. The three windows run three different recipes: a tight, fast configuration hunting the 2026 whipsaw, a wider configuration built to work a two-year range, and a patient weekly configuration with a trend exit riding the full cycle. Same rules discipline, three different jobs.
Backtested results. Past performance does not guarantee future results. For educational purposes only.
“The whipsaw year”
January 1, 2026 to August 11, 2026 · Daily chart
What the window contained: the worst stretch for MSFT in years. Down from roughly $484 to a one-year low near $345, a failed spring rally, a second slump into summer, then the record three-day surge in early August back to about $500.
Buy and hold survived it: +5.9%, entirely thanks to the August rally. Monthly dollar-cost averaging returned +2.7%.
The strategy ran a tight recipe that bought the declines in slices and sold the recoveries by rule.
+33.2%
- Buy and hold +5.9%, outperformance +27.3 percentage points
- DCA +2.7%, outperformance +30.5 percentage points
- Ending value
- $133,204
- Realized profit, banked in cash
- $16,401
- Unrealized profit on open shares
- $16,803
- Shares held
- 122 at $363.16 avg cost
- Cash
- $72,095
Backtested results, daily chart, valued August 11, 2026. Past performance does not guarantee future results. For educational purposes only.
“The two year grind”
August 1, 2024 to August 11, 2026 · Daily chart
What the window contained: an entry days before the August 2024 carry-trade shakeout, a year of choppy recovery and range, the run to the July 2025 all-time high near $555, the entire 2026 drawdown, and the August 2026 recovery.
Buy and hold earned +20.1% over the two years, with a 38% drawdown in the middle of it. Monthly dollar-cost averaging returned +4.7%.
The strategy ran a wide-scope recipe sized for bigger swings: deeper entry threshold, larger buffers, patient exits.
+68.4%
- Buy and hold +20.1%, outperformance +48.3 percentage points
- DCA +4.7%, outperformance +63.7 percentage points
- Ending value
- $168,432
- Realized profit, banked in cash
- $33,022
- Unrealized profit on open shares
- $35,410
- Shares held
- 335 at $395.17 avg cost
Backtested results, daily chart, valued August 11, 2026. Past performance does not guarantee future results. For educational purposes only.
“The full cycle”
November 1, 2021 to August 11, 2026 · Weekly chart
What the window contained: an entry at the worst plausible moment, weeks from the 2021 all-time high. Then the entire 2022 rate-shock drawdown to roughly $213, the two-year AI recovery, the 2025 top near $555, the 2026 slide, and the August 2026 rebound. Two full round trips.
Buy and hold, having entered near the 2021 top, earned +49.7% in almost five years and spent long stretches deeply underwater. Monthly dollar-cost averaging returned +21.3%.
The strategy ran a patient weekly recipe with a trend-break exit, accumulating through 2022 and 2025 to 2026 weakness and selling strength by rule.
+111.5%
- Buy and hold +49.7%, outperformance +61.8 percentage points
- DCA +21.3%, outperformance +90.2 percentage points
- Ending value
- $211,501
- Realized profit, banked in cash
- $61,212
- Unrealized profit on open shares
- $50,289
- Shares held
- 415 at $381.76 avg cost
Backtested results, weekly chart, valued August 11, 2026. Past performance does not guarantee future results. For educational purposes only.
Side by side
| The whipsaw year | The two year grind | The full cycle | |
|---|---|---|---|
| Window start | Jan 1, 2026 | Aug 1, 2024 | Nov 1, 2021 |
| Chart | Daily | Daily | Weekly |
| Strategy return | +33.2% | +68.4% | +111.5% |
| Buy and hold return | +5.9% | +20.1% | +49.7% |
| vs buy and hold | +27.3 pts | +48.3 pts | +61.8 pts |
| DCA return | +2.7% | +4.7% | +21.3% |
| vs DCA | +30.5 pts | +63.7 pts | +90.2 pts |
| Ending value | $133,204 | $168,432 | $211,501 |
| Realized profit | $16,401 | $33,022 | $61,212 |
| Shares held | 122 | 335 | 415 |
| Avg cost basis | $363.16 | $395.17 | $381.76 |
All windows start from $100,000 in cash and zero shares.
Backtested results, valued August 11, 2026. Past performance does not guarantee future results. For educational purposes only.
Why the results differ
Three windows, three answers, one algorithm.
The whipsaw year asked: can rules make money in a stock that goes down, up, down, and up inside seven months?
Answer: a tight recipe turned each leg of the whipsaw into either an accumulation or a payday, finishing at +33.2% while the stock itself barely moved net of the noise.
The two year grind asked: can rules beat a real recovery, not just a drawdown?
Answer: yes, by working the position through the range instead of riding it, ending +48.3 points ahead of a buy and hold that did fine.
The full cycle asked: what happens when you enter at the top?
Answer: the algorithm turned the worst entry date in five years into +111.5% by doing its buying in 2022 and 2026, when a human holder was busy regretting.
Buy and hold and DCA are single-decision strategies. They decide once, then hope. The algorithm made hundreds of small decisions, each one pre-committed, none of them requiring courage on the day.
The edge is not prediction. It never saw the August rally coming. It was simply positioned, by rule, to be holding cheap shares when the rally arrived and to get paid as it happened.
Backtested results. Past performance does not guarantee future results. For educational purposes only.
The real lesson
- 01
"Safe" stocks need rules too
The name on the position does not protect you from a 39% drawdown. A plan does. MSFT punished exactly the investors who assumed it never would.
- 02
Realized beats unrealized
Across every window the algorithm converted rallies into banked cash. Paper gains round-trip. Realized profits do not.
- 03
The best entries never feel like entries
Every buy the algorithm made in 2022 and 2026 happened on days when buying felt wrong. Pre-committed rules are how those days become cost basis instead of regret.
The client decision is not whether Microsoft is a good company. It is what job you want done on the position you already believe in: white-knuckle it and hope you never blink at a low, or hand the position a set of rules that buys weakness, sells strength, and pays you along the way.
Backtested results. Past performance does not guarantee future results. For educational purposes only.
A path forward
Arch Public's algorithms run on your account, at your brokerage, under your control. Zero custody: your assets never leave your account. You can pause, adjust, or stop any strategy at any time.
Across three windows on the market's most trusted stock, the same discipline showed up three ways: a whipsaw turned into +33.2%, a grind turned into +68.4%, and the worst entry of the decade turned into +111.5%, with profits banked in cash the entire way. The stock everyone holds for peace of mind finally came with some.
You already trust the company. The question is whether your plan for it deserves the same trust.
Buy and hold, the full cycle
+49.7%
The Market Wave Algorithm for Equities, the full cycle
+111.5%
$61,212 banked as realized profit
The comparison that matters most is the middle rows. The algorithm's edge over buy and hold grows with the length and violence of the window, from +27.3 points in seven months to +61.8 points across the full cycle. The harder the stock was to hold, the more the rules were worth. And in every window, part of the return is already realized cash, not a screenshot of a brokerage balance.