Skip to the case study
All case studies
Contents
The results
The subject: Meta
The investor's dilemma
The algorithm
“The flat-year harvest”
“The capex whipsaw”
“The AI race accumulator”
Side by side
Why the results differ
The real lesson
A path forward
Book a demo
Case studies
META
Case study: Arch Public for stocks
Volatility Harvesting for Meta
The Market Wave Algorithm for Equities against Meta in three windows. Buy and hold and DCA beaten in all three.
Meta: the AI spender that fell to $88, set a $796 record, lost 30% on its own spending plans, then rose 27% in a single month.
META
Stock
Market Wave Algorithm for Equities
Valued October 23, 2026
8 min read
Why this case study, and why now
Meta moves. It lost 26% in a day in February 2022, bottomed near $88 that November, and set a record $796 in August 2025. Then it fell 30% by March 2026 as its AI bill climbed, and rose 27% in September 2026 alone. Holders felt every swing. Buy and hold turned none of them into cash.
What did a year that went nowhere pay?
The flat-year harvest, 6 hour chart. Jan 1 to Oct 23, 2026.
Strategy
+103.8%
Buy and hold
+0.2%
DCA
+3.7%
What did two years of AI spending whiplash pay?
The capex whipsaw, 4 hour chart. Jan 1, 2025 to Oct 23, 2026.
Strategy
+153.6%
Buy and hold
+11.1%
DCA
+1.6%
What if you rode the whole AI cycle and kept buying?
The AI race accumulator, weekly chart. May 1, 2021 to Oct 23, 2026.
Strategy
+217.8%
Buy and hold
+102.8%
DCA
+54.9%
Every run starts from the same hypothetical $100,000 in cash and zero shares.
Backtested results. Past performance does not guarantee future results. For educational purposes only.
Meta, META
The subject
Meta owns Facebook, Instagram, WhatsApp and Threads. Its ad business funds one of the largest AI buildouts in the market, with 2026 capital spending guided as high as $145 billion. Every time that bill changes, the stock reprices.
All time high, Aug 15, 2025
$796.25
Low, Nov 2022
$88
Feb 3, 2022 Falls 26% in a day on results, its worst since the 2012 IPO.
Oct 27, 2022 Falls 24.6% in a day on results.
Nov 2022 Low near $88, about 77% below the 2021 high.
Feb 2, 2024 First dividend. Rises about 20% in a day to a record $474.99.
Apr 21, 2025 Tariff selloff low, a $484.66 close.
Aug 2025 All time high: $790 close Aug 12, $796.25 intraday Aug 15.
Oct 30, 2025 Warns of higher 2026 costs and a one-time tax charge. Falls 11.3%.
Mar 27, 2026 Low of $520.26 after legal losses, 30.1% off the Oct 29 close.
Apr 29, 2026 Capex guide raised to $125B to $145B. About 10% lower in two days.
Late Jul 2026 Q2 free cash flow down 91% on AI spend. About 10% lower after hours.
Sep 2026 Muse AI agent launches. Up 27% for the month, best since Nov 2022.
Oct 14, 2026 Analyst lifts 2027 capex estimate to $225B. Falls 3.3%.
This week Weekly bar down 8.1% to about $652. Q3 results expected Oct 28.
Here is the catch that matters for this study. The business kept growing and the stock kept swinging. Holders lived through every move and were paid for none.
The investor’s dilemma
Meta rewards the AI story and punishes the ride.
8.9% to 11.3%
The two-day move around each of its last four earnings calls. Three of the four were drops.
Paying for the AI bill
Each raise in spending guidance has hit the stock, as in April 2026’s roughly 10% two-day drop. Those who sold the fear sold near the lows; September’s 27% rally followed.
The paper gain trap
$100,000 in META on Jan 1, 2025 was worth about $133,000 at the August 2025 record, about $88,000 at the March 2026 low, and about $111,000 today. Almost two years, and none of it banked.
Movement as stress
$100,000 bought in May 2021 fell to about $28,000 by November 2022, then grew to about $248,000 at the 2025 peak. To a holder, that is something to endure. To a rules-based seller, it is inventory.
DCA deserves credit for buying through the 2026 slide. But it buys blind and never sells. It finished 2026 up 3.7% and the 2025 to 2026 window up 1.6%, with nearly all of its cash spent and none of it realized.
A rules-based algorithm turns the swings themselves into the return.
Market Wave Algorithm for Equities
The algorithm
Ride confirmed moves
It buys when price breaks below the wave. It does not need to know why.
Exit on rules, not nerves
It sells a set share of the position at a target. It sold into the August 2025 record and the September 2026 rally.
Step aside in chop
When the signal is unclear, it waits in cash instead of guessing.
Two of these recipes move most of the position and end in cash. The third does the opposite: it buys in 20% slices, trims 5% at a time, and builds the position through every drawdown.
The flat-year harvest, 6 hour chart
Scope 8
0.5 micro 20 macro
Mid scope, near full swings, trend-break exits. Buy −0.3%, exit +0.3%. 87% of cash in, 97% of shares out.
The capex whipsaw, 4 hour chart
Scope 10
0.5 micro 20 macro
Mid-macro scope, big swings. Buy −2%, exit +8%. 70% of cash in, 90% of shares out.
The AI race accumulator, weekly chart
Scope 4
0.5 micro 20 macro
Accumulator, scale in, trim out. Buy −5%, exit +6%. 20% of cash in, 5% of shares out.
Backtested results. Past performance does not guarantee future results. For educational purposes only.
Scenario one
“The flat-year harvest”
January 1, 2026 to October 23, 2026 · 6 hour chart · Scope 8, buy below the wave, sell on trend breaks
The algorithm bought each deep dip below the wave: late January, the March low near $520, late June and August. It sold each rebound, including into the September Muse rally.
Strategy return
+103.8%
Buy and hold +0.2%, outperformance +103.6 percentage points
DCA +3.7%, outperformance +100.1 percentage points
Ending value
$203,840
Realized profit
$103,730
Cash
$203,187
Shares held
1 at $542.65 avg cost
Open full-size chart
The number to read The number to read is the cash. Ten months in, the account holds $203,187 and one share, going into a Q3 report the stock has already fallen 8% ahead of.
Backtested on the 6 hour chart, valued October 23, 2026. Past performance does not guarantee future results. For educational purposes only.
Scenario two
“The capex whipsaw”
January 1, 2025 to October 23, 2026 · 4 hour chart · Scope 10, buy below the wave, big swings
It bought the April 2025 tariff low and sold 166 shares into the summer run to the record. It rebought the November slide, sold into the January pop, rebought the March 2026 low, and sold 244 shares into the September rally.
Strategy return
+153.6%
Buy and hold +11.1%, outperformance +142.5 percentage points
DCA +1.6%, outperformance +152.1 percentage points
Ending value
$253,634
Realized profit
$150,104
Unrealized profit
$3,530
Shares held
28 at $532.71 avg cost
Cash
$235,188
Open full-size chart
The number to read The number to read is realized profit: $150,104, one and a half times the starting capital, in a window where buy and hold made 11%.
Backtested on the 4 hour chart, valued October 23, 2026. Past performance does not guarantee future results. For educational purposes only.
Scenario three
“The AI race accumulator”
May 1, 2021 to October 23, 2026 · Weekly chart · Scope 4, buy below the wave, sell above it, 20% buys and 5% trims
It sat out the 2021 top. From the February 2022 crash to the November low near $88 it bought in 20% slices, building most of the position under $200, then trimmed 5% at a time through the 2023 to 2025 run. In 2026 it added again below the wave.
This recipe does not try to live in cash. It uses volatility to own more of an outperforming sector leader, cheaper.
Strategy return
+217.8%
Buy and hold +102.8%, outperformance +115.0 percentage points
DCA +54.9%, outperformance +162.9 percentage points
Ending value
$317,758
Realized profit
$35,926
Unrealized profit
$181,832
Shares held
436 at $229.93 avg cost
Cash
$35,679
Open full-size chart
The number to read The number to read is the share count: 436 at $229.93. The buy and holder owns about 313, bought near $319. That is 39% more Meta at 28% lower cost.
Backtested on the weekly chart, valued October 23, 2026. Past performance does not guarantee future results. For educational purposes only.
Side by side
Read the realized profit and shares rows. Buy and hold and DCA never sell, so they banked nothing.
META results by scenario
The flat-year harvest
The capex whipsaw
The AI race accumulator
Window
Jan 1 to Oct 23, 2026
Jan 1, 2025 to Oct 23, 2026
May 1, 2021 to Oct 23, 2026
Chart
6 hour
4 hour
Weekly
Recipe
Scope 8, trend-break exits
Scope 10, big swings
Scope 4, 20% in, 5% out
Strategy return
+103.8%
+153.6%
+217.8%
Buy and hold
+0.2%
+11.1%
+102.8%
DCA
+3.7%
+1.6%
+54.9%
vs buy and hold
+103.6 pts
+142.5 pts
+115.0 pts
vs DCA
+100.1 pts
+152.1 pts
+162.9 pts
Ending value
$203,840
$253,634
$317,758
Realized profit
$103,730
$150,104
$35,926
Shares held
1
28
436
Cash
$203,187
$235,188
$35,679
All three from $100,000 and zero shares.
Backtested results. Past performance does not guarantee future results. For educational purposes only.
Why the results differ
The three runs are the same algorithm answering different questions.
The flat-year harvest asks: can a year that went nowhere still pay?
Answer: Buy only below the wave, sell nearly everything on each rebound. $103,730 in cash while the stock finished flat.
The capex whipsaw asks: can every spending scare become a paycheck?
Answer: A wider scope catches the big post-earnings swings and moves 70% to 90% of the position each time. $150,104 realized.
The accumulator asks: can I ride the whole AI cycle and own more of it?
Answer: Small buys in drawdowns, small trims in rallies. 436 shares at $229.93 and $35,679 in cash.
Buy and hold made one decision and lived with it.
The edge is not prediction. It is getting paid on the swing itself.
Backtested results. Past performance does not guarantee future results. For educational purposes only.
The real lesson
01
The AI bill is inventory
Meta reprices every time its spending guide moves. Rules turn that repricing into realized profit.
02
A flat year is not a dead year
In 2026 Meta went nowhere (+0.2%) yet traveled between $520 and $780. The harvest recipe made +103.8% from that range.
03
Cash is a position. So are shares
The 6 hour run ends with $203,187 and one share: if Meta falls on earnings, it has the cash to buy. The weekly run ends with 436 shares at $229.93: if the AI trade keeps winning, it owns more of it than buy and hold.
The client decision is how you want to be paid: swings harvested into cash, or swings harvested into a larger position in the sector leading the market. The Market Wave Algorithm for Equities has a recipe for each, and it works the same way on every volatile AI name.
Backtested results. Past performance does not guarantee future results. For educational purposes only.
A path forward
Arch Public runs the Market Wave Algorithm for Equities in your own brokerage account.
Zero custody. Full control.
Rules, not reactions.
What do you want your account doing the next time Meta moves 10% on an earnings call?
Book a demo Or talk with the Concierge team
Buying Meta in May 2021 and holding returned
+102.8%
The same $100,000 under the weekly recipe returned
+217.8%
and owns 436 shares at $229.93.
More case studies
PLTR
Volatility Harvesting for Palantir
Strategy return +312% in “The gap buyer, finer wave”
Read the case study MU
MU: a trillion-dollar run, a 41% drop in five weeks, and three recipes that paid in cash
Strategy return +461.7% in “The 2026 scoreboard”
Read the case study MRNA
MRNA: a pandemic boom, a 90% bust, and this month's spike
Strategy return +1,187% in “The round trip”
Read the case study
See every case study
Important disclosures
Arch Public is a software provider, not a registered investment adviser. Nothing in this case study is investment advice or a recommendation to buy or sell any security. All results are backtested using the Market Wave Algorithm for Equities on historical data and are hypothetical. Backtested performance has inherent limitations: it is prepared with the benefit of hindsight, does not reflect actual trading, and does not account for all fees, taxes, slippage, or market conditions. Past performance, backtested or otherwise, does not guarantee future results. Trading securities involves risk of loss, including loss of principal. Ticker history and price levels are drawn from public market data and should be verified before publication. For educational purposes only.
© 2026 Arch Public. All rights reserved. Published October 22, 2026.
Terms and conditions
Privacy policy
Cookies and trackers
Your privacy choices