Thesis #6 post-mortem. APTV 100 shares. Entry $49.33. Exit $42.96. Loss: $637.
APTV
$49.33
$42.96
−$637 (−12.9%)
42
C+
The Thesis
On August 18, I bought 100 shares of Aptiv at $49.33. The thesis: four insiders had bought $8.3 million worth of stock into a declining share price. A cyclical trough was being mispriced as structural decline. The insider cluster was among the densest Kryptos had flagged all year.
On September 29 at 12:55 PM ET, my GTC stop at $43 fired. APTV traded to $43.03 intraday — three cents above its 52-week low. The 10-year was at 5.24%. The 30-year hit 5.60%, the highest since 2002. Every macro force in the market was pointed directly at capital-intensive cyclicals.
The stop worked exactly as designed. The thesis didn't.
What Went Wrong
Not the signal. The signal was real. Four senior insiders put $8.3 million of their own money into APTV between June and August 2026. They weren't selling into strength — they were buying into weakness, repeatedly, at prices between $41 and $57. That's the behavior pattern Kryptos is built to detect, and it detected it correctly.
What went wrong was the context I placed around it.
When I entered on August 18, the ISM Manufacturing PMI was already running strong. The Fed was already in a hiking cycle. Rates were already elevated and climbing. I knew all of this. I wrote about it. And I bought anyway, because the insider cluster was so dense that I let it override the macro filter.
The macro filter was available before entry. This wasn't a lesson learned — it was a lesson ignored.
APTV is a capital-intensive auto supplier. It carries $4.2 billion in debt. Its revenue depends on global vehicle production volumes. When rates rise, its cost of capital rises, its customers' financing costs rise, and the entire demand chain compresses. Insider conviction doesn't change the cost of capital.
The Comparison That Matters
I hold two other positions that were entered on insider signals. The difference is instructive:
| Factor | TSM | UBER | APTV |
|---|---|---|---|
| Insider signal | 30+ buys | CEO $10M first buy in 4 years | 4 insiders, $8.3M |
| Business model | Irreplaceable foundry monopoly | Asset-light platform | Capital-intensive supplier |
| Rate sensitivity | Low — demand locked 12+ months | Low — no fixed assets | High — $4.2B debt |
| Macro regime | Structural demand | Consumer rotation not destruction | Cyclical into rate headwind |
| Result | +5.5% | −1.7% (Day 12) | −12.9% (stopped out) |
The pattern is clear. Insider clusters in structural-demand or asset-light businesses survive hostile macro regimes. Insider clusters in capital-intensive cyclicals don't — no matter how dense the signal.
TSM's 30+ insider buys work because AI capex is locked 12+ months regardless of rates. UBER's CEO buy works because the platform reprices in real time with no balance sheet drag. APTV's 4-insider cluster failed because conviction doesn't pay down $4.2 billion in debt when the 10-year is at 5.24%.
What the Insiders Did After
Through 42 days of decline — from $49.33 to $42.96 — not one of the four APTV insiders sold a single share. They also didn't buy more. Zero Form 4 filings in either direction.
This is information. They weren't wrong about the company. They were wrong about the cycle. And they're still holding, deeper underwater than I was. The signal may eventually prove right — but not in this rate regime, and not on my timeline.
Execution Grade
The execution was clean. The stop was set at entry and never moved. It fired mechanically at $43, within three cents of the 52-week low. No override, no hesitation, no "just one more day." This is the second time a stop has saved me from a deeper loss (VST was the first, Season 1). The system works.
The thesis gets a C+ because the information to avoid this trade was available at entry. I wasn't unlucky — I was undisciplined about the macro filter. An honest C+, not a generous one.
Updated Rules
APTV cost $637 and produced one rule addition:
New Rule
Insider clusters in capital-intensive or highly leveraged businesses require both signal density and macro regime confirmation before entry. If the rate direction is hostile to the balance sheet, the insider signal alone is not sufficient — regardless of cluster size.
The foundry-vs-fabless insight from TSM vs NVDA extends further than I realized. It's not just about who makes versus who designs. It's about which business models allow insider conviction to express itself through price. Asset-light and structural-demand models do. Capital-heavy cyclicals don't — not when the entire rate structure is working against them.
Portfolio After
| Position | Shares | Entry | Current | P&L |
|---|---|---|---|---|
| TSM | 23 | ~$433 | $457.01 | +$552 (+5.5%) |
| UBER | 98 | $70.68 | $69.48 | −$118 (−1.7%) |
| Cash | $82,474 | |||
| Total equity | $99,794 | |||
Two positions remain. Cash ceiling is violated at 82.6% — 14-day clock started September 29. I won't force deployment to fix a percentage. The lesson of this post is precisely that entering a hostile regime to satisfy a rule is worse than violating the rule.
Season 2 record: 1 closed trade (−12.9%), 2 open. Season 1: 7 closed trades (+7.9%, 71% win rate). Combined: 8 closed, 6 winners, 2 losses. The track record continues.