The S&P 500 has stopped behaving like the S&P 500.

本帖於 2026-04-21 03:24:02 時間, 由版主 lionhill 編輯

 

The S&P 500 has stopped behaving like the S&P 500.

I noticed something strange this week and went digging.

I pulled three years of daily returns and ran rolling 30-day correlations between the S&P and its four primary macro factors — the 10-year yield, the dollar, oil, and gold. Three of those four correlations are currently sitting at the most extreme readings they've been at in the entire three-year sample. Not close to extreme. At extreme. 0th, 1st, and 1st percentile respectively. The fourth — gold — is in the 84th percentile. Normally uncorrelated assets are now moving in near-lockstep. Normally loosely-related assets are now mirror images of each other.

Put plainly: when yields rise, the S&P falls hard. When the dollar strengthens, the S&P falls hard. When oil rips higher, the S&P falls hard. And when gold catches a bid, the S&P goes with it.

That is not how equities usually behave. That is how treasuries behave.

Here's what I think is happening. Traditional safe havens aren't working. Treasuries have been a broken trade since 2022 — yield duration is punishing, the rate path is unreadable with CPI at 3.3% and Iran still hot, and the dollar is too political to hide in. So capital looking for a "less bad" place to sit has piled into large-cap US equity — specifically the Mag 7 — as a substitute safe haven. Mega-cap tech has become the quasi-treasury of 2026. It's liquid, it's deep, it has a predictable cash flow profile, and right now it's being priced like a duration bet, not a growth bet.

This is a regime flag, not a trade. Regimes like this don't usually break gradually. They break when the single binary variable holding them together breaks — in this case, the Iran ceasefire on April 22.

When four correlations all go extreme at once, the market has stopped pricing the asset and started pricing the story.

When the story changes, all four mean-revert at the same time.

 

所有跟帖: 

UNH, 厚厚厚,盤前漲了21刀 -甜酒甜- 給 甜酒甜 發送悄悄話 甜酒甜 的博客首頁 (0 bytes) () 04/21/2026 postreply 03:37:01

UnitedHealth tops quarterly estimates, hikes profit outlook -lionhill- 給 lionhill 發送悄悄話 lionhill 的博客首頁 (0 bytes) () 04/21/2026 postreply 03:43:25

是的,獅山早! -甜酒甜- 給 甜酒甜 發送悄悄話 甜酒甜 的博客首頁 (0 bytes) () 04/21/2026 postreply 03:48:36

沒敢賭啊,昨天把被套的unh留著了,把unhg全出了 -hahaha2025- 給 hahaha2025 發送悄悄話 (0 bytes) () 04/21/2026 postreply 04:05:35

想當年剛爬上200日均線,被建國同誌一推下萬丈深淵 -hahaha2025- 給 hahaha2025 發送悄悄話 (0 bytes) () 04/21/2026 postreply 04:07:22

至少經受住了建國的打擊,LOL -甜酒甜- 給 甜酒甜 發送悄悄話 甜酒甜 的博客首頁 (49 bytes) () 04/21/2026 postreply 04:12:54

吼吼,終於有點結果了,等的花兒都謝了 -flyingcandle- 給 flyingcandle 發送悄悄話 (0 bytes) () 04/21/2026 postreply 04:28:39

這一年多,他家 -甜酒甜- 給 甜酒甜 發送悄悄話 甜酒甜 的博客首頁 (95 bytes) () 04/21/2026 postreply 04:38:00

關鍵要判斷哪個點位出,有人說股票很好等400,就被套了。上季度-18%意想不到的 -hahaha2025- 給 hahaha2025 發送悄悄話 (0 bytes) () 04/21/2026 postreply 05:30:58

投坦季度選股,我上個月選了他家 -甜酒甜- 給 甜酒甜 發送悄悄話 甜酒甜 的博客首頁 (377 bytes) () 04/21/2026 postreply 05:58:57

投壇啥時候季度選股? -flyingcandle- 給 flyingcandle 發送悄悄話 (0 bytes) () 04/21/2026 postreply 06:20:48

別的壇子的事,還是放到別的壇子去說吧 -三心三意- 給 三心三意 發送悄悄話 三心三意 的博客首頁 (0 bytes) () 04/21/2026 postreply 07:00:44

很典型的簡中文體,即便用英文寫成都脫不掉的基因,數據和結論的脫節和跳躍 -楚懷沙- 給 楚懷沙 發送悄悄話 (0 bytes) () 04/21/2026 postreply 12:53:39

不是我寫的,linkedin上看到copy過來的,純美國人,那來的簡中文體,別那麽相信自己的判斷力 -lionhill- 給 lionhill 發送悄悄話 lionhill 的博客首頁 (0 bytes) () 04/21/2026 postreply 15:24:10

純美國人也吃中餐,灣區和夏威夷很常見,還有一些純美國媒體的報道唯一用途就是給簡中引用和轉載 -楚懷沙- 給 楚懷沙 發送悄悄話 (0 bytes) () 04/21/2026 postreply 18:54:53

這個文章提到的數據是我最關注,研究最多的領域,所以,評價一下 -楚懷沙- 給 楚懷沙 發送悄悄話 (0 bytes) () 04/21/2026 postreply 19:11:17

請您先登陸,再發跟帖!