Yu Sun

Head of AI. Former data scientist.
I express my macro views through trading.
Not financial advice. No trade signals.

[Yu Sun @ysun1993]

血汗工厂。

我至今忘不掉某大厂 OD 合同面试了我 6 轮然后 HR 得知我已经在原单位提了离职，直接腰斩薪资，当然我比较硬气，我也不签了，这还签个蛋。

[plantegg @plantegg]

小红书在三年前裁掉了广东华南区销售的一个负责人陈浩。陈浩一直不服，因为他觉得自己的绩效没问题，而且期权在快要到手的时候突然被裁员，认为公司明显是针对他。

于是这几年，陈浩一直在跟小红书打官司（打劳动仲裁的官司，打讨要期权的官司）。直到最近，小红书向港股递交上市申请的时候，陈浩祭出了王炸：香港交所举报小红书 VIE 架构有问题

首先是劳动仲裁，公司应该赔偿了大几十万，因为本来就是非法裁员，这点没什么好说的。

然后在第二个讨要期权的官司里，公司给出的证据是：期权是和境外的公司签订的，而他状告的是中国的运营实体，两者之间不相关。

这些互联网公司给的期权都是境外公司的，因为公司的架构就是这样。签合同/发工资的是境内实体公司(VIE 架构)

所以这次当小红书递交上市申请的时候，陈浩去港交所举报，指出上市实体、境内运营实体和境外的上市公司不一致，因为有法院的判决书为证。

这一下，相当于把中国互联网行业的灰色 VIE架构直接拿到法庭上来说了。

本来 VIE 架构是处在灰色边缘地带，政府也没有追究。如果这些互联网公司不要动这种歪心思，比如说期权这种，让人家去境外告，给的时候不说，到打官司了就开始扯到境外公司上去。

境外就是一个皮包公司，总不能去开曼群岛告吧？

浩哥是个狠人，瑞思拜，小红书的骚操作会不会毁掉整个行业的 VIE？

[作业借你抄 （吃瓜版） @chaozuoye]

太搞笑了，小红书📕因为裁员导致上市失败了

小红书为了上市财报好看，省钱裁退了几十个有原始股权的员工

其中一个员工走了劳动仲裁，小红书法务说法大概是：

期权是境外公司发的，跟国内公司无关，你要告，去告境外公司

翻译一下就是：

你起诉鲁迅，跟我周树人有什么关系🤣

结果现在问题来了，如果境内公司和境外公司真没关系

那港股上市时的 VIE 架构怎么解释？

港交所在审批时留意到了这个案件，导致招股书里VIE 框架直接崩塌

直接导致上市失败，沦为笑柄 😂😂😂

[Yu Sun @ysun1993]

@hiarun02 You could use higher reasoning mode

[Yu Sun @ysun1993]

My CPI base case is simple:

This CPI is unlikely to show inflation getting worse.

Headline CPI should cool meaningfully, possibly close to flat month-over-month.

Core CPI is still sticky, but my base case is not a blowout. I expect something around the 0.2%–0.25% area: uncomfortable, but not explosive.

The first-order war inflation shock has already been largely offset by the reversal in oil and gasoline.

The second-order risk is still real. I am not dismissing it.

But that risk is more likely to show up later through PPI, freight, services, and corporate margins — not through another obvious deterioration in June headline CPI.

That creates a messy setup for QQQ:

A mild CPI can help QQQ.
A weak labor market can hurt QQQ.
Warsh staying hawkish or neutral is not bullish for QQQ.
AI/HBM can still keep a bid under QQQ.

Too many crosscurrents. I do not like messy causal chains.

I prefer clean ones: that is why I am looking at TLT here.

Why TLT instead of IEF?

Because I do not just want a generic Treasury hedge here.

I want duration.

IEF is cleaner and less volatile, but it gives me a lower-beta expression. If the market only prices a mild front-end adjustment, IEF is fine.

But my thesis is that a softer CPI / cooler oil / weaker labor mix can push the market to reprice the intermediate and long end more meaningfully.

That is where TLT works better.

Longer duration means more sensitivity to every basis point move in yields. It also gives better convexity if long yields start moving lower quickly.

Under the current information set, the TLT chain is shorter:

CPI / oil / labor cooling
→ lower hike risk
→ lower intermediate and long-end yields
→ higher TLT

My expression:
TLT Aug 21 85C / 90C call debit spread.

Why a spread?

Because I am only moderately bullish: the market has already absorbed part of the post-payrolls relief and part of Warsh’s hawkish shock.

Before every event trade, I want to ask the same question:

How much of this is already priced?

I do not want to chase TLT upside with naked calls after the easy part of the rates repricing has already started.

The 85C gives me near-spot upside exposure.
The 90C caps the payout, but helps finance the debit.

This is not a moonshot trade. It is a defined-risk, moderate-bullish expression of a cleaner rates thesis.

Why Aug 21?

Because I want the trade to cover the real catalyst window:

Jul 14 CPI
Jul 15 PPI
late-July Fed repricing
the bond-market reaction before the next major labor/inflation window

Front-week options may look cheaper, but they are not necessarily better.

I want enough time for the bond market to process the data, not just one binary CPI print.

The invalidation is also clear, and I will not focus only on a mathematical stop.

The trade weakens if:

core CPI is clearly hotter than expected,
PPI shows ex-energy inflation spreading into goods/services,
Warsh strongly re-emphasizes hike risk,
or term premium keeps long yields sticky despite softer data.

Those are low-probability risks in my base case, but low probability does not mean zero.

Tail risk always matters.

Again left side, yes.
But with a clean causal chain and defined risk.

[Yu Sun @ysun1993]

@icecreamnong 我总认为经济学一定要有所涉猎的：https://t.co/VWmXM9RMXu

普通人一辈子就那么1~2次加超重杠杆的机会：把握住了就翻身；把握不住就做一辈子牛马。

一定要珍惜这得来不易的机会。

[Yu Sun @ysun1993]

完全认同。在有把握的时候上低杠杆、及时通过获利了结去杠杆本来就是投资的必要手段，巴菲特也承认他多年以来大量借入日元进行投资。

评论区很多质疑的，估计是还没想明白：人家卖房子给你，拿走你未来的现金流之余，换来了自由，他可以拿着这笔钱退休、去旅行、去丰富人生，而你只能闷头打工还钱给银行。

你想翻身吗？想找到下一个接盘侠来故技重施？也不看看现在是什么时代：居民收入与房价早就严重脱节，这个是不争的事实；你要卖一个好价钱（你当然要卖一个好价钱），需要很多人未来30年的现金流了——还有这么多人愿意做这种交易吗——人口结构已经变了。

普通人一辈子就那么一两次上重杠杆的机会，要么翻身，要么就一辈子看到头了。现在如果还铁着头去买小城市、非闹市区、区域内流通性不好的房子，不好意思，________。

[Yu Sun @ysun1993]

@z3f_666 认同😃
上周应很多朋友要求写了一篇长文：
https://t.co/LhRKtY574T
微信公众号也同步发了，结果上来就在几个群里被嫌弃：
“说这些没用，直接说买什么赚”
我可能才是歌谭市的那位吧🤡

[Yu Sun @ysun1993]

https://t.co/F19QQBaf3U

[Yu Sun @ysun1993]

Yeah it's hard, like we're suffering delivery pressure during work, I know that 😬

[Lucca Huguet @luccahuguet]

i want to live from first principles, but it's hard

[Yu Sun @ysun1993]

@icecreamnong @Shuton_leon 确实……它的词库好少，但在 NixOS 上做声明式安装也许只能选 rime 了，它可以同时支持中文和粤语，我还在发掘它的其他功能🧐

[Yu Sun @ysun1993]

@icecreamnong @Shuton_leon 可恶的 rime🤡

[Pallavi @Pallavi_345]

Bro created Linux.

Changed computing forever.

Still commits code like he just got his first internship. 😭

[Yu Sun @ysun1993]

@icecreamnong @Shuton_leon 那个最后一个字是怎么打出来的，是 zu 吗？我怀疑我的输入法坏了……

[Yu Sun @ysun1993]

@Rajveer23667132 🥴我知道这是干啥……但我会劝他们换完 U 拿去 PolyMarket 做 MM 提供流动性算了……

大资金最喜欢最舒服的还是 delta 中性的交易~

[Yu Sun @ysun1993]

我在很早的时候就参与了 Burn 的开发了。当时手搓自动微分，也是头都大了，时常怀疑自己算错……
时至今日，中间我也怀疑过自己，现在觉得要这样看待这个问题：

一般的程序员也许曾经很熟手，但突然就不会编程了。

我的同事和朋友们还处在混混沌沌的状态，一时间猛吹自己用上了 Claude，一问是某中转站买的；一时间全都用上了 Codex，却连什么是 Skills、什么是 MCP 都不知道，又抱怨“这玩意有什么好用的天天写屎山”，天知道他们经历了多少次恐怖的上下文腐化，一直在重复倒粪。

而你不同啊，你很容易就理解什么是 Context，如何高效管理、按需引入，什么时候该 fork a new session……

而又因为你懂 transformer，知道什么是 KV cache，所以你知道 HBM 的重要性，早人一步拿捏到产业瓶颈进行投资，普通人的生活和工作成本被昂贵的 Pro 20x 抬高的时候，你早就在金融市场收割了这份溢价。

知识和经验从来都不会太多。
马拉松冠军回首的时候也会有点诧异吧：我当初还练过那么糟糕的长跑。
这很正常。

[Yu Sun @ysun1993]

https://t.co/F19QQBaf3U

[Yu Sun @ysun1993]

@nini_incrypto_ 避免择时风险

[Yu Sun @ysun1993]

@wangwatchworld 是啊。但投资经验丰富的穷人也会觉得机会成本很贵。
所以经济学好啊，得学啊。

[Yu Sun @ysun1993]

@ally_foe 当我想对我的兄弟好，to be like:

[Yu Sun @ysun1993]

@Li665508Li 虽然但是，我在努力用这个账号做投资教育，我的宏观看法和交易逻辑毫无保留地拆解出来了，也许是推荐算法的原因，似乎还没什么人看。
但没关系，慢慢来吧。

[Yu Sun @ysun1993]

@Seraphine_Moyao 可怜我预测到了大盘下跌，但不会破位，就直接止盈了put，否则估计今晚要吃爽了🤡