@drggong
Born in China and educated in the West, sharing unique insights on the latest trends in business, tech, politics, and research. Ex-Capital Markets Economist | Ph.D. in Management Science | Business Scientist More About Me: https://garroshub.github.io/
Weekly #market take
1. #Gold? Not a long war yet. Funds are selling gold to cover equity losses.
2. #Blackswan? Short hedges are near 2008 extremes. Panic is high, but defenses are in place.
3. #Oil? Reserves cannot close the gap. This shock is already bigger than the 1980s oil crisis.
#econsky
Repo: https://github.com/garroshub/backtesting-skill
What if an #AIagent could turn a #trading idea into a backtest in minutes?
I built an AI-native backtesting skill on #Zipline: schema in, results out, fast OOS + grid runs. Bundle data now, with hooks for replaceable data and future live interfaces like #IBKR.
#techsky #econsky
New #USstatistical reality: the headline prints strong, then gets revised away
Now we have 13 straight monthly payrolls downward revisions, -710k in total, and 24 of the last 25 months revised lower.
At this point, the first print is not data; It is #patriotic adjustments. #econsky
What actually happens after oil prices spike?
3 charts worth a look:
1. Geo shocks: mkts recover in 6-12mo. Panic outlasts the damage.
2. Oil's inflation impact: overstated. Energy's share of spending keeps falling.
3. Best hedge: changes every cycle. Gold & commodities rarely hold up.
#econsky
This is the tragedy of Trump's #tariffpolicy. Some firms have already received subsidies, and now they are getting refunds to correct his mistake. #Consumers pay the price twice. It is a massive #wealthtransfer from regular people to firms. #EconSky
Trump would agree with this: "The #US has been stealing Canadaβs oil for peanuts. Total disaster. They take our energy at a huge discount and laugh. We need Fair Trade, not Steal Trade. #CUSMA is a very bad deal for Canada. We want our money back! Canada First!" #CdnPoli #Trump
The #WhiteHouse Press Sec literally said "Victory depends on #Trump." If he defines victory, he can easily define βunlimited" too.
#econsky
Claiming unlimited munitions has zero cost for #Trump, but spreading this #misinformation carries a real price for national readiness.
Fact-check: 3,000+ PGMs gone in 36h. Replacing one AN/FPS-132 takes 8 years and 77kg of #Gallium (98% of which #China controls).
These friction costs are real and they will be turning into productivity shocks. #CdnPoli #EconSky
Current #energyshock is unlikely to drive sustained inflation. Todayβs #NFP print confirms that weakening growth and rising unemployment will now dominate the narrative. The #Fed will be forced toward easing, not tightening.
Look at the re-entrants: up by 150k, returning to December levels. I noted in January that the drop in #unemployment happened for the wrong reasons, mainly because people were just dropping out of the workforce. This reversal confirms how distorted January's data really was. #NFP#econsky
I find #AIagents excel at large-scale empirical replication, posing a real threat to Tier 2-3 journals. However, AI still lacks the depth for mechanistic deduction. This raises the bar for creativity; researchers must focus on The Why over The How.
#AcademicSky
This is a wake-up call for Data Sovereignty in academia. Relying on centralized government streams creates a systemic failure. We must decentralize research data to counter increasing state interference and ensure academic continuity. #EconSky #AcademicSky
#econsky
Think all index #ETFs are the same?
This year S&P 500 equal-weight has at times beaten the standard index and looks more diversified.
It isn't. Equal weight means constant #rebalancing, higher costs and tax drag. Over time, those frictions make it lag the regular S&P 500 ETF.
Live demo (static): garroshub.github.io/smart_money_...
GitHub: github.com/garroshub/sm...
#techsky #fintech
I built an enterprise-style agentic framework demo for multi-account financial planning after seeing #Wealthsimple #AIBuilders post (interest only).
It shows how an agent can turn multi-account data into a consistent client snapshot. (see Github link in the 1st reply)
#econsky
Weekly #Market Heatmap Review (2026-02-23):
1. Macro: #USD up; credit firmer; long-end flat; liquidity slightly tight.
2. Sentiment: Vol down; risk-on; breadth improving.
3. Global: Ex-US leads; USD/rates still key.
4. Rotation: #Defensives/#energy lead
#econsky
#GlobalTrade 2025: Vibes vs Data
#Trump: βU.S. #TradeDeficit down 78%. 2026 surplus.β (no source, no method)
U.S. stats: 2025 deficit still near highs. China's deficit is lowest in 20+ yrs.
#China: 2025 #TradeSurplus a record $1.2T.
Real question: Who do you trustοΌ
#LunarNewYear2026: The Year of the hashtag Horse
Today marks the commencement of the Chinese New Year, designated as the Year of the Horse. The Horse symbolizes kinetic energy, strength, and rapid progression.
Wishing you a prosperous and successful Year of the Horse.
#econsky
#Googleβs 100-Year Bond: Vision or Hubris?
Why now? Ans: Securing ultra-long-term funding for #AI capex.
Why GBP-denominated? Ans: Large institutional demand and FX hedging.
History lesson? Ans: Motorola, but it collapsed after missing a tech shift.
Foresight or peak-cycle hubris?
GitHub Repo: github.com/garroshub/ai...
Live Dashboard: garroshub.github.io/ai-economist...
#econsky
#academia
How AI Should Be Used in #EconomicAnalysis
Iβve just released the #AI Economist Agent on GitHub. It anchors the Taylor Rule and Bayesian inference with a GDPCastNow module (R2=0.83) into a structural framework.
Interpretable macro alpha that beats generic LLM guesses.
2/2: My broader view is that the value of AI fundamentally depends on the userβs domain knowledge and methodological discipline.
BTW, I am still working on extending it to include a GDP CastNOW style module, and the full framework will be open sourced on GitHub once completed.
#econsky
1/2: How AI Should Be Used in #EconomicAnalysis
I recently built a project that may be a good case showing how I integrate #AI into economic analysis, combining the Taylor Rule with Bayesian inference into an Agent Skill.
Results align with consensus, unlike direct #LLM forecasts.
π Live Dashboard: garroshub.github.io/NBA-Enhanced...
π» GitHub Repo: github.com/garroshub/NB...
#NBASky
NBA Defensive Trends: Week 14 (1.25.2026)
Weekly EDI update. My Bayesian-based defensive model tracks how players defend.
Highlights:
1. Wemby (#1) is still in a tier of his own.
2. Mobley (#2) & Holmgren (#2) hold firm.
3. Big jumps from Robert Williams III and Zubac.
#econsky
#Gold Rally Interpretation
This gold rally may not be about risk aversion. With U.S. fiscal tightening dead, markets are pricing financial repression: #debt diluted over time.
In negative real-rate regimes, gold rises as a system hedge. Is this the new reality?
#NBAsky
#VibeCoding Project: NBA Enhanced Defensive Index
Modern #NBA defense is positionless, but metrics like DEPM / DLEBRON still infer individuals from the team on/off.
EDI is an open-source, mechanism-first framework built to explain defense, not just rank it.
