![Ben Charoenwong Profile](https://pbs.twimg.com/profile_images/1502628533180325898/rEb5gKCg_x96.jpg)
Ben Charoenwong
@BenCharoenwong
Followers
1K
Following
31K
Statuses
8K
Finance prof @INSEAD. @ChicagoBooth and @UMich alum. I mostly RT news and make snide remarks. he/him.
Singapore
Joined June 2011
@bennpeifert Some bonuses are paid out directly from the pass through. So in bad times, the variable cost is lower. Helps smooth returns and mother ship keeps more.
0
0
3
Cool to see @perplexity_ai packaged together with @Singtel here in Singapore. Actually made me think about switching back for a bit.
1
0
5
@bennpeifert To me the real question is why aren't they not in textbooks? Maybe because too many are written by financial math people without proper finance industry exposure? Also how we got some of the big blowups in the first place?
0
0
2
RT @nathanbaugh27: I read this last night, called my mom, played a game with my dad. Had a great evening. But man this idea hit me hard. ht…
0
52
0
@mariuszoican @CBCCalgary @HaskayneSchool And importantly, don't expect the brokerage to help you figure out your biases!
0
0
1
@bennpeifert Specifically for short dated opts, hearsay is they're hard to hedge so may cause price distortions in the underlying. Wnat to understand market maker considerations
0
0
0
Cool intervention and design. 5 years in the making.
Firms struggle to forecast, often making over-optimistic inaccurate forecasts. Running a 5-year 6,000 firm experiment finds that simple data interventions can yield large improvements, from @I_Am_NickBloom, @m_codreanu, and Robert A. Fletcher
0
0
2
That's new. But likely because R1 already has chain of thought well integrated, and that it's CoT is probably superior to what users input.
Deepseek prompting tip from their paper: “Few-shot prompting consistently degrades its performance. Therefore, we recommend users directly describe the problem and specify the output format…” Better to use a comprehensive single prompt then multiple (chain-of-thought) prompts.
1
0
2
Anyone at Google Scholar using LLMs for search? Or are we just waiting for @elicitorg or @arxiv to just whack em?
0
0
0
@_KarenHao I think about missing piece in this argument also is that convincing everyone you need lots of chips helps to create the mirage of a moat...since anyone who can't afford 80bn in Capex get disincentivized to try. Inherently anticompetitive. Tot agree on the neg externalities.
0
0
1
RT @predict_addict: **Thailand 🇹🇭 Discovers Conformal Prediction with a Bang, Open-Sources Python Package " The team from Thailand has dev…
0
15
0
@jasminewsun Seems amazing for creative stuff. I found chatgpt much less usable compared to Claude, and now deepseek seems to be the go to ideation
1
0
0