building Trata (YC W25), hedge-bench

San Francisco, CA
Joined June 2015
The coolest publicity we could ask for. I’ve been a fan of Money Stuff for years!!! @matt_levine 🤝 @trytrata Public investors: My DMs are open. Funding secured. As always feel free to reach out anonymously. Acceleration coming soon bloomberg.com/opinion/articl…
3
2
23
3,759
Eric Cho retweeted
tokenmaxxing is a disease 95% of you are building nonsense
230
44
44
933
212,792
Eric Cho retweeted
how do you become an arms dealer? do you just wake up and start cold emailing dictators?
409
4,117
382
96,634
3,132,792
Eric Cho retweeted
What @sama hated most about running yc lol
11
6
25
357
158,698
Fable 5 (@AnthropicAI) scores 22% and tops the Hedge-Bench leaderboard. Running Fable was roughly 2X more expensive than Opus 4.8 per trial. For an industry where accuracy is mission critical, human judgement isn't going away
2
5
28
3,693
Eric Cho retweeted
Within the finance industry, there is a lack of objective measuring around model capabilities outside of rote mechanical work. We mainly see reliance on anecdotal evidence and a growing amount of fearmongering from more vocal personalities on social media. The purpose of Hedge-Bench is to provide a better measure of model capabilities on higher-level, reasoning-oriented workflows that actually matter to industry practitioners. We intend to update Hedge-Bench to test model releases in the future.
2
1
8
1,419
For the Analysts: between the sub-15% pass rate and 80% hallucination rate, your job is very safe
Today, Trata is excited to present Hedge-Bench, the world’s first benchmark focused on evaluating open-ended reasoning in the finance domain. We curated 102 tasks derived explicitly from the reasoning traces of professional hedge fund analysts working with relevant information sources. Most benchmarks are geared towards evaluating naturally deterministic tasks (e.g. updating spreadsheets, calculating formulas). The result is a blind spot on the frontier labs’ part to the competency that matters most in the finance industry: reasoning. No frontier model scores above 16% on Hedge-Bench. We observe Claude-Opus-4.8 actually regresses relative to the last generation of Claude-Sonnet. We see hallucination rates go as high as 82% for multi-step reasoning. These implications extend beyond just the finance domain and into other domains where trust in agent reliability is equally critical (e.g. law).
1
1
9
1,287
Eric Cho retweeted
Does anyone actually run 20 agents in parallel?
24
89
5
2,438
91,924
This app is probably the biggest time sink of feeling productive without actually learning anything - self reflection after increasingly lurking on tech twitter for the past 3 months. Will default to touching grass
4
374
fr
You need to be insanity maxing. You need to be getting crazier. You need to be getting wilder. You need to be constantly out of pocket.
5
841
Trata is now live on X
Trata (@trytrata) is now on FinTwit. We host anonymous buyside to buyside conversations via voice calls. We work with 125+ funds (vs. 10 in March) and receive 25-30 human-to-human calls weekly. We compile these conversations into a library of investment research. EX: Fund 1 is bullish and Fund 2 is bearish on $OPEN, they ask each other questions they want answered surrounding the key debates for anywhere between 30-60 minutes. The average fund on the platform: ~$200-400mm in AUM, 3 investment professionals, fundamental research focused. These funds use tools like Bloomberg and third-party expert networks, and have regular access to management teams. To the qualified lurkers and anon accounts of FinTwit: Request a call on any ticker (bull or bear). If we pair you, we’ll give you: 1. Access to that transcript and any transcripts on that ticker published through the end of 2026 2. Airpods
6
813
when you’re young, you should minimize the degrees of friction between you and serendipity. don’t live 2 hours outside the city to save a few bucks on rent. you’ll miss the last-minute dinners and random meetups that could change your life, and pay back everything you “saved”.
84
461
37
8,961
238,893
you don't have a discipline problem. you have an environment problem
4
94
3
757
44,688
Bill Gurley on why “TAM conservatism” often hurts investors “I’ve found that TAM conservatism hurts you more than it helps you as an investor. If you feel like something is super disruptive and it’s unlocking things, your optionality to build on top of that is going to be pretty spectacular.” Bill recalls NYU professor Aswath Damodaran’s conclusion that Uber was worth only $5.9 billion given the size of the taxi market. Bill responded in a blog post titled “How to Miss By a Mile”: “There are multiple reasons why this is a flawed assumption. When you materially improve an offering, and create new features, functions, experiences, price points, and even enable new use cases, you can materially expand the market in the process. The past can be a poor guide for the future if the future offering is materially different than the past.” To be fair to Aswath, Bill already knew that Uber was 20x the size of the taxi market in San Francisco at the time. But Bill cites a McKinsey study as another classic example. In 1980, they predicted that cell phone penetration in the U.S. by 2000 would be 900,000 units. This turned out to be less than 1% of the actual figure, 109 million. As Box CEO Aaron Levie put it: “Sizing the market for a disruptor based on an incumbent’s market is like sizing the car industry off how many horses there were in 1910.” Today Uber’s market capitalization is north of $150 billion—almost 30x Aswath’s estimate. Video source: @tferriss (2023)
6
57
11
470
74,274
I don’t sleep very much, so I have been really enjoying reading a bunch of dotcom bubble/bust era books. this was written (by a software engineer) about the uselessness of mobile:
53
24
9
729
56,315
"Nobody is an atheist at 50x leverage" is the realest thing ever written on this app
140
1,719
134
21,039
1,221,201
Sunday nights at Trata HQ
6
21
1,490