Trata (YC W25)

We host conversations that drive public markets

[Trata (YC W25) @trytrata]

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

[Trata (YC W25) @trytrata]

A special thanks to @MaheshRam23629 and @zephyllr for their amazing work on this!!

[Trata (YC W25) @trytrata]

Today, Trata is excited to share the revised Hedge-Bench: our benchmark tying open-ending reasoning to prediction capabilities.

We constructed 92 real-world environments derived from hedge fund analysts’ reasoning traces. Agents reason from frozen information snapshots, and their forecasts are evaluated against future real-world outcomes. No model scores above 50% (GPT-6 Astra being the top performer at 44.2%). For frontier, state-of-the-art models, these results highlight a persistent and wide gap between producing plausible-sounding reasoning and making accurate, economically valuable forecasts. In a semi-verifiable domain like finance, where outcomes are delayed and reasoning is difficult to verify, intermediate expert reasoning traces are crucial.

We post-trained @Meta's Muse-Glimmer-30B on 165 environments with per-turn credit assignment using an LLM-as-a-Judge. We observe evidence of generalization: held-out forecasting accuracy at high effort improved by 10.06 percentage points, while out-of-distribution evaluation scores improved by 3.94 points on @OpenAI's GDPval, 2.29 on @Surge’s GDP.pdf, 7.07 on @Harvey’s LAB, and 4.96 on @Samaya_AI’s FrontierFinance under their respective metrics.

[Trata (YC W25) @trytrata]

Arxiv:
https://t.co/sI3Rpya3fC

[Trata (YC W25) @trytrata]

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

[Trata (YC W25) @trytrata]

Our paper:
https://t.co/sI3Rpya3fC

[Trata (YC W25) @trytrata]

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).

[Trata (YC W25) @trytrata]

Lastly, we at Trata deeply believe in the value of human judgement. In a future where LLMs continue to improve, our aim is to augment human capabilities and enable a fairer playing field outside of just the largest asset managers. Based on our findings, Analysts - your jobs are very, very safe.

[Trata (YC W25) @trytrata]

Hedge-Bench tasks are built from real world traces given our proprietary data flywheel and network of actual industry practitioners. No models have seen our solutions during pre-training. Prompts are deliberately short to reflect the communication style of actual professionals. Tasks are divided into themes and further de-composed into sub-themes to reflect the multi-step, open-ended work Analysts do when decomposing a topic into chunks of actionable reasoning.

[Trata (YC W25) @trytrata]

Which models should our users be leveraging for non-trivial reasoning work? Some quick experiments, graph made in 4 minutes with @Lovable

[Trata (YC W25) @trytrata]

$GS, 1.22.26

Analyst 1:
"I started my career at Goldman. Then we ventured out and did a spin-off for a few years."

Analyst 2:
"I also started my career at Goldman. I was there before and during the financial crisis. Lots of interesting stories got thrown in there. I was a banker, but I remember at the time bigger and bigger analyst classes. This was the heyday in terms of prop trading. It was pre-Volcker Rule."

lol

[Trata (YC W25) @trytrata]

$MDB, Bull/Skeptic 1.20.26

Analyst 2:
"I was thinking more of @MongoDB Atlas as the system of record. It's a weird application of the classic economic hoteling cost. If I have data sitting in there, there's a cost to calling an @OpenAI API to compute my embeddings. But if your Voyage thing inside Mongo does it locally for me, that's great. I don't need to worry about my data being used by OpenAI in certain ways, lower compliance, or Mongo not being approved. So that flywheel is reinforced. I don't need to leave the Mongo universe to achieve my outcomes. As a result, Mongo becomes the system of record.

Analyst 1:
Is this your view that it increases the legitimacy of being a system of record?

Analyst 2:
It seems that way to me as I think about it. I think that's what they're going for. They're talking about being a NoSQL data platform, emphasizing their important role in data readiness, and acquiring a capability or technology for generating high-quality embeddings within their ecosystem.

They've already got great vector data-based tools. So when it comes to agentic AI, which requires vector search and data sitting in vector DBs for inference, you've got that in place. To me, the pieces of the puzzle that you're laying out sound like they're trying to be a system of record in an agentic AI future. That's the bold view. Of course, someone might say, "It's not going to integrate, and it's going to come down like a house of cards.""

[Trata (YC W25) @trytrata]

$BME.L

"Alex Russo was a D-tier CEO. Everyone looked great in 2021-2022, but it became clear in November 2024, when I was seriously considering the stock for the first time, that he wasn't running the business with long-term performance in mind. He was making decisions based on his instincts, going by vibes, rather than using data to inform his decision-making. So, pretty negative on Alex.

It remains to be seen regarding the new one. He's a private equity CEO, who had mixed success in his position. Part of that was his fault, and the rest was due to external factors. That's one yellow flag, but his prior role was in a fundamentally different business, so I'd be careful to extrapolate his performance there too much. We've done reference checks on him, and it seems generally positive-to-somewhat of a mixed bag."

[Trata (YC W25) @trytrata]

"It was a really frustrating outcome. I sold after the announcement. When I saw the initial headline, I thought there’s no way that was the true value they had sold the company. You looked at the publicly traded comps, especially with the IPOs of American Integrity and Aspen.

I commend the activists for getting involved. That's somewhat of a Pyrrhic victory, because you're never going to get over the Maryland hurdle and the insider ownership stake. Today, it’s a cash box, and in the future, they'll invest in similar businesses they’ve seen before. Maybe they'll be the exit liquidity to somebody that's got something hung up, and you can build a platform. I think it’ll be an attractive contingent on that deal being announced. It could be worth a flyer to take a smaller position, and then, once the deal's announced, if you like the economics, you go for it because you're not going back to a holdco structure. That model hasn't worked for many "next Berkshires" other than Berkshire Hathaway, Markel, and maybe Loews and Fairfax."

$BRK $MKL $L $FRFHF

[Trata (YC W25) @trytrata]

Short $ASTS, 1.16.26:

"The bigger concern is that by the time these 72 AST satellites without Ligado compatibility are launched, @Starlink will be on the next generation and fairly ahead. We have yet to hear a fundamentally compelling argument for the next-generation Starlink satellites having structurally worse technology than AST. We appreciate that they have a launch advantage and a launch-cost advantage at this point, and maybe to a greater extent in the future, depending on how excited you are about Starship. Generally, it's less of a tech concern and more that they're just not going to be able to keep up with a very robust competitor like @SpaceX . We've seen continued delays with their satellite launches, which doesn't give me a ton of faith that they're going to be able to get those 72 birds, or even 30 to 45 birds, out in short order. The biggest concern is more competitive.

[Trata (YC W25) @trytrata]

The bear case for $MNDY, 1.13.26:

"For a company like Monday, it's very important for them to keep adding new customers to the funnel. What I expect to happen is basically you're going to have so many competitors flooding into pretty much every software category, the same way that happened with PowerPoint, which was untouched for 30 years. I expect it to be disrupted by things like Gamma, but PowerPoint is protected by the Microsoft bundle in a way; it's not its own company, and it's definitely not trying to add massive numbers of new users every year that it has to compete with other platforms.

...That's the main way I see risks to Monday specifically: being both an application company built on the old paradigm with much lower revenue per employee and a company that has to win a ton of new customers to continue its growth trajectory. You can argue that, given the current valuation, it doesn't need to win many new customers, but for now, I'm basically arguing whether Monday is still a growth company or should be valued as a secular growth company. It is much more debatable right now."

$MSFT @GammaApp

[Trata (YC W25) @trytrata]

@BillBrewsterTBB Brewing some business!