AI Trading Agents Need the Market Data and Research Quants Use

We believe that in a world driven by AI and agents, people will do fewer and fewer things themselves. That raises bigger questions about how we will operate as a society, but it is also what we have been doing for a while: trading effort for convenience. That is not going to slow down, and investing won't be spared.

What investors have been paying for

The financial industry, and investing in particular, is built on one cornerstone: the customer doesn't know about investment opportunities, so he needs a professional for recommendations, advice and, ultimately, performance.

In practice, that promise often doesn't hold up. S&P's SPIVA scorecard found that 79% of active large-cap U.S. equity funds underperformed the S&P 500 in 2025. Meanwhile, the fees banks charge on the funds they offer their customers are still unbelievably high for something mildly better (and often worse) than what an ETF could do.

The ETF and passive industry has its own rabbit hole. Plenty of ETFs do what they promise, cheaply. But creating an index that is supposed to give exposure to a specific thesis or industry has become extremely simple, commoditized and... expensive. Too many of those thematic products look like a sure way for the issuer to rack up fees, more than a way to deliver on performance.

Ultimately, the end user is still underserved when it comes to effective money management. "Most people would be better off buying the S&P 500 and being done with it" is a recurring argument, and not a wrong one.

When you want more than an index fund

But what if you actually want more?

If you don't have access to wealth management or private banking, many opportunities, from unlisted investments to structured products, remain out of reach. But those clients are still a tiny fraction of the people who want to manage their money themselves.

With Covid came an explosion of educational content about trading, and options in particular. It served some of the market's needs and opened the door to more DIY investing, with the promise of fairly simple techniques to boost your returns. The results are much more mixed. Education is nice, but you still remain a largely unsophisticated investor facing an almost efficient market, with limited room for error.

Why LLMs need better context

How does that change in a world dominated by agentic workflows and AI? People have already started asking LLMs to answer their investment and trading questions. The results are obviously not great. An LLM, as smart as it can be and as well-read on the topic, lacks the most important thing for making good decisions: context, and more specifically good data about what is happening right now. Without it, the model fills the gaps, often confidently yet... making absurd mistakes.

In our view, the gap between a PhD and an LLM is narrowing fast, and it is only going to get narrower. But the PhD quant at a trading house doesn't have to reinvent the wheel every time he comes to work in the morning. He doesn't have to rebuild all the models, data and insights that help the firm make the right decisions. Those data pipelines are automated and available to whichever team at the firm needs them.

Retail traders have screeners, charts and whatever their broker provides, but not that kind of infrastructure. Yet with APIs and MCP, it has never been easier to connect an LLM to the same type of highly processed data, forecasts and signals. The connection alone doesn't make anyone a good trader. But armed with that context, we believe the model has a far better shot at helping you make good trading decisions.

Answering "what is the trade?"

At Sharpe Two, our aim is to deliver the most sophisticated and comprehensive analytics an LLM could need to help guide you toward good decisions. But even that could be daunting. Some people are really happy geeking out on the most obscure metric, or recreating some of them themselves. Many just want to understand... what is the trade?

Because we must not forget the overarching need: how do we manage our money in a more dynamic way than what is currently available, and if possible without spending our entire evenings doing it?

Answering "what is the trade?" means doing part of the thinking for you. Good research should already have filtered out the noise, explained why a situation looks interesting, and said what would make it stop looking interesting. Choosing which situations to put in front of you, and putting a number on them, is judgement, and we should own that rather than pretend we are just passing data along. What it can't remove is the uncertainty. Markets move, our models can be wrong, and only you know how much you can afford to lose and whether the trade fits with everything else you hold.

Judged on what happens next

That is also why our data must be held accountable for performance. What we offer is, to the best of our knowledge and ability, a set of situations we think could be exploited by a human or... an agent. We should be judged on what actually happens to them afterwards, including the ones that don't work.

That makes for a very different relationship from the one most people have with their bank. What you pay for is access to the research and to how we read the market, and you can judge for yourself whether it is helping you. Look at the situations we flagged, check what happened to them, and make up your own mind.

If the service sucks, you should be able to stop paying for it and find someone better. That sounds obvious for most things we buy, yet in investing it has rarely been that simple.

Essentially, we put the decision power back in the hands of the customer. You are not a prisoner of some secret sauce hidden in a fund. And while our secret sauce may not be as secret as theirs, we strive to make it as good and a lot cheaper.