Following on my comments from a few days ago, any start-up that wants to attack the securities research space is going to face a few challenges.

First, is that there are already so many firms churning out research. The director of research of at one of the largest hedge funds once told me he had over 300+ brokers providing his team research. Over the past few years, I have encountered a lot of smaller brokers, and have been amazed at how many there are. Anyone who wants to tap into this market has to recognize that providing more research is not going to be the winning model.

Second, much of the research that is coming out is bad. There I said it. The truth is, if you ask any analyst they will tell you the same. Even good analysts do not always put out good research. This is an attention-driven business, and just like any ad-driven website out there, this means that analysts have to publish constantly, even when they do not have anything new or useful to say. So I still see analysts publishing daily notes with highlights of yesterday’s news. I do not think many people read it, but I understand why the analysts publish it. Couple this to the fact, that at many smaller firms, analysts are paid based on how much commission their trading desk does in the specific stocks they cover. This is a perverse incentive that drives many analysts to publish loud, attention-getting notes; and a disincentive to publish longer, more thoughtful pieces. Better to be noticed and wrong today, than forgotten and right a month from now. When you look at like this, your realize how similar financial research is to the broader web publishing industry.

A third problem is that there are no easy tools for solving the issue. Web-based media can apply all sorts of algorithms to optimize readership because the numbers of views they get are so large that statistical models work. For financial research, the target audience is small by web standards, with a huge array of things they care about. I think of this is a “columns and rows” problem. A typical, large web site measures millions of viewers on a dozen metrics – how long they stay on the site, what screen position they view first and most, etc. This is a database with millions of rows but only a few columns. By contrast, investors care about dozens or hundreds of stocks, have different extra requests (for different financial metrics, industry data, etc.) and any number of other subtelties, but there are only so many investors out there. This is a database with a thousand columns and only a hundred rows. My sense is that the data tools out there are not really optimized to solve these kinds of problems, and the sample sizes (i.e. the audience) is so small that statistical modeling does not yield great results.

Another problem is that research is extremely perishable. If an analyst has some hot new data, it gets disseminated in the market extremely quickly. Investors trade on it within hours if not minutes of its release, and within 24 hours the data is literally and figuratively yesterday’s news, and thus of no value. This is not only true of specific data, but of broader ideas or patterns of analysis. If an analyst figures out some clever new data source, investors find out what it is quickly and move to replicate it.

Finally, the biggest problem is that almost no one pays directly for research. The brokers “give it away” for customers who pay via trading commissions. This does not entirely eliminate subscription models, but it does make them harder. Over the years, I have known many analysts who have tried to hang their own shingle, but have struggled to find clients willing to pay hard dollars for a subscription. There are many of these shops who hold out hope for regulatory change. And every few years, we see headlines about some large investor who is moving to a model where they pay for trading and research separately. For the time being, anyone providing research has to contend with hundreds of competitors whose price is ‘free’.

All this sounds rather bleak, but I think there is a more optimistic conclusion to draw. As I mentioned in the last post, research is just another form of content. There are a lot of media companies struggling with the same problems, and many of the successful business models employed elsewhere, can be used for financial research. There are still plenty of technical problems to iron out, but also many potential solutions.

I plan to explore these more in my next post, but let me give one example. The way I see it, financial research is screaming for a freemium model of some sort. Someone should offer tiered research. Longer, less “actionable” notes (notes which do not demand immediate stock buying decisions) should be free. Clients could then judge the quality of the analyst’s knowledge and pay for the trading ideas, and pay for access to detailed financial models, and access to the analyst on the phone or in person. There are a few firms that are working on variations of this, but so far no one  has explicitly tried to copy any of the web-based models. My guess is that the next successful business to emerge in this space will not be a research provider, but a clever distributor of other people’s research.

My point is that technology and the web have the potential to make financial research viable again, and we are in very early stages of that happening.