based on what do you say it was 50 million sources?

That particular point came from a 2006 profile of the company at TechCrunch. (http://www.techcrunch.com/2006/10/30/monitor110-raises-11m-more-for-market-monitoring/) Even if the figure isn’t precise, I think it’s on the right scale. The company’s business was predicated on scanning *a lot* of sources.

Ultimately, this was key to their downfall.  Whether they were scanning 50 million or 500 million sources, it didn’t matter.  Their technology couldn’t determine what information was relevant to the investor and what as junk.  The result was just a whole lot of noise.

With the web a big, unstructured repository of global information (that is growing), it is premature to say that there is no value in the concept Monitor110 was hyping.  Institutional investors, who have been known to pay money to speak to local retail store managers to see if they can gain an information edge on market trends, know there is valuable information out there that they are missing amongst the noise.

There are firms, who may not have received the same fanfare as Monitor110, that appear to be executing on this idea with some success (see FirstRain, Connotate and InfoNgen as starters).  Just because the over-hyped Monitor110 story has come to an end doesn’t mean there isn’t a need in the marketplace.

Yeah, there’s a few doing similar stuff. SkyGrid is another one that’s gotten some attention. I see it kind of like with the Semantic search companies. Someone might get rich if they can figure out how to actually automate the understanding of content. In the meantime, most will die tryin’.

From what I have heard, SkyGrid may be on the Monitor110 track more so than the other firms in the space....

The key to automating the understanding of the content seems to be less in having a slick interface per se or even in the ability to find stories from X # of sources, but making the unstructured information easier to analyze.  This requires an inherent understanding of the investment research process and workflow and then the ability to deliver applications that make it very quick to see patterns in seemingly unrelated information.

It is not possible to “automate the the understanding of content” as you put it.  This is the general natural language understanding and AI problem and is not much closer to being solved than when the field began in the 1950s.

The problem with a number of financial unstructured content start ups is they are taking too much of an abstract view of the problem.  Sentiment detection is a case in point—again, very challenging problem, poor accuracy.  Thare no secret “big solutions” and it is not do with computing power.  It’s simply a very difficult, unsolved problem.

I agree with An Insider: there is a lot which can be done and is being done in this field (our own company included).  The key is to identify more specific problems of unstructured content.  Often these relate to the particular trading strategy in use and the function involved (trading vs. modelling vs. valuation, for example).  If you take the cross-product of trading strategies and available techniques, you come up with a rich set of opportunities.

P.T. Barnum once said"I don’t care what they write about me as long as they spell my name right.”

So the spelling is “W. Brennan Carley”, not “Brendan”.

On a more substantive note, I would like to clarify a few things:

- While the company once did try to consume tens/hundreds of millions of sources, we recognized that the signal to noise ratio was poor, and changed the approach to a hand-selected list of about 21,000 high quality sources.  From those we actually did extract good quality, relevant content.

- It is very hard, if not impossible, to automate understanding of content, and as others have pointed out, if you could you should start your own fund.  Not to mention that portfolio managers don’t outsource Alpha generation.  Which is why I am very skeptical of vendors offering sentiment or similar signal generation techniques.  Monitor110 was not trying to automate content understanding, we were NOT doing sentiment or even trying.

- What Monitor110 WAS doing, admittedly after a strategic re-focusing that I initiated when I joined in January, was taking a hand-selected list of sources, extracting relevant content from those in real-time, categorizing that content by company and by investment theme (e.g. macro topics that are typically catalysts for movement in a stock, such as mergers, supply-chain issues, regulatory, technology change, etc.) and delivering those to institutional investors.

- This is generally not “tradable” information, and so the target market is not traders (sexy as that might sound) but portfolio managers and research analysts.  The value to them is access to information that is well categorized and intelligently filtered, which saves them time and allows them to efficiently broaden their research coverage.

Hope that provides a more clear and accurate story of what Monitor110 developed.

Unfortunately, by the time we did that refocus, launched product (real product, not demoware), and got into the market, we ran out of runway, and were looking for funding in the middle of a very challenging market.

Hey Brennan
Sorry..corrected your name spelling..thanks!

Brennan, thanks for weighing in.  Must have been very tough to come in at such a late stage and try to pull a “rabbit out of the hat”, if you will. 

You obviously saw some potential in this type of service when you joined, and must have gotten to know the competitive landscape pretty well.  Curious about who you think IS on the right track - or a similar track you may have taken if you had more runway.

Thanks.

I just want to comment since I have been a witness on the sidelines (in the place where the actual research takes place) in what this company did wrong. They were trying to hard to market one product that was as Nick H put it, impossible to create with regards to current technology. On the other hand, they had another product in the form of reports which were distilled by a handful of researchers. These reports actually beat the main stream information spread which makes these reports much more valuable to investors as these reports provide a set of opportunities that could be diamonds in the rough.

Gau, while I agree that it is probably impossible (as Nick H says) to automate the understanding of content, it certainly IS possible to automate the filtering of content to extract content that is relevant to particular companies and particular investment topics.  That is what the web-based service actually did.

You are right that the company also developed a “research report” product.  That product was produced by analysts/researchers who actually used the web-based service to identify the most interesting information.... which they then filtered using human intelligence and summarized.  (This was the “Alpha Desk” product, and it certainly does have more value to most buy-side institutions.)

In the end, as noted by Integrity Research (http://integrityresearch.blogdrive.com/archive/1165.html), this made the business look more like a “research” business, while the investors had invested in a “technology” business, and it turns out that funding is a lot harder to get for research businesses than technology businesses.

Thanks for your comments “An Insider”.  The last 7 months have been challenging, but I guess I am a glutton for punishment, as I actually enjoy this stuff!  As far as saying who is on the right track, I can’t comment on specific companies, but I think what you will see is this class of technology bundled into distribution platforms that already have broad reach, more as “value add” than as a discrete revenue stream.  i.e. On the one extreme you will see firms like Bloomberg, Thomson/Reuters, Factset, Dow Jones, S&P;, etc. bundle this kind of capability into their standard desktops, and at the other extreme you will see consumer focused companies like Google, Yahoo, Microsoft, and even “traditional” media imbed this kind of service in their free web portals.

Anyone who is truly successful in automating any degree of understanding of content will use it to run a hedge fund.