weatherbill

I've been trying to come up with something eloquent to say about weatherbill.com, but I'm just blown away: it's just so damn cool! It embodies everything that I see as driving the future of online marketplaces. Just as intrade.com is giving economic derivatives a run for their money, I see this trend continuing with weatherbill as compared to the more buttoned down offerrings at CME.

Working at Root Markets I can't truly appreciate the perspective of soemone who is taking on a pre-existing market place and attempting to replace it with something much better. I do, however, have some theories on what it takes to succeed.

Last year I was lucky enough to spend some time with one of the founders of intrade, and I was curious to learn more about what was driving their success against the incumbet economic statistics market. The lesson was simple: you need to offer contracts that serve real needs. The message that I heard was that the payoff structure of the CME Auctions Markets were not successful in doing so. Skimming over the contract specifications and current volumes, this doesn't appear to be the case now - but it was back then. The point still remains - serve a real need - design to hedge a real risk.

Closer to home, for me, is the mortgage market. An early story that our chairman shared with me was of the evolution of the CMO market. Prior to securitization there was very thin trading in whole loans as it was commonly believed that to understand your risk exposure you needed to understand the plethora of intangible factors that could possibly drive prepayment and default risk. Buying a whole loan was thought to expose you to the risk of what side of the street the house was on, etc. Of course, with the tranche structure of the new instruments, these risks fall out leaving the holder of a CMO with exposure to the right kind of risks. Understandable systematic risks. Portfolio theory 101 at work.

As the market evolved the models improved as can be see if you compare the book for a recent issue to that of a 1980's era issue. Nowadays, with liquidity chasing liquidity, savvy investors are seeking more specialized exposure and are slicing and these tranches looking for an edge and rebundling whole loans where once the market was super thin. Specialist funds are able to do this because of the modelling power afforded by new technology. The computer I am writing this on can do a gazillion calculations per second. Thems alotta calculations.

Compare the ability to slice, dice and model on weatherbill with the handful of contracts available on the CME. If I am a small business owner, I don't want to hire someone with a degree in financial engineering to work out how to best hedge my risks. And even then, the hedge will be imperfect when there are only a limited number of contracts to choose from. Businesses want to be able to trade specialized and differentiated contracts to best approximate their risk exposure. Yes it may be difficult for an army of broker dealers to make a market in millions of contracts, but this is the sort of problems that computing power can be thrown at. If you have the right pricing models, you should be able to scale to a silly number of contracts far more cheaply than if you had to hire high rent dealers to make the market.

This sort of scalability is not afforded to all types of markets. In futures markets, where the cost of carry is infeasible, a prediction market forms. This is the case for weather, and likewise it is the case for consumer leads (my speciality). Prediction markets are special in that there is no first-order linkage between trading activity on the derivative and the current spot price. Compare this to storable goods: if the spot price today decreases, I can always buy a little more today and hold the good for use later thus changing the value of a forward contract, and vice versa. However, if it is hot today, I can't store the weather for the purpose of changing the climate in 6 months time. If the price of leads decreases today, the rapid half life of consumer attention means that a marketer will have a difficult time realizing any value from a 6 month old lead, thus leads, like weather, can't be stored.

To understand why this is relevant in determining whether a market can scale to a large number of differentiated contracts we need to look at how market makers work. If I buy a contract for a the future delivery of some good, the market maker must find some way to hedge her risk of being the counterparty to your contract. And when the volume of sales effects the spot price, liquidity risk is present. In prediction markets you don't have to balance the effect of future market activity against spot underlying values. If you have an edge in pricing future movements, you can scale with diminished exposures to liquidity risk.

In short, prediction markets are different to standard futures markets. And there is little reason to not offer highly specialized contracts to participants. Weatherbill does this very well.

By offering the ability to upload historical revenue numbers, weatherbill makes it easy for unsophisticated businesses to asses their weather exposure and to specify a contract that exactly matches those risks. It is just so easy to use and there is something, dare I say, very Web 2.0 about it all. While it is technology that empowers this, weatherbill different from, say, hedgestreet in that there is amazing simplicity to what is otherwise a complex product.

It's just so damn cool!