The Problem: Data
Since the inception of the weather market there has been a balancing act between using "tradable weather stations" and basing a hedge on what the client really needs.
All too often the quality of data required for pricing and settlement does not exist. Many transactions never come to fruition simply because data is not available.
Sometimes transactions are struck on alternative sites, which introduces unwanted basis risk to the hedger given a less than ideal correlation with their actual exposure.
Another component of the data problem is the potential for tampering with observations, moral hazard. Many over-the-counter (OTC) contracts are written on second and third tier
station networks that do not have the security of the first order sites which are often located at airports.
- Long time series of the weather variable allowing accurate pricing
- Reliable time series. Discontinuities may exist, but ideally the history should be adjusted for these.
- Few missing data: missing data are inevitable. Provision of a cleaned and filled data history should be considered.
- Reference site should correspond to the hedger’s risk.
- Provision of sub-daily data may help improve the efficacy of the hedge and reduce its cost.
- A "high integrity" reference site: one that is perceived as being unlikely to be tampered with and which will be reliable.
- A site which is not about to be moved or where instruments are to be upgraded.
- A site where fall-back measurements can be relied upon if the primary site has reporting problems.
- Real-time data updates to allow rapid settlement of the contract
Speedwell: How we can Help
- Having worked in the weather risk market since 1999, we understand what market participants require. Larger transactions that involve
the movement of risk to the secondary market will often benefit from an independent third party
verifying observations. Speedwell is the dominant provider of settlement data for OTC weather risk contracts.
- Speedwell has access to large library of tens of thousands of weather data stations around the world. We are therefore able to help select
appropriate sites based on criteria from client or hedge provider relating to location, data
completeness, update frequency, availability of hourly data, etc.
- Speedwell can help ensure that weather data used for transactions satisfies the requirements for both hedger and risk provider.
- Speedwell can act as an independent third party focused on guaranteeing the reliability and quality of the observations.
While respecting the need for transparency, we understand that market participants (end users and risk takers)
require a complete understanding of our data processing. We understand the importance of quality, provenance and auditability.
- We are able to install weather stations to act as fall-backs and to address concerns relating to possible moral hazard.
Speedwell’s Role in Addressing Data Problems
#1 Missing / erroneous historical data
Problem: Without reliable / accurate historical data pricing is difficult if not impossible.
Solution: Speedwell’s cleaning / backfilling methodologies can make previously unusable data usable. Beginning with the highest quality data available, a series of quality control checks compares observed values against surrounding
stations and the station itself, suspect values are flagged for additional review. Erroneous observations and missing data are then filled using modelled observations.
#2 On-going feeds for deal monitoring and settlement
Problem: Unreliable (and potentially erroneous) data feeds cannot be used for settlement.
Solution: Using technologies developed to create settlement values for the Chicago Mercantile Exchange, Speedwell can produce accurate and reliable settlement
feeds for most stations. All techniques are transparent to the user and can be pre-agreed in contract settlement language. Approaches to ensuring data
integrity include "surrounding station analysis", supplementing observations with forecast data and installing back-up weather stations.
#3 Moral hazard risk
Problem: Weather settlement values can be subject to moral hazard.
Solution: Prevention and detection of tampering can involve "statistical security", physical security, or a combination of both. Statistical
security involves the analysis of the inter-relationship between observations. Departures from known relationships can be cause for concern
(for example, rainfall observed with no increase in relative humidity). Methodologies also involve the analysis of observations as compared with
surrounding observations (possible station installation). Physical security involves the implementation of barriers to tampering (electric fence,
advanced instrument mounting, inaccessible locations) as well as monitoring of a site (motion detectors, cameras, security personnel).
The end product is a dataset produced by an independent third party acceptable to all parties.
#4 No historical data for new stations
Problem: Newly installed weather stations have little to no historical observations for pricing.
Solution: Advanced data modelling techniques allows for the creation of simulated data. Inputs of surrounding station data, re-analysis data fields,
and meteorological know-how all contribute to building a representative dataset for a
new station. As with all Speedwell products, the methodology is fully disclosed and agreed upon by all clients.
#5 Delayed settlement feeds
Problem: A common problem with observing sites is a delay in obtaining observations.
Solution: There are numerous techniques available for producing preliminary settlement values:
- Statistical modelling of an observation from surrounding stations
- Supplementing observations with surrounding readings
- Using forecast data as ground truth
- Using Reanalysis data