Forecasting Rules

Participation

Participants will be divided into five categories: public (0), faculty/staff (1), graduates (2), third and fourth year undergraduates (3), and first and second year undergraduates (4). Participants from the same institution will register through their local manager. To be an official team and qualify for a cumulative team victory, the team must consist of a minimum of five forecasters. Although the competition is open to the public, only active faculty/staff and students (categories 1-4) may be on teams.

Schedule

A season of forecasting will consist of ten first-order ASOS stations in total. This will be split between five stations in the fall semester, and five stations in the spring. Two weeks of forecasting for a specific station (Monday through Thursday) will comprise one full period of forecasting.

Forecasting

Each forecaster will enter a forecast for three variables: high temperature, low temperature, and accumulated liquid-equivalent precipitation for a 24-hour period beginning at 0600 UTC, with forecasts due at 0000 UTC. Rather than simply issuing a deterministic forecast, you will enter a range for each of the aforementioned variables.

For example:

  • High temperature: Minimum 81, Maximum 85
  • Low temperature: Minimum 62, Maximum 64
  • Precipitation: Minimum 0.06”, Maximum 0.24”

The range forecasts represent both the uncertainty in the forecast, and the risk tolerance of the forecaster.

Competition Management

There will be no central manager of the competition. Rather, each team manager will have a role in managing the competition.

Scoring

Each forecast has a maximum of fifteen points that can be awarded (five for each of the three forecast variables). If verification falls within the forecaster’s range, the equation below is used to determine points awarded.

Forecast Scoring Formula

In the equation above, P is the points awarded for the variable, α is a constant multiplier (0.5 for the high and low temperature, and 10 for precipitation), U is the forecaster’s upper bound, and L is the forecaster’s lower bound. If the verification falls outside the forecaster’s range, zero points are awarded. If the verification falls inside the forecaster’s range, points are subtracted at 0.5 times the range for the temperature, and 10 times the range of the accumulated precipitation.

  • High temperature: Minimum 81, Maximum 85
  • Low temperature: Minimum 62, Maximum 64
  • Precipitation: Minimum 0.06”, Maximum 0.24”

Let’s say that verification for all three forecast variables fell within the ranges above. The forecaster would receive:

  • 5 – 0.5 (85 – 81) = 3 points for high temperature forecast
  • 5 – 0.5 (64 – 62) = 4 points for low temperature forecast
  • 5 – 10 (0.24 – 0.06) = 3.2 points for precipitation forecast

Thus, the forecaster receives a total of 3 + 4 + 3.2 = 10.2 points. Unlike other forecast contests, points are good. Note that in order to receive 15 points, a forecaster would have to make a perfect deterministic forecast for each variable!

The forecaster should never input a range more than 5/α (10°F for the temperature, 0.5” for the precipitation) since that will guarantee zero points. Therefore, there is incentive not to forecast too large of a range every time, but also incentive not to be overconfident since a deterministic forecast is a very high risk forecast.

Missed Forecasts

If a forecaster misses a forecast day, zero points are awarded.

Individual Period Scoring

Cumulative points will be tallied after each biweekly period. Forecasters will be ranked from most points to least. The rankings will serve as the actual score (used for cumulative scoring…see below) for the period. For example, if Brian receives 87 points and is ranked 4 forecaster category will receive a digital trophy.

Cumulative Scoring

An individual’s cumulative score will be an average of each of the city ranks, with the poorest two ranks over the entire year dropped.

Team Scoring

The team score will be an average of eligible individual year scores. At least 70% of the team must be eligible at the end of the year to qualify for team awards. This allows teams to have a small portion of their forecasters who wish to participate but cannot commit to forecasting regularly to still participate without hurting the team.