Competitive Forecasting Basics
Scoring Rules
Strategy
Frequently Asked Questions
Competitive Forecasting Basics
You are competing with everyone else to contribute the most accurate forecasts as early as possible.
You make a forecast by entering a low estimate and a high estimate for the target variable.
This is the range of your forecast, and the difference between the high and low bounds is the spread.
Scoring Rules
The scoring rules are designed to shower the most rewards upon those who make the most precise forecasts ahead of other participants.
As such, your forecasts will be scored for accuracy, risk and duration.
ACCURACY: You win points if your high-low range captures the outcome, otherwise, you lose points.
- If your range captures the outcome, the number of points you win is the same wherever the outcome falls. Whether it is closer to an edge or closer to the center of your range makes no difference.
- If your range doesn't include the outcome, you lose as many points as you might have won otherwise. You also lose as many points no matter how close you range was to capturing the outcome.
RISK: The more risk you take by narrowing your high-low spread, the more points you may win or lose.
- How much risk you take is estimated by comparing the spread of your range to the average spread of everyone else's ranges during the same period.
- The number of points you may win or lose is inversely proportional to the risk you are taking. For example, you may score (or lose) twice as many points with a spread twice as narrow.
DURATION: As soon as you've entered your first high-low forecast, it starts accumulating potential reward (or loss) with every minute that passes.
- If you're taking exactly as much risk as everyone else (on average), then your forecast accumulates potential points at the rate of 100 per day, or approximately 1 point every 15 minutes.
- You can change your forecast as often as you like.
- As soon as you change your forecast, it stops accumulating potential points, its reward potential is frozen, and the new forecast starts accumulating potential points.
TOTAL SCORE: At the end of the game, when the outcome is measured, your final score will be the sum total of all the points earned and lost by each of your forecasts.
Winning Strategy
Besides obviously submitting forecasts that are as accurate as possible, you can optimize your performance by being careful to:
- Submit a forecast as soon as possible: Time is money! The earlier you start forecasting, the more points you may earn.
But be careful: a bad forecast may cost you, so it is sometimes better to stop forecasting than to make a wrong forecast.
That's why, in addition to being able to revise your forecast, you also have the option to stop forecasting for a while, and resume forecasting whenever you want.
For the period during which you have stopped forecasting, you will neither earn nor lose points.
- Revise your current forecast as soon as you have a good reason to change your mind: Each of your successive forecasts is
scored for accuracy and duration. So if some new information prompts you to change your mind, act quickly in order to
increase the duration of your new (better, more informed) forecast.
- Make your forecasts as precise as your knowledge permits: Take as much risk as you dare by narrowing your spreads, because that speeds up the rate at which you may accumulate points.
Frequently Asked Questions
How is the collective forecast computed?
At any point in time, the collective forecast is the average of all the current forecasts.
Do I earn more points if I'm right and everybody else is wrong?
Not necessarily. But if everybody else makes forecasts that are less precise than yours, your forecasts will have a greater reward potential. If your forecasts are correct, you will then earn more than others. Of course, if your forecasts are inaccurate, you will also lose more.
How many revisions to my forecast should I make?
You should change your forecast as often as you change your mind, or feel safe to narrow the range of your forecast. Each forecast is scored for accuracy
and duration, so "time is money": if some new information prompts you to change your opinion, you should act quickly in order to increase the duration of your new, better, more informed belief.
Am I penalized if I revise my forecast often?
No, there is absolutely no penalty for making revisions to your forecast.
How narrow should I make my forecast?
The potential reward - or loss - of any forecast is based on that forecast's spread (high - low) compared to the average forecast's spread during the same period. It is therefore to your advantage to make your forecast as narrow as possible. However, keep in mind that if your forecast is very narrow and the actual outcome falls outside of your forecast you will loose a lot of points instead of earning a lot.
Why would I want to put in a forecast with a wide spread versus a narrow spread?
If you aren't certain about your forecast, it may be better to enter a wide spread. You will earn fewer points if you are correct, but you will also lose fewer points if you are wrong. If you are very uncertain, it might be better to stop forecasting altogether, this way you won't lose any points (nor earn any).
