The guide to reading charts on Almanis by Dysrupt Labs.
Probability forecasts in macroeconomics and geopolitics.
The Almanis group forecast is represented by a solid blue line. This is the probability derived from the Almanis platform on the question being forecasted by the group of forecasters.
The 95% confidence interval of the group forecast is represented by the light blue shaded boundary. These confidence intervals are approximated using historical data. As such they represent the likely range in which the forecast could move if it were to change.
The activity scale of forecasts and forecasters for the particular slice of the time series spanned by the bins. This is an ordinal scale, with 5 indicating very active (large amounts of participating forecasters or forecasts) and 1 indicating low activity. The exact numbers of forecasters and forecasts represented by each point on the activity scale is detailed in the table below.
A frequently asked question is whether we are confident in forecasts produced by less than 20 forecasters on a given question.
The answer is yes, absolutely.
The forecast volume and the unique forecasters on a question are an indication of disagreement, not necessarily of the quality of the forecast.
Of the panel of 1,000 plus forecasters on Almanis, many do not forecast on questions that they identify they have no edge in. However, all are constantly looking for opportunities to “correct” a forecast with fresh information or insight.
Almanis is a private, operator staked, prediction market platform. This means that our private panel of forecasters interact with each other in the form of trading. An Almanis forecaster is directly incentivised to correct any “error” in the most recent forecast of the panel. If the forecaster agrees with the present forecast on a given question, this will never manifest as a trade, as there is no profit that they can make on the platform.
If you would like to know more about the system that underpins forecasts on Almanis by Dysrupt Labs leave a comment or contact us at Dysrupt Labs.