So far on almanis we’ve been forecasting the probabilities of specific events. Lately, some of you might have noticed we’ve begun to ask a new type of question in which we’re forecasting values for known parameters. It allows the crowd to forecast statistical releases, for instance. Or how many seats a party will win in parliament, rather than just whether it will win or not.
The process for forecasting these questions is a bit different so, before getting stuck in, let’s take a couple of minutes to go through it all.
Below is an example of a forecast I was making on our question on April’s US Nonfarm Payroll release. For the uninitiated, this is the monthly change in the number of employed persons in the US. Lately the monthly numbers have been averaging around 234k with a range of 149k-295k over the past year.
Rather than being asked for a point forecast, we are asking you to make a forecast range in which the actual number will hopefully fall. You input your desired range into the two boxes on the left marked “LOWER/UPPER RANGE BOUNDARY”. Once you’ve done this, type the number of points you would like to invest in your forecast in the box on the right-hand side.
We then get to the all-important return which is automatically calculated and displayed in the bottom left. In this case, my 20-point investment would generate 35.7-points in the event of April’s Nonfarm Payroll number coming out within my 220k-230k range – a return of 78.5%.
So how is your return calculated? There are two factors, the first is the concentration of forecasts already focusing on your given range. I’ve gone for a fairly conservative estimate range, right around the recent average where others have also forecasted and so this will give me a fairly low return. However, I’ve also gone for a fairly narrow range of just 10k. Given there’s been a 144k range shown over the past year, that’s quite punchy and my return is increased to account for this.
By way of example, if I were to double my range while keeping it centred on my initial conservative estimate (215k-235k), my potential return would shrink to 54.6%. But obviously there would be a higher chance of me realising that return.
If I were to keep my forecast range to 190k-200k, keeping my punchy 10k range but move the centre away from the long-term average where most forecasts are currently focused, my potential return would increase to 112.3%.
If you have any questions about this, just contact the almanis curators.