Below is a table containing our projected results of the 2021-22 BMW IBU World Cup season. We arrived at these results by simulating all the races of the 2020-21 season 150 000 times, adding up the points and checking how often the different athletes finished in the various positions. Read more.
You can navigate to check our predicted results for both women and men and also all the different disciplines as well as the Under 25 cup.
We have communicated with the IBU and had confirmed that the current plan is to scratch two results in the overall cup and none in the discipline cups. However it is of course possible that this will change depending on how the pandemic develops before the 2021-22 season will start.
No matter the sophistication of a model, simulating an upcoming season more than six months before it is scheduled to start, is fraught with uncertainty. Even if we are proud of, as well as very bullish about, the quality of our model’s predictions, we are well aware that it has shortcomings.
The results of our season simulations is based on the unlikely assumption that all the athletes will have the possibility to start any race. In some cases this is not realistic. To give an example, two of our model’s top five most likely contenders for the women’s under 25 cup, Ida Lien and Emilie Aagheim Kalkenberg are far from sure of being selected in the strong Norwegian women’s team. Somewhat likely both of these women will not get the chance to start the world cup races of their choice. This is something the model does not take into account.
The model has been upgraded from last season and is now in a better position to adjust for the age of the various athletes and make good estimations about which athletes are likely to progress or regress.
Still no model can be better than the input data it receives. The best we can feed our model before the season has started is result data from previous seasons as well as the age of the different athletes.
We expect the accuracy of the predictions to increase when the season is underway and the model is fed recent result data and can better evaluated which athletes have improved over the summer break.
We are continuously updating and improving our model and hope to use an improved version to make another set of predictions before the season will get going.
The way the simulations work now, which assumes that all the athletes have skill sets similar to what they had the past few seasons, removes completely the variance created by how the different athletes will development over the summer. As our season simulations at the moment is disregarding this natural source of variance, we have adjusted by increasing the expected result variance substantially when estimating the likelihood of the various positions the athletes are likely to finish in. We have however not increased the variance much at all when estimating the points they are likely to score.
As the season progresses and we have some results from the 2021-22 season to base our simulations on in addition to the recent seasons, the scale of this problem will decline and we can decrease the result variance towards the average historic levels also when simulating the athletes finishing positions in the different cups.
The older version of the model which was in use last season performed very well. An example of this is that it had Sturla Holm Lagreid as one of the main challengers to Johannes Thingnes Boe for the very first men’s race of the season. We were not at all surprised when he won even if the mainstream press wrote about a sensation and called Lagreid a “nobody”. Another example is the women's World Championship mass start. Our model saw Hauser as the main race favourite and she duly obliged.
If you are curious about the performance we suggest you check out these four articles 1,2,3,4 where we evaluated the race by race performance.
Like last year we will also publish our model’s race by race predictions.