We understood you implemented a large loss capping (or “market flat reinsurance”) to avoid too much swing in the leaderboards (which we fully support, in fact it was quite a frustrating issue which we identified as well)
But can you disclose at what amount the large losses are capped (or point us where you disclosed it - sorry if we missed!)? because - as you can imagine - this can impact our prices significantly, as currently they are trained based on the observed large loss experience which is not capped …
We actually have not begun doing this yet!
The reason you did not see the same effect as in week 4, in week 5 is that there was no very large loss in the data fro week 5
I will be making an announcement later today or early tomorrow about the cap and we will tell you the exact level as well
yep - an idea would be also to make the reinsurance “cost”, in the sense that - instead of capping the large losses - you could offer different reinsurance retentions at different prices …
In other words, it is up to the players to decide whether to “sacrifice” some of their profit to avoid large losses and the “level” of retention depends on their risk appetite …
I think this would make much more sense instead of capping flat the losses … and it is how really work real life!
Hi @alfarzan - we really thought this also with the team and yes: capping only large losses for “free” doesn’t make sense, better keep them inside anyway - eventually we are playing on 10k more or less markets…
… and if you want really to cap, to make realistic, you need to make the players pay - this is how reality works … otherwise we risk to go back to a fake (simulated) world
Yes this is true that it doesn’t make sense to do it for “free” however, large losses that appear in the real world are no where near the scale of what we had in the dataset given the entire market sizes.
So for example the largest loss in week 6 would have accounted for 25% of the revenue of the whole market, not just one company. That is unrealistic for any given market even with reinsurance. The reason our values are a little different to the real world is because real world markets are much larger than a few thousand policies (even tens of thousands).
So we’ve tried to balance this out by also capping the claims while taking inspiration from the real world on where to set the cap. See this announcement for the details!