However, I think there isn't really any logic here. A claim that something is a Poisson process should be followed by some statistical test. I wouldn't know which one, because we do not know the various probabilities and because they change over time (for example, P(Spain:whoever) seems to have dropped quite a bit recently, and nobody knows when that happened and by how much), it will be estimate them.
But a statistical test still is needed to make any kind of claim of something being a Poisson process. I guess that, if you posed a model for way in which such probabilities change, you might be able to use some ELO-like system to estimate those probabilities and, from it, do some test for Poisson distribution. I fear such a model might have so many degrees of freedom that it is too weak to prove anything. At the very least, it would be hard math to wring anything out of such a model.
However, I think there isn't really any logic here. A claim that something is a Poisson process should be followed by some statistical test. I wouldn't know which one, because we do not know the various probabilities and because they change over time (for example, P(Spain:whoever) seems to have dropped quite a bit recently, and nobody knows when that happened and by how much), it will be estimate them.
But a statistical test still is needed to make any kind of claim of something being a Poisson process. I guess that, if you posed a model for way in which such probabilities change, you might be able to use some ELO-like system to estimate those probabilities and, from it, do some test for Poisson distribution. I fear such a model might have so many degrees of freedom that it is too weak to prove anything. At the very least, it would be hard math to wring anything out of such a model.