Given what we have seen from Andy against Gasquet and Sandgren I think it is perfectly reasonable to be looking for wins plural here and even thinking about going all the way, although a match vs Gombos could be a decent test.
Does TennisAbstact do its analysis purely on objective measures? I was just thinking that Andy would be an interesting test for these.
I may be wrong, but I think they take the ELO ratings they assign players for a given surface and then run 1000 simulations for each match up in each round, and use the results of those simulations to do their predictions. Each time a result comes in, they re run all the tournie simulations and change the predictions.
Specifically, they say to generate forecasts for a specific matchup, use a 50/50 blend of overall Elo and surface-specific Elo. These 50/50 blends are shown in the table as "hElo," "cElo,", and "gElo." The 'default' match type is best-of-three, so in a best-of-five match, the favorite will have a better chance of winning, by a factor that depends on the best-of-three odds.
Andy doesnt seem to appear in the current listing, so not sure how they allocate him an ELO, but overall they rate him as the 7th best player of all time at his peak and the 7th best on hardcourts at his peak