Among the other Pokemon Showdown stats that are posted monthly are lead stats, showing how often each Pokemon is led in battle. I use the different stats on here for tournament preparation regularly and thought it would be interesting to incorporate lead data in this prep. However after quickly checking the data I realized the number of pokemon led in battle is about half of what you'd expect. The sum of leads is 1/6 the sum of usage, instead of 1/3. I assume this is because the lead stats check which Pokemon is in slot 1 in each battle. But double battles have two leads on each team, and one of these isn't captured per battle.
You could say that you can just double the listed number for each pokemon, but there's no way to check the accuracy of that. Also, which slot you lead your Pokemon in does actually have implication in double battles, albeit not often. Some examples of this include when playing against ditto (which has been very viable at times) and when playing closed teamsheet against a potential shadow tag pokemon (competitive Gothitelle has top cut worlds). There's also human bias involved in this, and while I have no empirical evidence to show this, I suspect people are more likely to lead their fake out in the left slot, which would skew the data.
However, I'd argue what's more important for VGC matches is how frequently a Pokemon is brought to a battle. VGC matches are bring 6 pick 4, so you get to bench two pokemon every match. And your pokemon are decidedly not brought an even amount of the time. This means its common for teams to have "matchup check" pokemon that are brought to battles infrequently, but are very important in a small percentage of matchups, in a way that isn't as feasible in DOU or singles. On the other hand, there are pokemon that are almost brought 100% of the time when you see them on your opponent's team, and knowing that information is hugely important when deciding your own four pokemon. This information would be insightful for VGC players, and I'm sure BSS would love having that data as well.
I think the simplest implementation of this would be to add a row to the chaos JSON file indicating how often a pokemon is brought in each position (1-6) in a match. 1-2 would mean it was led, 3-4 means its brought in the back, 5-6 meant it was benched, and it'd be trivial to compile that into something usable while not requiring specific implementations for different metagames.
You could say that you can just double the listed number for each pokemon, but there's no way to check the accuracy of that. Also, which slot you lead your Pokemon in does actually have implication in double battles, albeit not often. Some examples of this include when playing against ditto (which has been very viable at times) and when playing closed teamsheet against a potential shadow tag pokemon (competitive Gothitelle has top cut worlds). There's also human bias involved in this, and while I have no empirical evidence to show this, I suspect people are more likely to lead their fake out in the left slot, which would skew the data.
However, I'd argue what's more important for VGC matches is how frequently a Pokemon is brought to a battle. VGC matches are bring 6 pick 4, so you get to bench two pokemon every match. And your pokemon are decidedly not brought an even amount of the time. This means its common for teams to have "matchup check" pokemon that are brought to battles infrequently, but are very important in a small percentage of matchups, in a way that isn't as feasible in DOU or singles. On the other hand, there are pokemon that are almost brought 100% of the time when you see them on your opponent's team, and knowing that information is hugely important when deciding your own four pokemon. This information would be insightful for VGC players, and I'm sure BSS would love having that data as well.
I think the simplest implementation of this would be to add a row to the chaos JSON file indicating how often a pokemon is brought in each position (1-6) in a match. 1-2 would mean it was led, 3-4 means its brought in the back, 5-6 meant it was benched, and it'd be trivial to compile that into something usable while not requiring specific implementations for different metagames.