@timewalking I read the post and I get the math, my point was more philosophical in nature I think.
Numbers do not even need to reveal a problem for me, even though I believe they do when you take this article and compound it with the various ones we have had over time. The problem is rooted in player perception and feelings as much as win rate and format health, which are not dictated by stats.
A deck can be microscopic in its representation and still be an issue for the format and the game because of how it makes players feel. Let's just say there was a vintage card that was legal, of a comparable power level to shops, unrestricted, and hyper rare. Like copies numbering in the 100s as opposed to 1000s. Obviously this card would carry a huge price tag if you could even find it.
Now let's say you have a large event, like a worlds, where you get the top players from around the world bringing their A game. This card represents a unique deck archetype and a number of player bring that deck, lets say 20%, the ones who have it and/or can afford it. Over the course of the tourney the deck has about a 55% win rate against non mirror matches, puts 2 copies into the top 8, and perhaps even wins the tourney. Is that a problem? I submit that it is for the following reasons:
Feels bad - It does feel bad for some players to lose to a card they will never have an opportunity to own or play themselves. It makes the format inaccessible and lowers player confidence that the format is sustainable. It may also just be a card that is "un-fun" to play against, which is purely based on perception and not power level. Un-fun can not be quantified with win rate stats, because you can win or lose, with or against it, and it can still be un-fun, and because the scarcity of the card will make it only an issue at some events, giving tourney appearances a coin flippy nature between fun and Un-fun.
Chaos Orb is a prime example here. On its face the card is not that powerful in a vintage context, at least not powerful enough to be banned. It is banned because of the logistical problems it contains, but within those logistics are a level of feeling. Even if everyone was able bodied enough to actually play the card mechanically (IE, have functional arms, ability to stand when needed) the card feels really bad because of its randomness, both for the player who flips it funny and has it miss and for the player on the receiving end who fells like they just lost to a card that had a lucky flip or that they didn't space our their cards properly.
Why Bother Teching - Because of the cards scarcity, it raises the volatility in its appearance. Why would you tech for the card in a more local event when you know that no one who normally comes there owns the card? What happens if one person does show up with it and then the whole field is unprepared and they win as a result of the cards rarity/pricetag? Statistical analysis may show you what the best sideboard play will likely be for any given event, but a single outlier event can damage peoples perception, and perception in many ways is even more important than facts when it comes to influencing, as you can ask any politician.
Stratified tourney results - Because of the cards rarity and price tag, you would wind up with average tourneys where no one really plays the card at all, and top level tourneys like worlds where the formats meta looks and plays completely differently because the card becomes a factor. Not based on performance specifically, but price and scarcity. This skews numbers and furthers skew perception. We already have this to some extent, because budget players tend to play in smaller events that have less investment more frequently than major events. You see more blood moon builds in 16 man weekly events than at worlds. But it is generally accepted as not ideal.
With stratified results, the statics do not show the full story of the card. Once you average out all the events you would see that the deck list was not particularly overpowered because its win-rate was within expected limits, its representation is low enough not to represent significance, etc. Bringing it back to the gun debate as a metaphor, if you took every school in America, counted every day that every school did not have a gun violence event, and then showed that as a percentage, it would be very low. In America there are 131,890 Public and private schools. Students attend schools 180 days a year. There were something like 300 school shootings in 2017. Statically that means on any given school day there is a 0.0012% chance of being in a school with a shooting, an incredibly low likelihood you will be involved in one. That number is still too high however because of what it represents, even though a purely statistical analysis bears out otherwise.
Now, shops is not at this level, this is a metaphor. But when you consider the way the pricetag has been moving on the card, the physical number of copies or playsets of the card available to the world that is ever dwindling, and the nature of the game itself and how shops can make people feel like they didn't even play the game in some matches, I think it still represent a problem that it is not on the restricted list.