GS Warriors one win away from setting new NBA record for most wins to start a season

[TheRealJoeFish]

...

Baseball is sufficiently different that I think the 35-5 start of the Detroit Tigers is superior, but there's no objective way to decide this. I would also place the '03-'04 Arsenal start above the GS start despite having draws in their incredible run, but again that's in consideration of the differences in the game.
...

A reasonable mathematical way to go about this would be to look at winning percentages, and their spread.  In baseball, almost every team has a record between .400 and .600; you can extrapolate from this that even the best team would beat the worst team only about 2/3 of the time.  In basketball, the records generally vary between .800 and .400, so the best team will beat the worst team maybe 9 out of every 10 times.

So, you could go something along the lines of, very roughly, the chance you win 23 games in a row with a 80% win probability vs. the chance you win 35 out of 40 with a 60%  probability.

The first is (4/5)^23, which is about 1 in 170 chance.

The second is (3/5)^35 * (2/5)^5 * (40 choose 5), which is 1 in 8,633.

So, based on that quick calculation (and given the rather large difference in magnitude of the outcomes), I would say that, yes, it's easier for a really good NBA team to start 23-0 than it is for a really good baseball team to start 35-5, just based on the differences in the sports.

That calculation does make the difference look a lot bigger than it is, though, because it assumes that every team a good NBA team plays has a 20% chance of beating them and every team a good MLB team plays has a 40% chance of beating them.  In reality, the greater variations in game-to-game probabilities (ie, the Spurs might have a 48% chance of beating the Warriors in San Antonio and the 76ers would have a .02% chance of beating the Warriors in Oakland) in basketball compared to baseball are going to minimize the impressiveness of a basketball run if you assume (as I did) that the day-to-day probabilities are uniform.