Topics like this always remind me of fundamental misunderstandings about statistics--and I know most of you know these things. That said, even when we know them, the results are often jarring.
As an example: If there's a 30% chance of something happenings, then that thing happening is an expected outcome. 3 out of 10 times is a lot of times. The same is true of a 5% chance; even if its a less frequent outcome, it is still an expected outcome. In sports it's often easier to see the inputs that lead to such an outcome: particularly egregious fouls (or referee's calls), injuries, sub-standard play by key players, etc.
The frequency of the event impacts our reaction to the statistics.
In my profession I see this a lot: "We have a 70% chance of winning at trial." This is a single event, so the reaction is often something like, "Awesome, we're going to win!"
But when we go to the baseball game: "This guy hits .300." There are tens of thousands of at-bats per season, so: "So glad he's at the plate for us!"
Same person reacting to the same odds drawing exactly opposite conclusions. (And that's not even touching how subjective a lot of these odds are.)
And of course, when there is a 90% chance of something happening, but it doesn't happen, that doesn't mean the 90% was wrong.
In politics people often confuse the odds of an event, with the percent of the vote. A 5-point win in a national election (with approximately 130 million voters) is a blowout, but receiving 52.5% of the vote doesn't mean the winner had a 52.5% chance of winning.
And the casino business model is pretty straight forward: it's math. It requires capital to float the math, but with tens of thousands of events (depending on the time period you're measuring), the casino wants you to win because that will convince the other people to come (and lose). While 60% is a coin flip on a single game, 52% is a lot of profit over the course of thousands of events (and ~48% is basically the best odds you'll get in a casino).
So, in the roulette example, assuming the statistics presented earlier were accurate (and it's been a long time since I've done any of that math), with a 94% chance that you win $50, there's a 6 percent chance that you lose $1000. Winning $50 94 times means you've won $4,700. Losing $1000 6 times means you lose $6000. So for every 100 tries at this theory, the casino wins $1300. And that's a theory in which the winners show restraint, and walk away after their 5% payday on $1000. Few people head to the casino to make 5%.
Sports betting has always been a mystery to me because I don't understand how they set the line (I'm sure I could learn it, I just don't care that much), but regardless of the inputs, I'm confident in saying that the casinos know how to do it such that they encourage a lot of winning--just so long as it's a little less than the losing. Your local, college bookie may not have the capital to float the losses, but the big kids do, and they rely on it to keep raking in the dough--and building more capital.
