Hairyducked Idiot

Then what is it?

Only if they were expected to go 6-10 on average, so that 10-6 results in four wins of variance. But yes, *every* stretch of games should change your projected final result by the amount they varied from expectations. If you have a precisely .500 team on paper, every time they win that should add 0.5 wins to your projected total, and every time they lose it should subtract 0.5 wins from it. (Obviously you could make that model more complicated by adding in strength of schedule, and Fangraphs even has a nice page where they do something similar but also update their estimate of underlying talent level in real time, but strength of schedule just isn't a strong enough factor in baseball games to adjustments like that meaningfully change the answer over small samples).

OK, that's wrong. Not in an "I have an opinion, you have an opinion" sort of way. It's objectively wrong in a way that a statistics professor would mark your answer incorrect on an exam sort of way. We're gonna have to go back into some academic-level statistics to explain why. A true-talent 78 win team will expect to win 48.1% of their games. But even if God Himself tells you that the team's true talent level is 78 wins, they won't win exactly 78 games every season. Or even most seasons. That's not how probability works. In mathematical terms, their observed win percentage will trend toward their true win percentage as the number of games stretches forward toward infinity, but that doesn't mean it will hit that line at exactly 162 game. This is the same thing as God Himself telling you that a coin is a true 50/50 coin, but you will still frequently get 6 heads and 4 tails in any 10 flips. The observed rate will trend toward 50/50 as you approach infinite flips, but it will frequently diverge from 50/50 for long stretches. Those divergences are what we call "variance." Expected variance approaches the limit of 0 as iterations approaches infinity, but it never touches zero because iterations are never infinite. If you flip a 50/50 coin 162 times, you expect to get 81 heads and 81 tails, but it's extremely probable that you will observe variance. If you start out with 10 heads and 6 tails, then in order to get to 71-81, you'd have to 71-75. If you *expect* that to happen, that means you're expecting the properties of the iniminate coin to somehow have knowledge of how the past flips have gone and adjust physics so that the coin is no longer 50/50. Coins do not have knowledge and cannot change the laws of physics, so this cannot be true. Expecting probability going forward to change in order to revert observed variance is a human cognitive bias known as "gambler's fallacy." Our brains are sometimes wired to think that way, but it's wrong. If God Himself tells you that a team has a true-talent level of 78 wins and they start out 10-6, absent any other information, the correct answer to how many games they are now expected to win is (rounded off) 80. None of this has anything to do with your original argument, which was about strength of schedule, so I *think* this might be one of those situations where I'm not supposed to be taking your answers literally but generally you're just expressing the vibe that you think I'm wrong and the actual arguments you use aren't supposed to matter.

If you have a team that you are told by God Himself has a true talent level of 78 wins per 162, and they start the season 10-6, but you have no other information, how many games would you project them to win for that season?

In a sport where almost every team is between .425 and .575 true talent, and you play a mixed schedule, the odds of this making a non-negligible difference in the analysis are minuscule. You can do all that math if you want, but if it changes your result over 16 games more than a small fraction of a win, i would be rechecking the math.

How many games should a 78-win true talent team win, in average, if you know they started 10-6? If you say 78, then that's gambler's fallacy.

Yes. If you flip a coin 100 times, you expect 50 heads and 50 tails. If you start out with 8 heads and 2 tails, your new expectation for the 100 flips is 53-47. You still expect to go 45-45 the rest of the way, but the banked results have to be accounted for. 16 games isn't much sample size to start changing your prior expectations for a baseball team. You could probably tease out something meaningful if you really worked at getting down to fast-stabilizing metrics, but that's a lot of work for minimal gains on something so trivial For message board posters, saying "16 games isn't enough to change my opinion of the team, but the fact that we have a few extra wins already puts us on pace to be a little better than expected" is perfectly reasonable.

I feel like this must be one of those situations where I've accidentally implied something very different than the literal meaning of my words, because this makes absolutely no sense to me as a response to my proposal, which was that banked wins can be added to original projections for the rest of the season rather than committing gambler's fallacy and assuming the team would lose extra games to reach the original expectation. You're talking about stabilizing results for future predictions when my idea specifically avoids trying to make updated future predictions.

But ... my idea didn't include run differential, so why would we do that?

Implying a bet? How on earth would you even try to measure that? Do we have access to unlimited alternate universes where we can identify thousands of teams with 75-win projections that start 10-6 and see if their final win total ends up closer to 78 or whatever we calculate with a more complicated formula?

Well then you're just incorrect. It's perfectly fine for back of the napkin message board posting and acknowledging that refusing to include early results in your final exoectation is gambler's fallacy.

It's extremely simplistic. But the differences in baseball teams are small enough that making it more complicated would give you pretty limited gains in accuracy.

16 games should statistically have a small impact on your assessment of team quality. Not zero, especially with the run differential, but not a ton. But if you thought this was ~75-win team before the season, they've now banked 3 extra wins, so you would project them to 78 wins without changing your opinion on their talent. Fwiw, fangraphs now has them projected to finish 80-82 with a 27% chance of making the playoffs

I know it's a hot streak but dang does wisdom smash baseballs

Darnell Nurse had himself a night On the tying goal in the closing seconds, he had the guy in front of the net tied up. Pass comes through to the back door and he panics and lunges toward that guy, which is 100% the wrong thing to do. That guy got the shot off before he had even turned around, he had no chance of influencing that part of the play. But the guy he left wide open at the top of the crease puts it the rebound. You have to trust your goalie to make the save and protect him from the rebound. The ot goal, he got caught in the weirdest no man's land. I understand the pk formation they were playing requires him to stay loose and not just go one-on-one the shooter in the middle, but there's no way he was supposed to be so far back that his feet were in the blue paint. Second goalie isn't a position.

I thought that the dumba hit on pavelski was pretty clean and just unfortunate that Pavelski fell awkwardly and hit his head on the ice. It was the fall that did the sickening damage, not the impact of the hit. Only thing I wish is that they would eventually legislate out the "exploding upward" style of hot. Dumba comes in low and makes contact with the shoulder, but as soon as contact is made he drives upwards with his legs to explode through the hit. Those kinds of hits are legal, but they add way more force than is needed to separate player from puck and frequently (though not this time) lead to head contact

Apparently the answer is yes

All the top guys on ol seem like "maybe could be a good starting tackle but would need to hit upside to get there" and you can usually find that later too.. I would fine with taking one of the top ol guys. Just not my literal first choice

Really not interested in a question-mark laden DT at 9. Trade down would be fine with me. Maybe I'm being dumb but I really want JSN. An unguardable slot receiver gives this offense a dimension it doesn't have now with our three good outside threats. Offensive line would be my second choice. But it seems like all the options there are flawed by top-10 pick standards and we can do just as well there in the 2nd round.

I like Yan Gomes. He hits home runs for the Cubs.

Case by case basis. If I'm extending I want short term and I want a big discount

Even though I'm a Ducks and Blackhawks fan, I hope bedard goes to some random team near the back of the top 10. Gimme them tank-happy tears

Nobody's ever happy with the draft odds. It's like playoff formats

Matt mervis is a bat-only first baseman who still hasn't made his MLB debut when he turns 25 this weekend. We own him through age 31. We will be ok if he doesn't extend

Re: toews, it's really hard for me to let go of what happened and while he wasn't directly a part of it, I absolutely believe he's lying when he says he didn't know about the harassment that came after. And his reaction when the news broke was absolute trash. I really want every single piece that was in any way connected to that era to go away. Re: draft position, this is a really good draft to be locked into top-4. From what I understand reading reports of people who are better scouts than me, there is one McDavid-level talent in this draft and three more that would have a case to be top overall picks in other years, sort of eichel-level. Obviously fourth means you might have to stomach the Russian situation but that's probably worth it over settling for the next tier down.