On predictions

The Nats won yesterday, but since the Dodgers won as well HFA is still up in the air, though heavily favoring the Nats. The most likely scenarios for the weekend are the Nats win 2 over the Marlins, who they are better than, and the Dodgers win 1 or 2 over the fighting for their lives Giants. That would be enough. I will worry though if the Nats lose tonight and the Dodgers win.

Really I only want to see one thing tonight. Bryce playing. I know some of you think he should sit as long as it's slippery out there because he coud aggravate his injury. Me, I think he should get a few games in because it's a long wait until the NLDS starts next week. Plus if you're that worried about rain-caused injuries to susceptible players I have a first-baseman, third-baseman, and left fielder that I need to talk to you about.

Murphy? He's not playing. They have him as ready for the NLDS, which is good, but not great. First, because with each game he's out the next game becomes suspect. If he plays in this series you can be fairly certain he will play in the NLDS. If he doesn't, you can't. We'll get updates next week that might clear it up - he'll run or play in a fake game or something, but on Day 162 you can't say for sure you'll see him in the NLDS and that's scary.  Second, assuming he is ready he'll go from sitting for 3 weeks to facing Clayton Kershaw. Good luck with that. There's normally an adjustment time coming back from injuries and Murphy will have none. Even if it takes Murphy just a couple games to get back to MVP level that could put the Nats in a hole that they can't climb out of.

OK on to more fun things, like mocking me (sort of)

When the season started and we were putting up predictions I put up my own. Nats win 90. Miss playoffs.  Instead, they'll win like 95 and the NL East. Oops!

Now if you want to stop there and make fun of me I guess that's your right. But I'm going to do a little deeper look at what I said to see how bad my prediction really was. Here's a teaser - the Nats win total would not be what I consider my worst prediction.

So being more specific what I did was basically say three things and add a caveat. The three things were :

• The Nats would win 90 games
• The Mets would win 91 games
• The 2nd WC would win 91 games.

The caveat was "if anything bad happens to the Mets (and nothing to the Nats, or really more bad to the Mets than the Nats) the Nats should have an easy time winning the division."

Basically all predictions should be made expecting even luck with injuries for the teams involved. Anything else would be akin to wild guessing. This is not the same as saying "all teams will have the same amount of injuries". Some teams (older, injury-prone players, pitching dependent) are more likely to be affected by injuries that others.What it means is that these teams have the amount of missed time due to injuries you could reasonably expect given their make-up. For the Nats that's about what happened*. For the Mets things were worse. So taking the caveat into the equation the Nats season played out as I expected it would if the Mets were to suffer more injury losses than the Nats. The Nats would easily win the NL East. They did.

Now if you asked me what it meant for win totals I'd toss a couple more on for the Nats - say 92 and cut a handful from the Mets - say 86. For the Mets that's going to be close to dead on, but I'm still enough away from the Nats total, especially given the late season injury rush costing a win or two, that I do feel like I undersold them a bit. We'll go over it sometime after the season as a whole but I think, reading through the projection, that I didn't get the pitching right. Most of that was not figuring Roark for the year he had and pen being better than I thought. Offensively it feels like the surprises balanced the disappointments, so the read I had on that stayed pretty solid.

So, in my estimation, I had the Mets about right and undersold the Nats a couple games. Since I had the Nats losing the division by one game that probably means I should have had the Nats winning the division anyway. Oh well. I'm not bothered.

But what does bother me is the whiff on the second wild card. I said the Nats would miss the playoffs because the second WC team would win 91 games. This was mostly because in 2015 the second wild card in the NL won 97 games. 97! Well looking back that was an extreme event. In reality there have been only three times in the past four years of two Wild Cards that a 90 win team would definitely not get in. Last year's NL and the AL in 2012 and 2013. That's as many times as a 90 win team would have been first WC (2014 both leagues, 2015 AL). That's not counting this year, where it looks likely that the AL WCs will be 89/88 game winners and the NL WCs will be 87/86 game winners. Even with the prediction of 90 wins, I really should have had the Nats in as a Wild Card, maybe even hosting it. That would have been more likely with 90 wins than missing it alltogether.

Whether you consider it a mild miss, like me, or a big one, can I make up for this miss with playoff predictions? I don't know. I don't love making playoff predictions because of the inherent variability of them. I've checked past years and I can only find one in a post since 2012 where I bothered. But I guess I'll try again because I nailed that one - picking the Royals in 5 over the Mets in the Series.**

*For the season as a whole. For the playoffs it looks like they will have bad luck with injuries.

 ** Don't do the same! The safest prognosticator route is to take the favorite in the maximum number of games possible. Why? Your pick has a built in one-game cushion, meaning if the favorite wins in 6 or the underdog in 7 you can point to your prediction and say "basically right!". If the favorite wins in fewer than one less game than predicted you can still say you picked the winner correctly. Also sweeps are hard. Think about how hard it is just to sweep a random series against a nobody team mid-season. Now you have two teams of around equal talent trying to win at all costs. Hard to have it go your way three or four times in a row without it ever going the other way. You can't worry about them. So really you are only likely to "lose" if the underdog wins in a shorter series. That's a good bet to make. 

However you should throw in a random underdog winning here and there just so you can point to it for the rest of your prognosticating lives, even if it wasn't all that crazy a prediction. No one remembers you picking against a heavy favorite that wins, but you'll never let anyone forget you picked against a heavy favorite that lost.