We're winning, which is good, and the wins that we've accumulated thus far are in the bank. Can't nobody take 'em away from us. Let's not worry, for the moment, whether we'll continue to win. Let's figure out how the heck we've been this good so far.
Win Probability
We'll start by checking out our win probability added. If you're not familiar with WPA, you can read about here. The basic idea is that every event in the game either decreases or increases your chance of winning. If you record, for each event, how much it increases your chance of winning, and then add up the values for each event, you will end up with the number of wins you currently have. The awesome site Fangraphs records all this information for us. Since the probability of winning any given game starts at 50%, one half of a WPA "point" counts for one win.
WPA is really neat because it captures clutch. When the Angels scored a run off of Dallas Braden in the ninth inning of Wednesday night's blowout, it didn't increase their chance of winning by very much. On the flip side, Jack Cust's tiebreaking single against Brandon Morrow last Sunday was worth a ton because it broke a tie game in the late innings.
So, let's see how the A's have been winning, per WPA (as of Friday morning):
Hitters: -0.38
Starters: +0.98
Relievers: +2.39
Before we start, we do a quick sanity check by adding these numbers. -0.38 + 0.98 + 2.39 = 3. One half of a WPA "point" counts for one win, so the difference is 3*2 = 6. And how many games over .500 are we (as of Friday morning)? Six. Perfect.
Despite the fact that we've scored a ton of runs, the hitters have been slightly below average, registering a net negative WPA. I'm guessing this is due to the fact that A's hitters have actually been not so hot in clutch situations. Yes, it seems like every time we have hitters on that we're driving them in. That's good, but driving in runners isn't the only aspect of clutch hitting - getting on base helps, too. In some of the close games we've lost, we just haven't gotten on base at important times. On the flip side, we may be driving in lots of runners, but how does that really help when you're blowing out the other team?
The starters and relievers have been quite good, with the relievers in particular being totally awesome. Casilla, Brown, Devine, Street, even Foulke have been awesome. Street's occasionally shaky performances haven't really cost us except for the Japan opener. And of course, Casilla, Brown, and Devine have been...well, divine.
Batting Runs Above Average
Remember how WPA records the change in your chance of winning for every event? Batting runs above average (BRAA, also recorded at Fangraphs) does the same thing, but it records the change in your chance of scoring a run for every event. Think of BRAA as WPA, but without accounting for how clutch the situation was in terms of the game situation. So, hitting an RBI single to bring home a man on third with one out is counted the same regardless of inning or score. You can think of BRAA as still accounting for the clutchness of the base situation, but ignoring the inning and score.
And the A's?
Hitters: +5.3
Starters: +19.0
Relievers: +17.4
Let's add 'em up: 5.3 + 19.0 + 17.4 = 41.7. And we've outscored our opponents by...40 runs. Almost spot on.
As we suspected, the hitters do quite well when we ignore the context of inning and score: their contribution by WPA is not as much as their contribution by BRAA. They may not be hitting really well late in close games, but they're hitting well when there are runs to be driven in.
The pitchers account for 36.4 BRAA and 3.37 WPA, meaning their performance in clutch situations has been really good (36.4 BRAA ~ 3.5 wins, 3.37 WPA ~ 6.8 wins). You'll notice, however, that the starters and relievers look much more equal by this measure, for the very basic reason that relievers are used toward the end of games when a lot of WPA points are doled out.
Context-neutral runs
What if we ignore all aspects of clutch, so that we ignore not only the inning and score, but also whether or not there are men on base. You can do this a number of ways, and I won't get into the gory details. I'm going to use a Tangotiger method - those are my favorite since they're always open source. The basic formula is to take all the team's hits, doubles, triples, homers, etc., ignore when they happened and consider only how many there are.
To the numbers!
Hitters: -14.5
Starters: +28.0
Relievers: +19.1
Ugh. The hitters look pretty bad here. We know in our heart of heats that our offense is below average. By ignoring the context in which we hit, we can see it pretty clearly. Our hitters have been -14.5 runs worse than average in a context-neutral setting, and yet they've produced 5.3 runs in a semi-contextual sense (BRAA). That means that we've been hitting out of our minds with men on base: .297/.384/.411 (AVG/OBP/SLG). Is it luck? Skill? Some kind of Ty Van Burkleo pixie dust? Without doing a super in-depth analysis, it's hard to say for sure. Yes, guys definitely change their approach when there are men on base, so we can't dismiss the BRAA number out of hand. But there's also a heavy does of randomness involved. If the team continues to hit the way they've been hitting, we may see an offense that struggles and sputters to score runs.
Our starters have actually been kind of crappy with guys on base, as their context-neutral runs are much better than their semi-contextual runs (BRAA). I'm not sure how that compares to league average, though, since I'm pretty sure that pitchers in general do worse with men on base. The bullpen has been basically even.
If you add up the context-neutral numbers, you get -14.5 + 28.0 + 19.1 = 32.6 runs. In reality we've outscored out opponents by 40 runs, using BRAA our run differential "should" be ~42 runs, and using a context-neutral measure it "should" be 32.5 runs.
The A's have gotten off to a great, great start. We're scoring runs with very timely hitting, particularly with men on base. We're preventing runs by just plain old good pitching. Analyzing our hot start shows that it's not a fluke - at worst, we might conclude that we've only been good enough to outscore our opponents by 32.5 runs. But even if that's the conclusion, that run differential suggests a team that's six wins above .500...which is right where we find ourselves today (Friday morning ).
Now, that doesn't mean that we're going to continue being this good or this bad. We might get better and we might get worse (probably worse is my intuition). But that wasn't the point of this analysis. The point of this analysis is to show you that our hot start was NOT driven by luck. It may (or may not) have been driven by unexpectedly good performances. But that's not what we're looking at, either. Given the performances we've received from our players, this team has not been lucky.
The A's have been good.