The Rockies are Hot. Is it Time to Re-evaluate?
Fifteen-ish games later, there are indeed three NL West teams in playoff position. The Dodgers are there, of course, and the Padres — no surprises here. But then there are the Colorado Rockies, 11-4 and leading the National League. It’s early — although with a quarter of the season already in the books for many teams, how early is up for debate. But regardless of the time of year, the Rockies are in first place, and I wanted to learn more.
Going into the season, we pegged the Rockies as a .461 team after accounting for their extremely difficult schedule, somewhere between 27 and 28 wins. That gave them a 28.4% chance of making the playoffs, per our odds. That’s a projected win probability for the full season, though, not the first 15 games. And the Rockies had something going for them in those first 15 games; one of the easiest projected slates in baseball:
Team | SOS |
---|---|
MIA | 0.472 |
STL | 0.472 |
MIN | 0.474 |
COL | 0.475 |
OAK | 0.478 |
CHC | 0.479 |
DET | 0.481 |
WAS | 0.486 |
CIN | 0.490 |
TBR | 0.493 |
LAA | 0.494 |
HOU | 0.496 |
TEX | 0.497 |
LAD | 0.497 |
BOS | 0.499 |
MIL | 0.499 |
KCR | 0.502 |
NYM | 0.503 |
NYY | 0.504 |
SDP | 0.506 |
CLE | 0.507 |
BAL | 0.508 |
PIT | 0.508 |
CHW | 0.510 |
ATL | 0.518 |
PHI | 0.519 |
SEA | 0.521 |
TOR | 0.528 |
SFG | 0.540 |
ARI | 0.548 |
Short of the two teams with suspended seasons and the Twins, no one has had it easier than the Rockies (using our rest-of-season opponent-neutral winning percentages to determine schedule strength). That sounds surprising, because Colorado is in a division with the Dodgers and plays interleague games against the Astros. Even the second-tier teams in the two Western divisions, the A’s and Padres, are formidable opponents.
But the Rockies have so far avoided the two powerhouses, while playing the A’s only twice and the Padres three times. That leaves their remaining strength of schedule as the second-toughest in baseball, behind only the Marlins, who suffer from the double curse of playing in the NL East and not getting to play those perpetual NL East doormats, the Marlins.
Merely reducing the Rockies’ start to “Oh they haven’t played anybody” is unfair, though. If the Rockies were a true-talent .500 team (and we think they’re close, at .489 the rest of the way), they’d go 8-7 on average in the games they’ve played. They’ve outdone that by three games, which is huge; if they’d undershot by three games, they’d be sitting at 5-10, and we’d be calling this season a wrap for them.
Some of that is the vagaries of a short season. Starting hot is important when starting is a quarter of the battle. But I wanted to tackle a slightly different question. Let’s say, first, that you share FanGraphs’ view of the Rockies; a sub-.500 team in expectation, but pretty close to that dividing line. How likely would they be to pull off this hot start? Next, what if you disagreed with us and thought they were better? How likely would this be then?
To do so, I built a lazy approximation of the season so far. First, I gave the Rockies a static winning percentage — in our first example, I used that .489 number from our projected standings, but I’ll vary it later on, so don’t get too caught up in that. Next, I built their schedule as it happened — three in Texas, two in Oakland, three against the Padres, four against the Giants, and three in Seattle.
At this point, I needed to come up with an approximation of home field advantage, which has so far been absent from my calculations. That’s a tricky nut to crack with no fans and with the Jays performing a traveling road show this year, but I went very simple and took the total league-wide home winning percentage, which came out to 52.5%. Good enough for me — we’ll use a 2.5% home field adjustment.
From there, I went to an old standby in my articles: simulating a snippet of baseball an arbitrarily large number of times. In this case, I simulated the Rockies’ first 15 games a million times. If the Rockies are what we think they are, this is kind of a boring simulation: they’re 7.1% likely to go on this run (or do better), which doesn’t sound too high or too low. If there are 10 teams of roughly Rockies-level talent and schedule, it’s roughly a coin flip that one of them would start this hot.
Again, though: that’s boring. Let’s start messing with the Rockies, making them better or worse. First, move them up to .500. Now, they’re 8.3% likely to start this hot or hotter, hardly a difference. But let’s keep tinkering; how about a .520 Rockies team? .540? .600?
True Talent | Odds of 11-4+ |
---|---|
.489 | 7.1% |
.500 | 8.3% |
.520 | 10.9% |
.540 | 14.1% |
.560 | 17.8% |
.580 | 22.1% |
.600 | 27.0% |
While we’re at it, let’s put in the bottom of the scale, too:
True Talent | Odds of 11-4+ |
---|---|
.400 | 1.50% |
.420 | 2.20% |
.440 | 3.20% |
.460 | 4.50% |
.480 | 6.10% |
.489 | 7.10% |
Why do all this? Mostly for show. It’s neat to see all those numbers stacked up in tables, and to know that even dominant teams don’t go 11-4 very often, while even awful teams can do it sometimes. But I promise, there’s a punchline after all of this.
If you’ll recall, Rockies owner Paul Monfort declared the Rockies a 94-win team before the season. To be blunt, that looks wrong. That’s a .580 winning percentage, and the team simply doesn’t have the horses. Now that they’ve started 11-4, how much should we re-evaluate Monfort’s claim?
There’s an easy tool we can use for this, though it requires a few assumptions. Given that this article is already pretty much assumptions all the way down, though, that’s fine with me. We can make a Bayesian inference to see how much more believable Monfort’s claim is in the wake of a hot start.
First, let’s set some priors. Say before the season that there was a 95% chance that our projections were right; the Rockies were a sub-.500, but somewhat competitive team. There was also a 5% chance that Monfort was right, that he was sitting on a .580 true-talent juggernaut that every model simultaneously missed. That’s giving him too much credit, I think, but we’re optimists here at FanGraphs.
Next, it’s time for a little math. If the Rockies were 95% mediocre and 5% great, they’d start 11-4 or better against this slate roughly 8% of the time. How did I get that? 95% times 7.1% (the odds of 11-4 at their current .489 projection) plus 5% times 22.1% (the odds of doing it if they were a .580 team).
Now that that 8% likely outcome has happened, it’s time to re-evaluate the odds we give the team of being mediocre or great. 86% of the probability mass of going 11-4 or better comes from the .489 team, which means we should adjust our guess; we now think they’re 86% likely to be a .489 team and 14% likely to be that Monfortian juggernaut.
That’s with an optimistic 5% probability of the Rockies being a 94-win team. Shift that probability down, and Bayes is harsh. Start with a 99% likelihood of mediocrity and 1% of greatness, and we’d now still be 97% certain that the Rockies are mediocre; a two percentage point drop, but still overwhelmingly likely.
What does all of this mean? Not much, to be honest. What I think of the Rockies doesn’t affect how they’ll play. But if you’re wondering how much you should adjust your view of them, it depends on what you thought in the first place. If you already thought there was a decent chance our predictions undersold them, you should be far more bought-in now — heck, if you thought it was 50/50 that they were mediocre or great, you should think there’s a 76% chance they’re great now. If you thought our projections were extremely likely to be correct, well, this is probably just a hot streak. That’s a fun little mathematical trick, and what’s the baseball regular season good for if not fun little mathematical tricks?