Projected 2016 Strengths of Schedule

I write some version of this post every spring, and each time, I’m more excited at first than I am as I get more involved. Whenever it comes back to me, I always like the idea, but then eventually I remember it just isn’t that important. It certainly isn’t something people are keeping in mind all season long — no one really worries about the standings until, I don’t know, July, and the league-wide landscape in baseball is pretty even, relative to other professional sports. Most fans operate under the assumption the schedules are more or less even, and they nearly are. Differences are subtle.

But, you know, differences are there, and remember that this is an MLB environment that considers a win on the free-agent market to be worth something like $8 million. Every single win is important, in some sense, and because the schedules aren’t truly identical, there’s no harm in examining the projected advantages and disadvantages. Acknowledging from the outset that this is all based on projections, and that you don’t agree with all the projections, let’s quickly go over the various schedule strengths.

FanGraphs makes this really easy, to be honest. If you navigate around, you’ll find our Projected Standings page, which shows what it says. But you’ll also find our Playoff Odds page, which shows projected standings that are ever so slightly different. This is not a bug! Many people frequently come to the conclusion that this is a bug, but the reason is that the Projected Standings page shows projected records in isolation, while the Playoff Odds page shows standings that fold in the various team schedules. Now, to be honest, I don’t totally know why things show up this way, but I’m not going to complain as long as they do, because of posts like this one. It becomes trivial to evaluate schedule strengths — you just look at the differences between projected team winning percentages. You could do it yourself in 15 seconds.

I’ll assume you haven’t already done it yourself. Which is fine, because I did. Let’s start with the American League schedules. In the plot below, you see the differences per 162 games, in terms of extra wins. A positive number reflects a softer schedule, while a negative number reflects a tougher schedule.

strength-of-schedule-al

Let’s go ahead and acknowledge the modest elephant in the room: community consensus is that our Royals projection is way off, and these calculations depend on the projections, and on nothing else. So that would be a factor in here, although in fairness it would be a pretty small one. If you want to adjust the Royals’ team projection, then they would get a little boost here, and the Indians would get nudged down. Every team would get slightly nudged down, on account of the games scheduled against the Royals, though of course the rivals in the AL Central would feel it the most. I don’t want to dwell on this too much, but I would feel silly if I didn’t bring it up.

And as for the Orioles, I think there are actually two things happening. One, I think the Orioles have a slightly tougher schedule. But, two — annoyingly — I think maybe the Projected Standings page reflected the Pedro Alvarez acquisition when I grabbed data, while the Playoff Odds page didn’t. I’m given to believe the Playoff Odds page refreshes itself less frequently, so if this is indeed the case, then the Orioles wouldn’t really be as poor off as they seem here. Alvarez makes them a little bit better. Whenever I’ve analyzed schedule strengths before, the Orioles always seem to come away with a tougher slate than their rivals, but I don’t think it’s a real difference of two wins.

Truth be told, the AL isn’t the interesting league, because the AL is characterized by its parity, and the landscape really is pretty well balanced. This projects to be maybe the most even year in recent AL history, and so differences between schedule strengths would be small. Even after adjusting for the above, it seems like the Indians might have a slight advantage. The Angels would have a slight disadvantage, and I think the Orioles would, too. It all matters when the wild card is likely to be so hotly contested. Anyhow, let’s move on to the National League:

strength-of-schedule-nl

A funny side effect of the way the NL is set up is the teams with the disadvantageous schedules are the teams that mostly aren’t even trying to win. Like, the Reds have the toughest schedule in the league by this measure, but I doubt they really care. The Rockies might care a bit, because it seems like they’re at least in part in denial, but it’s not like the Brewers care, and it’s not like the Phillies or the Braves care. They’re all going to lose. An extra loss or two is no big deal. Conveniently, there are even two bad teams per division, so that kind of balances out.

But it doesn’t balance out entirely, because the Phillies and the Braves seem like they’re going to be worse than the other four last-place contenders. So once again, you see an advantage for the Mets and the Nationals. They’re both essentially at +2 wins, and the Marlins also get to benefit. It probably won’t make them feel better about losing Carter Capps, but it’s not nothing, in a year when the Marlins think they might have enough talent to go someplace. The three biggest advantages are in the NL East. Then you find the Cubs and the Dodgers. Their benefit, as you understand, is that they don’t have to play against themselves.

One thing I can’t help but hone in on is the difference between the Marlins and the Diamondbacks, here. They both want to at least fight for the wild card, but the Marlins are given a relative schedule edge of almost two wins. That feels like it could be a big deal, and though there’s a difference between this and starting the Diamondbacks with an actual two-win handicap, this could end up a contributing factor. It would never be the first factor someone would talk about, but an advantage is an advantage.

Just like the two-win difference between the top teams in the NL East and the Cardinals, Giants, and Pirates. Only one of the Mets and Nationals can win the division, so if the other is also good, that team will be in the wild-card hunt, and, you see where this is going. It’s the same as the previous paragraph.

There’s nothing worth freaking out over. Projections can be wrong, and no less importantly, rosters change. As people like to say, it’s not so much about who you play as it is about when you play them. Maybe some team is going to be extra lucky or unlucky facing the Dodgers and Clayton Kershaw. Literally the entire season has yet to play out, and schedule strengths aren’t going to singlehandedly determine the playoff picture. But the schedules, you know, are never completely even. That can’t not matter.



Print This Post



Jeff made Lookout Landing a thing, but he does not still write there about the Mariners. He does write here, sometimes about the Mariners, but usually not.


Sort by:   newest | oldest | most voted
The Jestaplero
Member
The Jestaplero
2 months 19 days ago

I have a vacancy if anyone is looking for a league.

This is a league I started with my friends last year. It’s competitive with a good mix of beginners and experienced players.

It’s Yahoo, roto, 12-team, mixed league.
Draft is Sunday, March 20, at 8pm.
Buy in is $30.

R HR RBI SB AVG OPS
W ERA K SV WHIP K/9

Post your email if interested.

Beer
Member
2 months 19 days ago

Um, you need to post these things on Notgraphs.

The Jestaplero
Member
The Jestaplero
2 months 19 days ago

The same Notgraphs that shuttered in 2014?

Beer
Member
2 months 18 days ago

Hell yeah. Hopefully he’s still trying to get his comment to post.

Shirtless Bartolo Colon says in Serbian to Vietnamese to French and back to John Elway
Member
Shirtless Bartolo Colon says in Serbian to Vietnamese to French and back to John Elway
2 months 19 days ago

I too have 19 “fantasy baseball” vacancies.

Standard 5X5 mixed league, 20 teams. Auction draft 3/16. $150 buy in, $3000 payout to winner.

Please post account info if interested.

Remain blessed,
Prince Achebe Surugaba
Cotonou Benin Republic West Africa

bluejaysstatsgeek
Member
2 months 19 days ago

Do you need help getting millions of USD out of the country?

John Elway
Member
2 months 18 days ago

LMHAO!!!!!!!!!!!!!!!!!!!!!!!!!!

Steiner29
Member
Steiner29
2 months 18 days ago

I would be interested. Email is masterj2700@aim.com

Ernie Camacho
Member
Member
Ernie Camacho
2 months 19 days ago

What I’m guessing confuses a lot of people is that your “Projected Standings” aren’t really your projected standings. They are actually your estimates of team relative true talent, expressed in a w/l record scaled to a 162 game season. By definition, good teams have easier schedules than bad teams, so the playoff odds (the actual projected standings) will elongate the distribution of W% and could even tip the relative order of wildcard contenders.

Jack B
Member
Jack B
2 months 19 days ago

I’m still confused as to how the strength of schedule plays as big of a role as it looks like it is playing for certain teams. I don’t see why it should be different for teams that are (a) in the same division, and (b) evenly matched in projected WAR on the Depth Charts page.

For instance, in the depth charts, the Cardinals are at 41.3 WAR and the Pirates are 41.1 WAR. Virtually identical. The play in the same division, so their schedule should be nearly identical as well, and they’re evenly matched so playing each other shouldn’t cause an imbalance.

Yet, the Playoff Odds page has the Cardinals at 85.3 wins and the Pirates at 83.9. How is that possible based on a 0.2 WAR difference in projections, with same divisional schedule, and no real imbalance from the teams playing each other?

You see the same thing with the Orioles at 36.3 on the Depth Chart actually ahead of the Rays at 35.9, but Orioles projected at 79.1 wins to the Rays’ 80.0 on the Playoff Odds page. There may be some weird updating thing there as you pointed out in the article, but even before the Alvarez signing the Orioles were 0.6 WAR behind the Rays on the depth chart and 2 full wins behind them in Playoff Odds.

Is there something weird with how the strength of schedule is coded that I’m not seeing, or have these teams really gotten that unlucky in their draw for opponents relative to their divisional opponents?

phoenix2042
Member
Member
phoenix2042
2 months 19 days ago

Could it be whom they play in interleague? There are some Al/NL rivalries that get played every year, and they might happen to be easier or harder. Like how the Yankees are going to play the Mets a bunch, and the White Sox are going to play the Cubs a bunch.

phoenix2042
Member
Member
phoenix2042
2 months 19 days ago

Also I bet the Orioles will be playing the Nats a bunch for the same reason. That could account for it too.

Jack B
Member
Jack B
2 months 19 days ago

This could be part of it, but is that handful of interleague games really causing this large of a difference? I suppose it could but you’re talking about an extra 3-game interleague series. The interleague schedules are the same for all teams in the division except for an extra series vs. the “rival.” I don’t see how a 3-game series could cause a full 1-win difference.

Ernie Camacho
Member
Member
Ernie Camacho
2 months 19 days ago

See my above comment. The team WAR totals aren’t estimates of team quality. If you look at the “Projected Standings” you’ll see that Fangraphs projects the Cardinals’ schedule-neutral team talent as .525 W% and the Pirates at .517, a one win difference.

Jack B
Member
Jack B
2 months 19 days ago

Are the Depth Charts not schedule-neutral? It’s just an accumulation of projected WAR. Why would that not be schedule-neutral?

My understanding is:
Depth Chart: Schedule Neutral
Projected Standings: Schedule Neutral
Playoff Odds: Includes SOS

So maybe a better question is, why is there a discrepancy between the the Depth Charts and the Projected Standings?

Ernie Camacho
Member
Member
Ernie Camacho
2 months 19 days ago

Fangraphs does not use aggregate projected WAR as its team talent estimator. Fangraphs literally doesn’t use any WAR estimates in their team projections.* The team projections are based on the projections of underlying components, which FanGraphs converts to runs (using a BaseRuns formula), and runs and runs against are converted to W/L using PythagenPat.

*[Incidentally, that’s why the complaint about WAR supposedly unerestimating relief pitchers or whatever is totally irrelevent to the team projections.]

Jack B
Member
Jack B
2 months 19 days ago

That’s the answer I’m looking for, thanks. I didn’t realize that’s how it worked. So correct me if I’m wrong, but the difference is in the bullpens – with WAR, higher leverage relievers are rewarded, but that’s not a component of BaseRuns – all innings are valued the same. Therefore teams with good pens in the Depth Charts such as the Orioles, Yankees and Pirates are not seeing that reflected in their Projected Standings via BaseRuns?

Ernie Camacho
Member
Member
Ernie Camacho
2 months 19 days ago

Fangraphs could definitely do a better job explaining how all these different projections and projected standings link together. Took me a while to unpack it all.

I think you are exactly right about bullpens. Which is kind of ironic, because the overall critique about properly weighing bullpen quality is probably right to some extent, even though it has nothing to do with WAR.

That said, based on my own playing around with historical team performances, my sense is that people over-estimate how much “good bullpens” can “beat” BaseRuns or run differentials. For one thing, every other team is also using its best pitcher(s) in high leverage situations (more or less), so the effect is somewhat symmetrical. For another thing, its a crapshoot how even good relief pitchers distribute their own performances. Even elite closers give up some runs, and it matters a ton whether it’s in games with 3 run leads vs 1 run leads.

evo34
Member
evo34
2 months 18 days ago

“If you look at the “Projected Standings” you’ll see that Fangraphs projects the Cardinals’ schedule-neutral team talent as .525 W% and the Pirates at .517, a one win difference.”

I’d be careful with using Projected Standings here to compare true team talent. If a team has a lot of players who have faced an easy schedule last three seasons, it will be over-valued by the current system (Steamer does not adjust for past schedule — only league and park).

soddingjunkmail
Member
soddingjunkmail
2 months 19 days ago

Looks like the Indians, White Sox, Twins, & Tigers are all benefiting from getting to beat up on those patsies the Royals.

phoenix2042
Member
Member
phoenix2042
2 months 19 days ago

How do the Red Sox get such a big bump when they have a division with no bad teams? The Yankees, Rays, Blue Jays, and Orioles (in no particular order) are all at least decent and have a chance at contention. The Red Sox, I would think, would therefore have a really tough division. Why do they then have the second easiest schedule in the AL?

Ernie Camacho
Member
Member
Ernie Camacho
2 months 19 days ago

Because the other teams in their division have to play the Red Sox a bunch of times, and the Red Sox don’t.

Ernie Camacho
Member
Member
Ernie Camacho
2 months 19 days ago

Blah, sorry, that wasn’t exactly responsive. The bump (or penalty) for each team comes from the size of the projected talent gap between it and its opponents, and since the Red Sox are projected to be very good (the best in the AL), and the rest of their division about .500-ish, that gap is fairly large.

suicide squeeze
Member
suicide squeeze
2 months 19 days ago

Also, four games against the Braves that no one else in the division has

Ernie Camacho
Member
Member
Ernie Camacho
2 months 18 days ago

Wait a sec, is the math Jeff did in this post really a measure of schedule strength? Isn’t it really capturing the delta between a team’s true talent and its opponents’ true talent? I’m not sure it really matters much–this measurement is arguably more relevant–but it isn’t really what most people have in mind when they think “strength of schedule.”

evo34
Member
evo34
2 months 18 days ago

This is a really complex topic that Sullivan brings up. Several points to make:

1) Steamer projections are NOT based on anticipated strength of schedule — not directly anyway. What they do is assume that the player has played a league-average schedule for the last three seasons and that he will do so again this season. For players on the same team for a while, there will be some unintentional schedule adjustment baked in. Players who played a rough sked last three years will be projected to do worse than similar players who played an easier sked.

2) Therefore, there is a ton of inconsistency on Fangraphs as you move from player projections to team win forecasts. Here is a list of what projections are at least partially schedule-adjusted (none of them are adjusted correctly, btw):

— Steamer player projections: SLIGHTLY [unintentional player L3Y schedule effect]
— Depth Charts team WAR projections: SLIGHTLY
— Projected Standings: SLIGHTLY
— Playoff Odds: FULLY/TOO MUCH?

The reason the playoff odds are over-adjusted is that the player projections were *not* neutralized for past (3 years) schedule. That is, if a player played an extremely difficult schedule bc his team was in a tough division, Steamer does not know this and will project his numbers as if he was playing the same tough schedule this season. I.e., some unintentional schedule adjustment at the player level.

Bottom line: to correctly project team wins, one must either: a) start with fully schedule-neutral player projections and then apply SoS adjustment to teams, or b) apply SoS adjustment directly to the player projections (Steamer would have to use expected schedule strength) and then build team forecasts without further adjustment.

As it is, it’s a hodge-podge of improper adjusting, which would be fine if the issue was fully documented somewhere on the site. But unfortunately there is zero documentation — other than some reference to using the same algo coolstandings.com used. And yes, getting schedule effects right does make a big difference in win forecasts.

David Appelman
Admin
Member
2 months 18 days ago

Fair on the documentation point. Maybe we can put that somewhere in the glossary somewhere.

The Depth Chart projections are supposed to be SoS neutral, and the playoff odds are adjusted for SoS since they are a season sim and not just straight BaseRuns.

Do you know of any projections systems that SoS adjust prior seasons? I’m somewhat skeptical that this is significant issue, but I can be convinced.

evo34
Member
evo34
2 months 18 days ago

No, I do not know of any projections that explicitly factor in past or future SoS off-hand (but then again I am not intimately familiar with each algorithm).

So there are really two issues:

1) Projection models are not adjusting historical stats for past schedule [which, ideally, would take into account not just team opponent strength, but individual pitchers/batters and their handedness].

2) They are not factoring the effect of a player’s upcoming schedule [regressed for uncertainty however much is appropriate].

As a result, there are two sources of error at work, and they are not insignificant in aggregate.

As an extreme example of the first issue, Derek Holland’s average opposing batter’s OPS last season was .761, whereas Shane Greene’s was .710. That’s worth about 0.65 runs of expected difference in ERA, which to my knowledge is not being accounted for in any of the mainstream projection systems.

evo34
Member
evo34
2 months 18 days ago

Same thing with average road park factors for pitchers on the same team. For example, last year, Vogelsong had an average park factor of 94 (all parks pitched, incl. home games), whereas Lincecum’s was just 88.

wpDiscuz