The Brotherhood: We Just Haven't Debated Sarkisian Around Here Enough

Very sad to hear the news of Don James's death. He was a great man and a great coach.

Darin changes the subject

In an effort to talk Husky fans in from the ledge, I'm going to describe as best I can something that I think is true of college football but that most of us don't really consider.

Going into the Arizona State game, I thought the Huskies were a touchdown better than Arizona State. I suppose what that means is, if we play our average game and they play their average game we win by a touchdown, but the precise meaning isn't important for this discussion. Now, everybody recognizes that sometimes teams play better or worse than their average game, so there's some chance Arizona State would win even if the Huskies are "actually better." This is called an upset.

One way to picture this state of reality as of Saturday morning is that the Football Gods will roll their dice, once for the Huskies and once for the Sun Devils. Whichever team gets the higher roll wins the game; however, since the Huskies are a touchdown better, their dice are weighted to give a higher number. The technical language is something like, the probability distribution of outcomes for the Huskies has a higher mean than the one for the Sun Devils.

The graph below shows how we tend to picture this state of affairs. The outcome for each team is a bell-shaped curve, meaning there's some chance of a good game or a bad game, but the most likely thing is a game around average. The Huskies' curve is shifted out to the right, by seven points in this made-up example. There's some overlap between the curves, reflecting the possibility that on a given Saturday the Football Gods will roll a higher number for ASU, but most of the Huskies' curve is to the right of the ASU curve.

You can do a little math to figure out that, again in this made-up example, the chance of the Sun Devil roll being higher than the Husky roll is 16 percent. I.e., the Sun Devils have a 16 percent chance of winning the game. There are a few other things you might be interested in, too.

• If the Sun Devils win, their expected margin is about one point. (This isn't necessarily points on the field, it just means that ASU will be only a little better.)

• The probability of a 14-point differential in favor of ASU, which means something like the blowout we saw last weekend, is 0.5 percent. Basically impossible.

That isn't right, though. Although the Huskies actually are seven points better than the Sun Devils, the curves don't look like that. They are much "flatter," meaning the chance of a very good game or a very bad game is higher than we like to admit. Maybe something like this.

In this version of reality -- the real reality -- the Huskies are still seven points better on average, but the overlap between the curves is much more substantial. If you do the same math as above you find the following things.

• The chance of a Sun Devil win is 38 percent (more than double).

• The expected margin if the Sun Devils do win jumps to five points.

• The probability of a 14-point blowout by the Sun Devils increases to 15 percent -- 30 times more likely!

What are some takeaways for Husky fans?

Takeaway No. 1 -- Relax, we haven't really learned anything...

It seems clear that this game, even though it was a blowout, doesn't tell us anything existential about our program or our coach. It could have been bad luck, it's almost impossible to tell.

Takeaway No. 2 -- ...unless we have.

Arnold Palmer used to say that the difference between his golf game and yours wasn't that his best shot was so much better than your best. It's that his worst shot was a lot better than your worst. As Husky fans, we had hoped games like that one were behind us. I guess not.

The point here isn't that this is exactly reflective of reality. I'm just trying to highlight the fact that the variability in the performance of a college football team and the outcome of a game are probably a lot higher than we'd like to admit. There are all sorts of discussions that come from this. For example, does WSU have a wider curve than Alabama? Can you cause your team to have an asymmetrical curve, with a big upside but not the equivalent downside? How much control can a coach have over what the roll will be in a given week? Any?

I sure hope we beat Cal.

Brad Mostly Ignores the Notion

Okay, fine. I get the takeaway. And if everything goes the Dawgs’ way the rest of the season, this one game will be an aberration.

Arguing this way is totally ignores the human aspects of a football season. Dice are dispassionate; one roll has no effect on the next. That’s not the case here. The Gods have a tremendous impact on what roll they’ll produce, and the cumulative effect of a series of bad (or good) rolls can have a tremendous influence on the next one.

There are too few data points to simply boil college football down to statistics for a single season. Going into Saturday’s game, the UW was “better” than ASU on both sides of the ball (FEI and S&P+). Now, ASU’s better. Maybe it’s just because ASU had a really good roll on Saturday, and the Dawgs had a bad one. That’s not the most logical conclusion, though, because the odds are so highly stacked against it (especially given the way the game played out – an absolute physical ass kicking). No, there’s likely something else at play that kept the Huskies’ heavily-weighted die from being able to come up with anything bigger than a one.

Maybe the problem actually is that the Huskies aren’t just playing dice. That they can’t just go into a game, and have any reasonable expectation of achieving the most likely outcome.

We did learn something on Saturday. We learned that in Year 5, Sarkisian now has to win out to avoid a run in-conference that looks like this: 4-5, 5-4, 5-4, 5-4, and 5-4. I’ve always held that conference play is the really measure of success for a team. And while I definitely think this team is better than it was last year, and probably the year before that, the fact is that relative to the teams it’ll play year in and year out, there’s nothing you can point to that suggests progress. The conference as a whole has gotten better in that time. From the top almost all the way down to the very bottom. But……so?

If I asked you to make a list of all of the things that Sarkisian has improved upon as a coach from 2009 to today, what would be on it?

Darin Sticks With the Science

I agree with your basic point, that there's a "human" element that my analogy ignores. I just think that element is greatly exaggerated and that simple randomness explains more than we'd like to admit. We mortals, and especially football fans, are expert at telling stories to explain what happened. Unfortunately, we're a lot better at predicting the past than the future.

Okay, you say that there are too few data points to make big statements about "goodness" in college football. Then you say that the ASU game should make us believe something big about the Huskies -- that our die is badly weighted. I agree with the first statement, not the second one. Of course the ASU game doesn't mean nothing. It told us that the Huskies are capable of playing a worse game than I thought they could, for example. But the most we should do is make an incremental adjustment to our assessment of where things are. You have to look at everything.

I completely agree that Sarkisian's conference record looks pretty dismal. It's not what we wanted or expected. But here we are. There are two important questions for Husky fans to ask themselves before they start calling for Sarkisian's head or jumping off the Aurora Bridge.

First, what do we actually believe about the trajectory of the team? If you believe my story about randomness, it's very hard to tell the difference between a team that goes 7-6, 7-6, 7-6 and one that really "should" have been 6-7, 7-6, 8-5. If the last three years had followed that trend, we'd all feel much differently, even though we probably shouldn't. The fact that we have such small sample sizes is why we can't just look at wins and losses, especially in the beginning. We have to look, for example, at whether the coach is doing things right, whether the team is improving, and most importantly what our best guess of future wins is. The fact that we laid an egg at Arizona State should have some effect on our assessment of those things, but it's not the only datum.

Second, what's our next-best option? Although I believe randomness is important when it comes to the games, I believe it plays an even larger role in hiring coaches. As near as I can tell, there are a few coaches who seem to transport success wherever they go: Nick Saban, Les Myles, and... that's about it. Everybody else, I mean everybody, is rolling dice when they hire a coach. Pete Carrol was arguably the single most successful college coach in the last fifty years. That's why USC was in a bidding war with ten other schools for his services, and he was able to name his price -- because everybody knew he'd succeed in a big way. Oh, wait. That didn't happen. He was USC's fourth choice. He was Cal State Northridge after they didn't get into Stanford, Berkeley, or UCLA. Even the revered (late) Don James was not exactly number-one-with-a-bullet when he was hired at the UW. And here's the more relevant part for Husky fans today: it would have been impossible in the first part of Don James's career to say, "This guy is the greatest coach ever at Washington. We should make sure to keep him forever." We're perfect in our assessments, but only in retrospect.

So, let's say you look at the world with a cold eye and conclude that Sarkisian is probably not going to be Don James. Now what? Do you have a better plan? I made the point a couple weeks ago that the cost of changing coaches, including the risk of hiring somebody who doesn't work at all, is probably higher than the cost of continuing forward. I would keep Sarkisian without much hesitation.

What would be on my list of things Sarkisian has improved? Fair question. Recruiting. That's first on my list of what matters anyway, and pretty much nothing is in second place. It's not been perfect, of course, as we've discussed at length.

I'm not saying all is well. It clearly isn't. Here's what you should remember:

1. It's really hard to tell how well things are actually going.

2. If you make major changes in an effort to improve things you're at least as likely to make them worse.

Brad Wraps it Up

I mostly agree about the randomness. I guess what I’m getting at is that the human influence on the randomness is such that there’s a much greater chance of ending up with what we saw on Saturday than your mostly made-up stats. The probability of what we saw on Saturday is significantly higher than 15%. The distribution is way too wide, and the asymmetry is too weighted toward the “bad” end.

When I talk about the datum, I’m talking in a single year. But the ASU game is representative of a trend that’s gone on for years. This non-competitive type of loss that shows up when you least seem to expect it. This wasn’t a “trap” game (nor was Arizona last year). This was a game that in a lot of respects was actually bigger than Oregon or Stanford, because it was against a team that the Huskies should’ve and needed to beat to show the ability to separate themselves from the middle of the conference.

What’s the trajectory? Right now, today, it looks pretty flat. I get that there are going to probably be 6 more data points that are going to be needed before you can accurately plot the trend line for this season. And, should the team win out, things actually will look pretty good (no intellectually honest Husky fan would’ve turned their nose up at a 9-3 regular season last August). Over the first 4.5 years, though, it’s definitely flattening out after a pretty good start. Is Sarkisian doing things right? It’s a mixed bag, on all fronts. The scales aren’t tipped in either direction, really. The team is better than it was last year, and better than it was in 2011. Is it enough better is the question. The problem with predicting future wins with Sarkisian is that you’re far more likely to be wrong than right in the “swing games” based on what we’ve seen to date. Few people would’ve picked the Dawgs to beat Stanford or Oregon State last year. And few would’ve picked them to lose to WSU. Few picked them to lose to ASU. And predicting that sort of thing is very, very difficult to do dispassionately.

Of course I can’t answer the question of the “next best option” right now. Names get thrown around, but that’s mostly a futile exercise for fans. I totally agree that identifying and being able to hire the “right” coach is largely luck. So the question is either, How worried are you that you’ll end up doing worse, or How confident are you that you’ll at least match Sarkisian’s results? If the cost of keeping Sarkisian is lower than the cost of firing him, at what point does the pendulum swing?

Ah yes, recruiting. To me, it’s probably the best of Sarkisian, and the worst of him. The thing that’s going to preclude his success – no matter what – is the line of scrimmage. Let’s not count 2009. Sarkisian did great on the lines on both sides of the ball in 2010. Since? It’s been mostly misses. It’s setting up to be another one again in 2014. And that great line class of 2010 occupied all five positions on offensive line Saturday. That’s one of the most depressing facts about that game. The guys that were supposed to be the backbone of Sarkisian’s tenure got punched in the nose over and over and over again, and couldn’t do anything to stop the bleeding. What happens to the defense if Danny Shelton gets hurt? What if he goes pro after this season (not likely, I know…)? The fact of the matter is that the replacements and the reinforcements aren’t on the roster right now. I’ve said before that not being able to fix the offensive line in particular is a fireable offense. I have no doubt that Sarkisian is trying as hard as he can to fix the line. I also have no doubt that he’s failed categorically. Cozzetto may very well be fired after this season. After 5 seasons, is that enough? If it is, how long can you afford to wait and see?

I’m not reacting to this game as emotionally as it might sound. If the team wins out, I’m more than happy to have Sarkisian as the coach in 2014. Even though there were a lot of things to dislike about the team’s performance in 2012, there was enough background noise (mostly the injuries on the line) to, not ignore, but overlook the warts that seemed to plague Sarkisian and the team (mostly the same ones that show up every year). And things got off to a good start this year. But as things have gotten tougher, and the grind of the season has set in, my biggest issue is that the tunnel vision seems to set in every year with Sarkisian has set in again. The thing that most hit that point home was hearing him on KJR on Monday – he talked about how the team can’t make a living just throwing bubble screens, and that they had to get the ball down the field. I agree. You’ve got to throw the ball down the field. But it’s the way he’s doing it – like USC. And like he has USC’s offensive line. The model should be Oregon with Darron Thomas at QB, or the Boise State game plan. Instead, it’s the attack that puts the premium on the weakest link of the team. The one that lead to 10 3 and outs Saturday. The one that’s gotten his QB pummeled. It’s what he knows, and it’s what he’s going to do. It’s not going to work.

Steve Sarkisian has fatal flaws as a coach. What are the odds he corrects them?