One key to success for the Oilers next year is a decent year from Milan Lucic.
What does ‘decent year’ even mean though? What can we reasonably expect out of Looch next year?
Rebound or continued plummet?
Some are convinced that Looch’s points total next year will be in the 30s at best. Because he’s done – too slow, too old, too beat up for the new NHL.
Some are adamant that it will be in the 40s (meh) or 50s (good) or 60s (whaaat!) next year. They are very much in the “rebound” camp, arguing that Lucic’s scoring shortfalls are something that can be cured by the passage of time or an attitude adjustment.
I’m not very optimistic in terms of the number projection. It’s easy to see Looch doing better next year, but hard to see Lucic doing a lot better next year, and time is clearly ticking.
But for a variety of reasons I do happen to be in the rebound camp, and wrote about it previously.
That seems contradictory, no? Not so optimistic on the numbers, but still believing in a rebound?
This is because I believe there are two opposing forces at work regarding Lucic.
The (Simplest) Case for the Rebound
On the one hand, the single biggest reason for believing in a rebound is that his scoring fell off a cliff starting in January. Not at the start of the season! January.
While he finished 2018 with an absolutely miserable 8 points in 43 games, it’s worth remembering that he started the season with 26 points in 39 games. That’s on pace for a 55 point season. That’s a ‘prime Lucic’ pace, on par with his LAK season and better than his average in BOS.
That tells me there’s something more than just an aging curve or an off-season training issue or “it’s a faster league” situation at work. Those things didn’t suddenly happen over Christmas.
There are other mean-reverting factors (his sh% of 6.8 was the lowest of his career) at play, but the seasonal split is the simplest data point that can’t be easily countered with an argument based on aging or speed of NHL.
The Case for the Decline
The offset to that is that I also legitimately accept that there are a lot of hard miles on Looch, and so it’s reasonable – based on the arc of his career – to believe he is well onto the downslope of an aging curve already. Arguably, that’s an effect that was starting to show before he got to Edmonton, and the Oilers’ tendency to play a skating style rather than a dominating possession style (like BOS and LAK) perhaps has exposed that.
So my working hypothesis is that Lucic will indeed rebound, but he’ll rebound towards a ceiling that is declining.
Today’s article is to try and estimate where that ceiling might be.
The Simplest Aging Curve
The simplest aging curve assumes that a player follows a literal curve – a steady improvement when young, a broad peak, and a steady decline when older. The most basic statistic to use for this curve is something like points scored (normalized for games played of course).
So let’s model this with a curve (technically: a quadratic regression) on points per game. It looks like this:
So the first observation is that this represents a pretty decent fit. The curve runs nicely through Looch’s points history, and the usual metric for assessing fit (R², shown on the chart) indicates that about 50% of Looch’s points history history can be explained by aging. That’s quite a high number for any sort of curve fit that involves one person!
Notice that the chart shows the curve extended from past data to one year in the future. We can use that to create an estimate for Looch’s points per game pace next year.
And those results … are not so encouraging … to say the least.
Plugging next season (Looch’s twelfth) into the equation shown on the chart, we get an estimate for next year of 0.33 pts/game, or just 27 points in a full 82 game season. YIKES.
If we adjust that to an estimated range by looking at a 20% range of variation around that number, you get 0.27 to 0.4 points per game, which equates to between 22 and 33 points. Given Looch’s history of bouncing above and below that curve, maybe I’d say he’s more likely to outperform than underperform that number – this is where we see the bounceback in other words.
But 33 points. That’s basically a repeat of his miserable 2017 season.
And that’s *after* accounting for a bounceback.
Before we get too down in the dumps, let’s dig a little around that number to see if we can find something more encouraging.
Lucic had a singular points peak quite early in his career. Could that be a reflection of the strength of that Cup winning team … which then makes any sort of curve fit artificially severe on the downside?
To address that, let’s see what happens if we only model the post peak years. What does the decline look like then?
Results: well … better, though still not great.
The fit is still decent. But now the prediction sits at 32 points (0.39 points/game) with a range of 27 to 37. Still highly problematic if you’re counting on Milan to be a Top 6 forward.
Using the early season run rate
One more thought exercise – let’s dig into that early season/late season split. The one where he was at a solid 55 point pace up to December, then scored at approximately the same rate as his own cardboard cutout after that.
Let’s take Milan at this word, that he fell off a cliff in January because he was down in the dumps. And now – hallelujah – he has seen the light, and that’s the Looch we’re going to get all of next season. Subject to the aging curve of course!
What if instead of his actual seasonal numbers, we pretend* he scored those 55 points last season? Yes, I’m getting a bit desperate over here!
If we refit the aging curve using this new … um … “assumption” … it now looks like this:
*This whole section is something of an exercise in wild ass assumptions, I highly recommend against trying any of this at home
Ah, well, now we’re cooking with the most optimistic of gas!
If he can deliver a full season at the pace he was playing at before the New Year, subject to an aging curve, Lucic could deliver around 0.54 pts/game, or about 44 points (with a reasonable range of 35 to 53).
Hey, Looch isn’t pushing up the daisies after all, it turns out he’s just been pining for the fjords!
The massive blinking red-light cautions on this optimistic little side project though:
a. we’re manipulating the data because the initial results are unpleasant. While we are rolling in some additional information regarding Lucic’s career, it’s not exactly rigorous ya know
b. the fit is no longer as good as before, which is a bit of a warning sign on top of that
c. the new curve also means this last season was above the curve – so next year, far from expecting a bounceback, maybe we rationally expect a decline below the projected number. More rain on the parade even when we’re pretending it’s a sunny day.
What does all of this tell us about Looch?
It tells us that … with all of the ups and downs of his career scoring … and his two lousy (5v5) years in Edmonton … and the hot-cold pace of last year … when it comes time to try and project what’s going to happen next year, uncertainty seems to be the order of the day.
A straightforward (albeit well fitting) aging curve suggests that last years crappy 34 points would represent a bounce back year for Looch next season, with downside risk to a miserable 22ish points.
Manipulate the data a little bit, generate a few optimistically minor variations of the aging curve using some additional information about Lucic’s career, and we get a full seasonal projection range all the way to a higher-than-career-average 53 points.
Twenty two to fifty three points!
Now, if you put a gun to my head, I’m of course going to ignore the manipulations and just go with the original estimate. Let the data speak, as it were, even if the data are miserable rat finks who would snitch on their own mother for smoking a little pot. Data that I hope are really really wrong.
What it all really says to me, though, is that anyone who predicts with any measure of certainty what Lucic might do next year is probably full of caca!