Breaking Down Duncan Keith’s Tenure With the Oilers Thus Far

Photo credit:Timothy T. Ludwig-USA TODAY Sports
1 year ago
One of the most frequently-discussed topics this past off-season among the Oilers fanbase was the trade for Duncan Keith.
It all began in late June, when Elliotte Friedman reported that a Western Canadian team was interested in Keith. As days moved on, it was reported that Edmonton had indeed been in talks with Chicago for a potential deal involving the 38-year-old defenceman.
At the time, many speculated that Chicago would send assets back alongside Keith. After all, he had a lofty cap-hit of $5.5M. At least, it seemed likely Chicago would retain some salary. 
That wasn’t the case.
On July 12, the Oilers sent 23-year-old Caleb Jones and a 3rd round pick for Keith with no salary retained, alongside prospect Tim Söderlund (who was recently put on waivers for the purpose of termination). Keith was projected to play alongside Cody Ceci on the 2nd pair.
Subsequently, a large number of arguments and debates ensued across social media. Opinions ranged from Keith still being a capable two-way, top-four defenceman, while others felt he was hardly a good 3rd pairing player.
Personally? I thought Keith would struggle mightily in a top-four role. 
I despised the trade at the time, and I felt the most suitable role for him would be as an offensively inclined 3LD. I feared he would weaken the top-six, and would get shelled against top opposition.
So, how well has he played thus far in Edmonton?

*All stats via EvolvingHockey and PuckIQ unless stated otherwise

Keith’s numbers have seen a significant improvement from his tenure in Chicago

The primary concern with Keith was never his offensive ability, but his defensive play. His offensive results were generally above-average in Chicago, but his results at the other end of the ice were awful.
In 20-21, his GA/60 (goals against) relative to teammates was 229th out of 231 defencemen with at least 200 even-strength minutes played. His xGA/60 (expected goals against/60) relative to teammates was 231st, or in other words, dead last
Have those dreadful numbers continued in Edmonton so far?
The answer is no.
Statistically, Keith has largely improved from his most recent seasons in Chicago. He currently has a 52.2% goal differential, and a 52.2% expected goal differential at 5v5, ranking in the top-six among Oiler players in both categories.
His offensive numbers stayed the same generally, but his defensive results have vastly improved. No player in the entire NHL saw a more significant jump in their xGA from 20-21 to 21-22 than Keith. The Oilers have evidently struggled defensively this season, but they allow shots, goals, and scoring chances against at a lower rate with Keith on-ice.
These results have certainly been a surprise to me. But what if we look deeper than just his raw totals?

Let’s dive deeper into the numbers. How well has Keith played against top opposition? What about his play without McDavid?

Here’s how Edmonton’s defencemen have performed playing against elite competition, or in other words, the league’s top lines/top defence pairs.
*Note, it’s important to keep the sample size into account, which is why the TOI bar is beside each defenceman
Although William Lagesson ranks atop this list, he’s played a mere 25 minutes against elite competition. Consequently, it’s fair to say that Darnell Nurse has been Edmonton’s best defenceman when deployed against top opposition. Evan Bouchard also ranks pretty high in regards to DFF%, although he’s had a bit of bad luck in regards to GF%.
As for Keith, he’s spent roughly 32.6% of his TOI against elite (which is nearly ⅓), but he’s been quite poor when doing so. His metrics against elite ranks near the bottom of the team.
Only Markus Niemelainen and Slater Koekkoek have allowed DFA against elite at a worse rate than Keith, although neither Niemelainen nor Koekkoek have even played 50 minutes against elite. With sample size in mind, it’s fair to deduce that Keith has been Edmonton’s worst defenceman when deployed against top opposition. 
Here’s how Edmonton’s defencemen rank against players in the “Middle” QOC (quality of competition) tier.
Once again, Nurse ranks near the top of the team in regards to DFF%, although his unusually low oiSH% drags his GF% down.
Keith performs much better when deployed against the Middle QOC tier, in terms of both DFF% and GF%. Keith’s relative DFA/60 (Dangerous Shot Attempts Against/60) ranks 2nd among defencemen, just behind Broberg, who’s only played 25 minutes. This is a big step up from his dreadful defensive metrics against elite.
Finally, here’s how Edmonton’s defencemen have performed against players in the “Gritensity” QOC tier.
Keith’s DFF% against Gritensity is even better. Edmonton largely improves both offensively and defensively with Keith on-ice when the opposing team’s 3rd/4th liners are on-ice.
Most of Edmonton’s defencemen have had similar metrics against each QOC tier. Nurse, for example, has been above-average when deployed against all three tiers, while conversely, Barrie has been poor against all types of opposition. By contrast, Keith’s performance has been the most variable depending on the type of competition he plays.
Now, how does each defenceman play without McDavid? How reliant is each defenceman’s performance on the amount of TOI they spend with McDavid?
*Note: DFF% and xGF% are very similar
Back to Keith, he’s shown solid chemistry with McDavid, as Edmonton currently has a 69 GF% and 68 xGF% when Keith and McDavid are on-ice.
However, no defenceman sees a higher drop-off from their metrics with McDavid, to their metrics without McDavid, than Keith. As one would expect, every player will perform worse without the best player in the league, but relative to the rest of Edmonton’s defencemen without McDavid, Keith’s SCF% and xGF% totals aren’t exceptional (the GF% totals here are quite odd, though. It’s not uncommon for GF% to seem unusual/bizarre in limited samples such as these).
Overall, these are certainly fascinating results.

So, what’s the explanation behind these results?

On-ice metrics like GF%, xGF%, etc, can be exceedingly useful. These stats are useful in giving us an idea of a player’s performance, and can be a very descriptive tool.
However, the main flaw in these stats is that they can’t explain how the player achieved those numbers. 
To quote Eric Tulsky, who’s currently Carolina’s assistant GM: “Public models generally rely on inferring a player’s impact from looking at how the team’s results change when he is on the ice – and that can be done very well, and can be powerful. But when you want to move from talking about what happened to talking about why it happened or what could have been done differently, you often need a richer data set.”
Results matter, but it’s always important to recognize how those results were achieved in the first place to gain a superior understanding of a player. Context is always important.
The explanation behind Keith’s results in Chicago is an issue that seems to have persisted for quite some time; his zone entry defence and gap control.
 (Stats via @ShutdownLine and InStatHockey, a private NHL stats company. The stats from InStat are generally unavailable to the public, but thanks to @JfreshHockey, we have a small glimpse of them)
The visual shown above is from Keith’s season in 20-21 in Chicago. 
Keith’s primary strength is his passing/breakout ability. Jason Gregor has also mentioned how Keith ranks near the top of the team in completed passes this season.
On the other hand, as stated before, Keith’s defensive metrics in Chicago were terrible. The major reason for this was his inability to prevent chances off the rush and deny entries/carry against, as shown in the above visual. Keith might have excelled in this facet of the game in the past when he was younger and had quicker feet, but that evidently isn’t the case now. 
These issues have persisted this season. Per Corey Sznajder, out of the seven Oilers defencemen with at least 150 EV minutes played this season, Keith ranks 2nd last in entries with chances allowed/60, only ahead of Slater Koekkoek. 
However, what if his sustained inability to defend chances off the entry is the major reason for his odd results in Edmonton? Let me explain.
The quality in forwards has probably been the most significant change for Keith when moving from the Blackhawks to the Oilers. Edmonton’s forward group is a vast improvement on Chicago’s forward core. 
The forwards that currently surround Keith are spending more time in the offensive zone, which likely means they’ll spend less time in their own zone, at least in comparison to Chicago. The Oilers’ forward core is superior in maintaining possession in general. Chicago had an EV CF% of 45.2% last season, while Edmonton has an EV CF% of 51.4% this season (CF% is often used as a proxy for possession).
In addition, the vast majority of Edmonton forwards have a much better CF% against “Middle” and “Gritensity” competition than Chicago forwards in 20-21 (only Ryan Carpenter had a positive CF% against “Middle” last season).
So, what if this means that Keith just has to defend chances off the entry at a lower rate with Edmonton’s forwards, as opposed to Chicago’s forwards?
He still struggles to defend the rush. That hasn’t changed much, and it’s still an issue that has carried on from his time with Chicago. At this age, it’s exceedingly unlikely he’ll improve in this facet.
But perhaps his defensive metrics have improved because he just has to defend the rush less as a result of Edmonton’s forwards being superior in puck possession.
As stated in this piece by Sportsnet, Keith led the league last season in defensive zone blocked passes; his in-zone defensive play isn’t his biggest concern. Keith’s poor zone denial abilities were probably a more significant cause for his dreadful results in Chicago, as opposed to his ability to defend in his own zone.
Combine Keith’s potential lower workload in terms of defending the rush, alongside his above-average in-zone defensive play, and the result is a player with solid defensive metrics.
This theory is further supported by his results against different levels of competition, and his play without McDavid as shown prior.
Keith’s poor results against top opposition could be due to his inability to hold his blue-line, being exposed against elite transitional players who excel at generating chances off the rush. Meanwhile, he doesn’t have to defend rush chances at a high rate when playing against 3rd/4th lines, while Edmonton’s forwards are superior in possession.
Furthermore, McDavid is by far the best player in the league in the offensive zone. Again, Keith’s workload isn’t as high when he’s deployed alongside a prominent transitional and possession-based player such as McDavid, and McDavid’s mistakes in his own zone can be covered up by Keith. Every defenceman will perform worse without McDavid on the ice, but this is probably why Keith suffers the most when he isn’t deployed alongside him.
This is merely a theory/hypothesis, but several facts support it.


Keith has exceeded my expectations so far. I’d be lying if I said he didn’t. I feared Edmonton’s top players would perform worse with Keith prior to the season, but he’s turned out to be a fine complementary player to the top-six.
But has he performed at a level that justifies Holland acquiring him at full salary? Has he been exceptional to the point where Edmonton doesn’t need to search for another LD?

The answers to both of these questions should be no.
A general summary of Keith’s play in Edmonton so far is that he’s an excellent passer with above-average defensive ability in his own zone, but he struggles with holding the blue-line and denying chances off the rush. He thrives when deployed alongside the best player in the league and/or against weaker competition, but his play massively deteriorates against tougher opposition.
Per PuckPedia, a cap-hit of $5.5M ranks 43rd among defencemen contracts, and that definitely doesn’t sound like a top 43 defenceman. Not to me. 
That sounds like a player who could still play on a 2nd pair with the right partner, but a team that wishes to contend should undoubtedly acquire an upgrade over him. A good top-four defenceman just can’t perform this poorly against elite competition.
To add on, I think Keith has massively benefitted from the breaks in the schedule, but can a 38-year-old defenceman like him sustain these numbers as the schedule gets tighter, and the competition gets tougher?
More significantly, if Keith doesn’t retire following this season, it could lead to some cap issues. Per PuckPedia, Edmonton has $7.9M in cap-space this coming off-season. Puljujarvi and Yamamoto require contracts, only eight forwards are under contract for 22-23, and Edmonton also needs to acquire a goalie.
Although Keith has performed better than I thought, the trade still isn’t a smart decision.
At the time of the trade, the Blackhawks had practically no leverage. Chicago needed to clear cap-space, Keith had an NMC that he would only waive for a short-list of teams, and Edmonton was the clear front-runner for Keith. They had plenty of time and cap-space to negotiate a better deal.
If the Oilers retained 50% on his salary and weren’t the team giving up assets, the trade could have been fine. The fact that nearly the exact opposite occurred is simply poor asset management from Ken Holland. 
The player has surpassed my expectations to an extent, but he hasn’t been exceptional either. 
The trade was brutal at the time, and it still hasn’t aged well at all.
Find me on Twitter (@NHL_Sid)

Check out these posts...