Wednesday, February 20, 2019

Caleb Jones is the Bakersfield Condors MVP

I was going to wait to call something like this until I had every player's data worked out; on ice and individual; with-you and without-yous, video review too and so on - but by the time I got to sorting everything out for Jones I'd seen enough and started to alter the outline and frame of this piece to reflect his status as the most valuable player for the Bakersfield Condors this year.

It was a bit of a worry, Jones's player-identity lines up with my preferences in a defenseman (the kind I would ice six clones of) and we'll get into what that means, but the trouble was essentially that I was getting ahead of myself based on the eye-test before I had the data to challenge and adjust the perception that was forming, which is always something you want to avoid in either direction. This was also after the early part of the NHL audition he earned, where I was very impressed with his third pairing work and had no expectation of success to disappoint when it came to his entirely too early 'promotion' to Larsson's pairing.

In fact, I remember the exact play when I realised the player had - crucially - taken his eyes and his sense with him to the NHL, while keeping the heart rate in a healthy range. It's that point when you realise that someone gets it.




It's kind of funny that at the time, and still now, I can't really tell if this is something of little significance that bias is blowing up because of the trauma we know as Oilers fans from watching defensemen puck-watch or puck-chase or be puck-paralysed in this exact situation so many times, or if instead it was a cue that an unbiased, normal, non-Oilers-watching-brain-having person would take note of.

After tracking another 16 games where the shot-share dominance persisted (probably more-so than I have recorded because of the score-effects of the recent win streak) I started to want to pick out the strong-links, find who the drivers were. Marody and Benson were obvious suspects, but they've always been outscored understandably by the glued-together veteran line that was better at busting cycles in the defensive zone, and generally more consistent at puck-recovery as a trio all of elsewhere in the rink.

But I also had noticed Jones many times, doing things that you seem to pick up on when you're a tracking project like this. A lot of the stuff that we went through in my Sekera post, the back three-quarters. Picking the puck out of traffic in the defensive zone and wheeling away, testing forechecker's cardiovascular facilities, solving the puzzle of the neutral zone with a neat pass. The type of play that you encounter if you're watching a strong possession defenceman for the first time in awhile.


...


A fortunate part of my confidence in the identification of Jones as the main engine on the blueline at 5-on-5 is courtesy of the fact that his tracked games were split up in such a way that's sometimes rare for regular defenceman. In my sample - which is an accurate reflection of his season; all but five of Caleb Jone's games in the time period are tracked, and the time period tracked is up to February 9th - he's played on a default pairing with Stanton ~43% of the time, Lagesson 48% of the time, and the other games were with Gravel.

The reason that this is fortunate is because of a problem that's always run into: Whether you're model-building, pursuing a wider view by informing yourself of a player's stats (advanced or traditional), or even just using the eye-test, is that the difficulty of accurately dividing up blame and praise for any player's contribution to any given situation is a large part of the fundamental and daunting barrier of analysing the game - that's much more easily scaled when you have both a large sample size to draw from, in which there's a large array of teammates (of varying qualities and identities) to help isolate the player's effect on the game.

So, what'll happen a lot with defensemen is that they'll end up playing a ton with just one partner, meaning their time apart with which you're trying to solve a question based on an assumption (they're both good together, but who is better apart is the better player) often has a couple major maladies: the sample size is small, and the player that is better in the coach's eyes plays the harder minutes away from their regular partner. Say there's a defense pairing with one savvy veteran and one talented rookie - when the rookie is already in the process of sheltering by virtue of playing with the smart established player, it logically follows that that rookie's situations away will be soft ones where the coach is capitalising on the tactical vulnerabilities of the opposing team, and the veteran's ones will be mostly spot duties where the vulnerability is his team's own.

This is a solved problem in my sample when it comes to Jones's defense partner, and with the forward line teammates. Jones' pairing is typically the top/tough comp. one (my partially-educated guess, though, is that Woodcroft runs a top-four style Dcorps as opposed to there being a top-pairing and two others) and so they do play more with the top-six forward group the Condors ice by virtue of both who they're deployed against and the fact that they play the largest portion of game-time and so do the top-six forwards.

In terms of shot-events, I manually went through and counted each that occurred where Jones was on the ice with one centre and no others. Around 16.9% of such events with Marody, 20.1 with Currie, 30% with Malone, 17.7% with Vesel, and 15.3 with Esposito.

Given that our sample then has a good division between both defense partners and forward lines, I'll assert that if, for the majority of them, their better minutes are with Jones as opposed to without him, we can safely assert that (within our sample) Caleb Jones is the man most pushing the possession margins for the team on an individual basis.

Let's find out if that's the case, starting with the two defense partners that Jones started and spent the vast majority of his 5-on-5 minutes with for 90% of the games in our sample:





This is pretty stark. I imagine that it's somewhat likely the inequality of possession in the AHL between teams and their players is greater than in the NHL, as teams have raced (for longer) to close those gaps in the highest league where they were first noted, discussion, and subsequently addressed to the point where CF% is less likely to predict GF% than it was in the past - and this discussion (in public) hasn't really occurred in regards to the AHL.

What I mean is that a player with Jones's ridiculous shot share (64.47 CF%) is perhaps a lesser standout (percentile-wise) in the AHL than the NHL.

I'm attempting not to fawn.

Below is the same deal, but with the centres. One detail of note is that since these are events that occurred when each centre was the only centre on the ice, excluded is the minutes spent the recent line of Benson-Marody-Currie (which were probably positive shot-share) and the season-long phenomenon of DZ faceoffs with whatever centre plus Luke Esposito taking the draw (which were probably negative shot-share). Another is a reminder that we saw Benson, who played on one line for a majority of my sample in my previous posts (the profile on him and the update on the team) crank up the shot share for the team in a major way. Does he or his line with Marody pull up Jones's numbers?




There's a sore thumb here (and in the next) that I'll address real quick: The Josh Currie numbers are a result of a few very successful lines deployed in the beginning of the season when the Condors were especially dominant on the shot clock, combined with four later games where the Benson-Currie-Puljujärvi line essentially torched every 2nd shift for 3-7 shot attempts, resulting in dumb stat-lines like 26-5 CF-CA goosing some already strong numbers to the point of absurdity. It should be mentioned, though, that most of those dominant shifts were (as we see here) spent with Jones and given what we know so far, it's fit to wonder how much the defenseman figured into that equation.




Again, the uniformity here does a lot of the work for you in terms of figuring out who's doing what. The important bit here is that the two taller blue pars are representing a pairing that, as we've seen above, the centres do worse away from. This is a boat-floating player, in terms of possession.

Let's wade into goals. More specifically, shot-shares against goal-shares, team-wide, Jones-on versus Jones-off.

Something that has to be noted before we do, though, is that the numbers I'm now going to be using are going to be the ones that are recorded officially by the AHL itself. They're not limited to my sample, and so I won't be speculating on the difference between the shot and goal-share as if the delta is from the exact same set of games, just games from the specific time period (start of season up to and including the game on February 8th, 2019). Essentially the question to be answered is, how much were the shot-shares from my sample representative of goal-share of the team over that period of time in totality?




Interestingly enough, goals outperform shots when Jones is off the ice, and shots outperform goals when he's on. The shot-share dominance is still too large - the goals can only wander so far from it (typically goaltending only goes between .890-.930, shooting 5-11, you're not going to end up south of 50% goals if you're pushing 60% of the shots very often once the sample size grows large enough) for it to go into an awkward place where the possession is essentially for naught. Still,  Jones not standing out in 5v5 primary points per hour either (that stat ranks the Condors regular six in the order of Day, Stanton, Jones, Lagesson, Bear, Lowe) prompts an exercise in establishing whether or not there's a bunch of inflationary corgis baked into this thing offensively.

Turns out, in terms of conversion, Jones has the typical talented young defenseman effect of increasing the GF/60 to a great extent and the GA/60 to a lesser one en route to his net beneficial effect to the team's goal differential over time. Credit should be given to Woodcroft for cashing this, because it's a gambit that commonly has been rejected outright by traditional hockey coaching and management staffs - to their profound disadvantage - for at least the past decade-plus that we have the data to reference, and likely many before.



(For those interested in how I got these numbers - you're not going to find them by googling - these estimations were the product of grabbing the estimated TOI and player-level and team-level on-ice goals for numbers from prospect-stats.com, and then using the fact that NHL teams played 48 minutes 5v5 per game on average this season and using that as a rough reference to give the Condors 2352 5v5 minutes played given their 49 games played at time of writing. Then I just divide and conquer the on/offs from there.)

Remember the breakdown of partners from earlier? The defensemen Jones was partnered with for 90% of the games I tracked (and, since I still watch the games I don't track, I can tell you that ratio applies to all Condors games, or close) are 4th and 6th among the Condors six regular defencemen. That primary implications (if these estimations are close to accurate) of that, in my view are population two:

  1.  Jones' GF/60 would likely be even greater if he were partnered with better offensive facilitators.
  2.  The theory of Jones' numbers being inflated by the possibility that he could have been deployed in the most favourable situations (most minutes with the top-six, offensive zone draws, OTF abuses, weak competition) takes a bit of a hit when you see a lead this large over his partners in those minutes - for the record, I'd also wager the third pairing with Logan Day gets the most red meat in terms of deployment. 
Even I didn't expect that he alone would be the primary driver of actual goals per hour on the blueline.

...

An exercise of interest for me, under the umbrella of post-mortem team analysis, will be to dig into how much of the Condors shots-for at 5-on-5 come from defensemen, or from the point in general - and then slice that up into each different player and how they effect that. From a visual standpoint, Jones doesn't have the best shot of the Bakersfield Dcorps but he certainly finds himself or others in tighter on a more regular basis and is exceptionally (among this group) aggressive in jumping up in the play. Only Logan Day is similar in that regard, but he certainly doesn't pick his spots as well.

It'll be interesting to observe whether or not Jones' on-ice GF% will come closer to his CF%, and if it does whether or not that's via the share of the shots decreasing or the conversion-rate/save-percentage increasing; and all of the same in regards to the inverse situation of the Jones-off hours.

Last thing on the hard possession numbers: There's a split in performance from before he departed to the NHL, and after. There was some talk from Ken Hitchcock right after it was announced Jones was to be sent back down, the coach putting forth that the player had lost confidence and was in a bad place on the ice and off. That could be one explanation for the splits, but Jones has also put up 9 points in 10 games since returning to the AHL. I'll suggest the team is playing different games now, they're leading more, their overall possession advantage has eroded slightly (but goal differential has improved) as you'll learn about in my next team update if that difference is still present at the time of publishing. This likely combines with a readjustment to the game he plays in the AHL as compared to the one he does in the NHL; the shift that's likely more comfort than confidence based.

Let me give you a look at the early game logs; and mind this is also a good demonstration of what the game-to-game CF-CA splits of a player who puts up near 3:2 ratios really looks like. We'll move on to the video after this - and you'll see what I mean when I talk mentioned the different 'game he plays in the AHL' - but check out this run:





...



I want to start by reintroducing the division of a defenseman's game that I used to contextualise my look at Sekera's return to pro hockey, that you can read about in the beginning of that piece.

Jones lays waste to the AHL in this area, but I want to cut the back three-quarters in half again - though not by area. I want to instead mark the difference between a set, proactive play and a broken, reactive play. It would be significantly more difficult to communicate this distinction so cleanly with one clip of one ~30 second run of hockey time, but luckily there was a great example within those parameters to be found.

To quickly brief you on it, the first part is a clean breakout by the Condors where they lure each of three Roadrunners, pulling them far enough to make a backcheck mightily inconvenient and sendin Jones on his way for a clean-to-clean exit-to-entry. The brain is there on this play, but so is the head-up, four-way mobility by Caleb himself, especially in the last pass out of the zone where he seperates from his check far and fast before receiving the puck. Later, you'll see an example of Jones taking advantage of the timing and spacing of the opposition, on a more broken play, that he must react to.




When I'm tracking games, the primary goal is to get all of the data down accurately and quickly. Aside from that, I've also made a habit of marking down specific time-stamps of plays to return to and clip, with just short words or phrases making an umbrella under the massive array of unique sets of play that happen in the game of hockey. One of them, is 'stop'. It's when a player that's in front of the puck stops the flow of play in their direction; 'stop and go' is when in that same stopping motion there's an immediate and successful pivot that sends the play safely (for their team) and dangerously (for the other) back the way it came.

One can break down every little part of a player's game on video in equal-ish partitions, but it's much more efficient to take a combination and blow it up for the majority of the analysis if your goal is to give the best impression of both a player's identity and how effective they are with it and I'll show you now how it works: Here's a compilation of Caleb Jones stopping, going, and his 'stop and go'.




There's a lot of common themes here. The opposition generally doesn't know if he'll pass or skate; when or where he'll do either. There's a useful hesitation (Oilers fans will know what useless hesitation looks like) that fakes out opposition without Jones' forward linemates running out of track. This area of the game (the reactive half of the back three-quarters) is just so important in the environment that is the neutral zone in today's professional hockey game.

An argument that plays like in the clip above are the majority reason that Jones drives possession is a convincing one; the same if you're arguing that he sets himself up well for an NHL future by doing this at twenty-one years of age. Add to that the commonality between what Jones brings and what the Oilers lack and it's hard to imagine a 2019-20 season that doesn't see Jones in Oilers silks for much of the campaign, barring injury.

...


The offensive zone game is where Jones's ability is less unique, but still effective enough to become an NHLer - though if he'll be powerplay-playing defenseman is better left to a scout or one of those sort than I. He's not being groomed to be one, I haven't done a ton of work on the Condors powerplay but often they'll play two defencemen and in points it's a split between Logan Day (who's essentially deployed as an offensive specialist), Caleb Jones, and Ethan Bear. Of the 26 PPP earned by defensemen on the team, those three make up 22 of them: Day with 9; Bear with 8; Jones with 5.

I'll top off my doubts with a two-parter that I have of Jones' powerplay game that I feel pretty accurately represent what his impact is at the 5-on-4 game-state. Phrased shortly, he's as good as you'd assume (based on his 5-on-5 back three-quarters game) at facilitating the zone entries, which is critically important, but just regular good as a distributor once everything's set up. Here's an example of the Condors entries, two unique-ish ones regular speed and the default, go-to ones sped up:




...


Below is an opening clip of what illustrates Jones' ability to take and make a pass that's beneficial on at 5-on-4, followed by a series of clips that gives a good reflection of what's the game-by-game, more unassuming participation of Jones on the Condors' powerplay. He's not Erik Brannstrom (who has a billion powerplay points in this league as a teenager) so it's more a matter of getting your teammates to the latter stages of what becomes a scoring play than directly participating in them.



...

I want to make a quick pivot to the player's shot, and goal scoring. Jones' goals come from hard slapshots, and that's primarily the setup for his shots - where he leads the Condors with 2.29 shots per game all-strengths - as opposed to trying to sift wristers through, and he'd rather use a slapshot for an inside-out deflection (from the centre-slot, say) than a slap pass for a outside-in deflection - backdoor tap-ins and the like. Here's the player's clean goals.



...

As you could probably guess seeing as the player facilitates offense as opposed to being a point-producing passenger, he's pretty close to full value on his assists. Some of them are more remarkable than others; being plays that essentially span the whole ice surface, turning a blocked shot or pass against into a fast break goal. Here's some of my personal favourites.



...

At this point, I hope to have hypnotised the reader to exhaustion, and that their brains are now fertile to any conclusion I'm willing to put forth due to the sheer enormity of labour in writing this piece overpowering any opposition . That the mountains of evidence are so vast that even if I've interpreted them incorrectly, there's no easy defense against a turning-handle mounted machine gun spitting out innumerable falsehoods.

Seriously though, I do hope that I've made a persuasive argument for Jones as the most successful Condor, but before that or anything else I hope the most that I've given a good snapshot of what Jones has been this season, and the type of AHLer he is. I hope that someone who's never watched Jones or the Condors at all this season gets a good impression of what the prospect does well and how well he does it - and that regular observers of the team or closer examiners of Oilers prospects have their knowledge base expanded as well.

I'll end now, as in my profile of Tyler Benson, with my prescription: It'd be wise to not outright gift a third-pairing role on the Oilers to Jones outright, but believe your eyes if you see that he torches all competition for that slot; between NHL hopefuls and NHL veterans alike.

3 comments:

  1. extremely well done. Lots of run-on sentences, but great info.

    ReplyDelete
    Replies
    1. haha, thanks. i still write exactly like i speak, still figuring out how to remedy that without losing too much of my voice.

      Delete