Castagne and the quest for a more fault-tolerant system

Using the surprising links to the Leicester full-back as an excuse to better understand how to fortify the Arsenal squad. Plus: notes on Cancelo, Fresneda, and Boey.

Hey you. A lot of you found me on the Gunners subreddit, to which I say: thank you. I hope you’re holding up OK with the subreddit offline and your daily routine interrupted during the blackouts. Feel free to use the comments section in lieu of all that. I’ll be here. 🙂

---

To better understand squad-building in the beautiful game, and gain insight into how Arsenal can better thrive within it, let us first borrow two concepts from the glamorous world of computing. (And they say we reduce players to numbers…)

The first concept is fault tolerance:

Fault tolerance is the property that enables a system to continue operating properly in the event of the failure of one or more faults within some of its components. If its operating quality decreases at all, the decrease is proportional to the severity of the failure, as compared to a naively designed system, in which even a small failure can cause total breakdown.

These issues are commonly referred to as a “single point of failure.” See:

This, in computing terms, is bad.

For one way out of that problem, we introduce our second term — redundancy:

In engineering, redundancy is the intentional duplication of critical components or functions of a system with the goal of increasing reliability of the system, usually in the form of a backup or fail-safe, or to improve actual system performance.

In many safety-critical systems, such as fly-by-wire and hydraulic systems in aircraft, some parts of the control system may be triplicated, which is formally termed triple modular redundancy (TMR).

By creating this “fault-tolerant” system, engineers can prevent these points of failures from torpedoing the whole project.

Alas, this is an Arsenal newsletter, so you probably see where we’re going with this.

Despite the glistening highs of the Arsenal season, with a full 50 points through 19 games, some faults were gradually introduced into the system. Some were proven to have ample redundancy. With Gabriel Jesus injured, Arteta hot-swapped a backup (Eddie Nketiah) for minimal downtime; when that backup picked up a ding and started to falter, Leandro Trossard stepped in to fill the void.

Some were more mixed: the team was able to cope without Oleksandr Zinchenko’s progressive passing better in some games than others. Likewise, the dipping form of formerly-instrumental Thomas Partey had varying impacts down the line.

But most Arsenal fans would point to a worst-case Europa League loss against Sporting CP as the day the worm turned. It went a full 120 minutes; it was squandered on penalties; it made us all Big Sad; and most portentously, the team lost both Takehiro Tomiyasu and William Saliba for the season.

In the weeks to come, the system was proven to be insufficiently fault-tolerant. We learned, the hard way, why the big projects require “triple modular redundancy.”

Fixing this is the primary goal of the summer window.

---

Enter Timothy Castagne?

Dear readers, we are gathered here today to discuss a transfer report by Belgian journalist Sacha Tavolieri.

Supposedly reliable on such matters, the scribe is reporting Arsenal interest for the Leicester City full-back:

Is it likely to happen? Who knows. We’ll grow dizzy with transfer speculation in the months to come, as we do every summer, and we can’t (and shouldn’t) cover them all.

But with some interesting questions about depth and quality being posed here, and a player we could serve to learn more about, I thought we’d dive in a bit more.

Castagne started at Genk, where he “exploded” the physical tests. Originally a defensive midfielder, his work-rate and speed suited the touchline a little better, and his time overlapped with that of Leandro Trossard, Wilfred Ndidi, and the sole managerial stint of Arsenal assistant Albert Stuivenberg. (I also just realized that Stuivenberg coached Edon Zhegrova, who was another surprising transfer report last summer. Airpod Albert clearly has some pull.)

From there, he moved to Atalanta in a €6 million deal, and spent the next three seasons rotating on both sides, playing as an attack-minded wing-back. He crashed the box to score a pivotal first goal in the Champions League to send the Nerazzurri through to the Round of 16. He displayed a skill and dynamism that earned him a £21.5m move to Leicester, a decision at least partly informed by a lack of communication with the manager at the time:

“After the game I was asking: ‘How did I do? What did I do well?’ either to the assistant or the head coach, that would have helped with my confidence but I didn’t get that at Atalanta. Now I’m receiving it and it has been very good.”

The same article mentions an interesting nugget about moving to Leicester — which wasn’t really about moving to Leicester at all:

[H]e had several reasons for welcoming the move. One was a yearning nurtured in childhood, when the appeal of the Premier League owed a lot to a glorious striker at Arsenal. “I’ve always loved Thierry Henry, he has been my idol since I was a young boy,” says the 24-year-old, who will fulfill a dream of playing against Arsenal if picked for Wednesday’s Carabao Cup tie at the King Power.

He has since played in 112 matches for Leicester, starting in 101 — including the FA Cup Final win in his debut year. He’s also played for the national team, securing three starts in the World Cup. Along the way, he’s battled some injuries, with the nastiest being a double eye-socket fracture in Euro 2020:

This year, he’s made 37 appearances for a thoroughly disappointing and thoroughly relegated Leicester side.

Overall, the time has most often been spent at the RB/RWB position — but he’s done about half as much work at LB/LWB, and non-trivial minutes as a CB in back-threes as well.

---

A deeper look at his game

They say centre-backs are notoriously difficult to sift through statistically, as so much depends on the style and form of the team, and the role of the player; over the course of this season, I found full-backs to be an even tougher nut to crack.

While players are always dependent on manager-led factors, there are ultimately more commonalties among groupings of strikers, wingers, central midfielders, and centre-backs. For full-backs, almost everything comes back to some external variable: are they on the left or right? Are they inverted? How inverted are they? Do they overlap? Are they a wing-back? Are they defensive? How much possession does their team have? And so on.

Put another way, a full-back might be a full-back, midfielder, central defender, or winger. Oy.

Nonetheless, I created a pool back in March:

So I pulled a list of around 80 full-backs worldwide, added some flags for desired characteristics — progressive passing relative to age, etc — and will now present you with a more manual list of potential additions to the team. I’ve watched all of the players we’ll actually discuss to varying degrees, but may still have some blindspots, and despite calling this “exhaustive,” I’m sure I’ve forgotten some dudes.

A few players led the way (Mazraoui, Livramento, Fresneda, Boey), while dozens of others got mentioned.

I couldn’t evaluate everyone, so started with a list of of players under 26 and then added players who’d caught my eye otherwise.

It is perhaps telling that Castagne didn’t make that list. This wasn’t because I dislike his game, but more because he fell into the less-descript tier of “solid Premier League full-back.” There were a bunch of others in that category — including players I really like, such as Rico Henry — that got a passing mention or less.

Meanwhile, during the World Cup, I pulled a less age-restricted list of global full-backs to see how Benjamin White’s performances stacked up to the best. I called the stat “BLANCO,” and White ultimately slotted in at 7th at the time.

For his efforts, Castagne came in at 44th worldwide, ranking similar to Carvajal, Danilo, Álex Moreno, and Caio Henrique. Not bad.

Still, I felt like I had an incomplete view of Castagne’s game, and wanted to free myself of preconceptions while checking him out with new eyes. So I did what I usually do: speed-run through some relevant full-90s, and then see how the analytics can contextualize what I saw.

To better understand, I watched the following on ~2x speed:

• Liverpool, 15 May: Castagne plays LB against Salah

• Arsenal, 25 February: Castagne plays RB against Martinelli

• Croatia, 1 December: Castagne plays LCB (in a back-3)

• Canada, 23 November: Castagne plays RWB against Davies

I then watched some highlights from his Atalanta days, and a few other relevant playlists on crossing, dueling, and dribbling. Here’s what I found.

Some notes on the above:

• Caveats apply: Castagne is the eldest. Castagne and White play in the hardest league. Castagne and Fresneda play for relegation fodder, while Boey and White play for excellent teams.

• We should always beware of defensive counting stats for players on weaker sides; White is the best defender here, and has the fewest successful defensive actions. It may be an indication of skill, or it may just be an indication of having to defend a lot. Castagne is in the middle: Brendan Rodgers stuck to more attacking style of play than was probably practical, so Leicester still held 47.7% possession, but was also isolated a lot in a sloppy defensive shape.

• The big concern for Castagne are the bottom two numbers: accurate progressive passing, and loss index.

• You’ll see why I’m still drawn to Boey here, even given league-adjustment. Wanna see him jump?

But what did I see when watching him? Let’s debut a new segment called…

Sometimes maybe good

👉 Running and work-rate: In the ~4 games I watched more closely, and all the clips in-between, I didn’t see a single example of him having to conserve energy. He can run for the entire 90 and then some. This shows itself in a few ways. In defensive transition, he is active and engaged and direct, though occasionally imperfect (he probably challenges too much at times, and gets spun). One skill I probably overindex is shot-blocking technique; I like it as an example of a player successfully overriding some base human impulses with practice and technique, and he’s good there. His work-rate also shows in aggressive, exploitative runs forward. When he plays against players that cheat forward, as he did against Alphonso Davies, he’s excellent at immediately sprinting into the space behind on a change of possession. He also did this against Man City:

Along with improved crossing, White was starting to add that element to his game in the closing months and it gave the opponents one more thing to think about on Saka’s side.

👉 Positional versatility and readiness: He’s genuinely versatile, and not just versatile out of need. He’s good with his left, and is probably most interesting as a RWB doing some runs like the above. But when he settles down into the RCB/LCB of a three, as he does at times for Belgium, his game changes, but doesn’t fall off a cliff. In the game against Estonia that secured them their World Cup berth, he slotted in at RCB in a back-three, and finished 127/130 (98%) passing.

A “CB in a back-three, who tucks in and occasionally overlaps” is probably the best analog for White’s particular blend of full-backing, and for that, Castagne is well-suited. I’m not sure the best comparison, but two players kept coming to mind when watching him: Takehiro Tomiyasu and César Azpilicueta. His blocky athleticism reminds me of Tomiyasu, and his general work-rate and positionlessness has some Budget Azpi vibes. While I wouldn’t forecast him in the Zinchenko role on either side, it’s important to acknowledge the two types of inversion in the Arsenal shape: the Zinny type and the aforementioned White type. Castagne would be able to provide depth in the White role on either side, which increases optionality across the pitch — and could even serve as emergency RCB/LCB depth.

👉 Few weaknesses: He’s pretty good at most things.

👉 Arsenal love: He apparently is particularly enamored with this project, and is willing to wait for the opportunity. This may help ease out any concerns about playing time or role.

Sometimes maybe shit

👉 Pretty good: This is the least scientific way of looking at his game, and came up as a strength, as well. Looking through his tape, it seems to be permeating with pretty-goodness, at least by new Arsenal standards: he’s pretty fast; he’s somewhat technical; he’s pretty defensive; he’s pretty strong; he can carry the ball; he’s pretty good at passing. Is this an overly-practical mid-table starter, or the player to challenge the starters and truly lift Arsenal’s levels? It’s a worthwhile concern, and one I share.

👉 Passing and control: My biggest concern, by far. While he is excellent at making himself available for passes, he can often try to get it away at the first sign of danger, which can result in hurried passes into the middle of the pitch. He clears it a lot; controlling and passing such a ball is a big adjustment from a lower-table side. He also isn’t terribly accurate in medium (80.2%) or long-balls (45.5%), and doesn’t have a diverse array of switches or clips in his locker. He looks solid in some of the crisper short stuff. His crossing shows a potential for goals with a better side. Because of some of his shortcomings, there can be games where his teammates don’t actively look to feed him the ball, and he can become a secondary character as a result. There would be hopes that this cleans up on a better team, as I didn’t see quite as much of this in his first year at Leicester or his tape from Atalanta, but I don’t know.

👉 Lack of high-end speed: He’s plenty fast, and probably faster than White? But in terms of additive qualities for the squad, I’d be concerned about acquiring another full-back who can’t nullify high-end winger speed down the line. It would have been great to have a Boey on hand against Brighton.

👉 Dueling: I like how he approaches duels but it may be a touch impractical at times, which has led to a low challenge percentage (48.8%) this year. To his credit, his managers seem to move him around to blunt the best opposing attacker, as I saw against Liverpool (at LB for Salah) and Canada (at RWB for Davies).

👉 Dribbling: His open-space carrying looks fairly bursty and dynamic, but I didn’t see a lot of examples of him being able to dribble himself out of trouble (nor trying to).

---

Signing a defender, in conclusion

In all, here are some loose notes on the situation:

• João Cancelo, recently linked, is kind of a Catch-22. People can quickly forget just how special he is on the ball. We have little actual insight into what kind of a teammate he is, only assumptions (which may be true). If ever there was a player to galaxy-brain, it’s him: for all we know, he may have a few good seasons as a winger or CM left in him — he is special in every phase, and has the potential to be one of the most undervalued assets in the world right now. So why is he a Catch-22? Well, I’m fully on board if we get Rice/Caicedo to blunt the defensive issues; in that case, though, I have no clue where the funds come from.

• Cancelo being in a category by himself, my top four long-term, pure full-back signings are, in order, and in a vacuum: Tino Livramento (Southampton), Iván Fresneda (Valladolid), Sacha Boey (Galatasaray), Arnau Martinez (Girona).

• That said, questions abound — starting with Livramento’s health (which is a black box to us plebes), and Fresneda’s readiness (which isn’t). Depending on the details, these strike me as worthwhile risks to consider, given their special talent trajectory — especially when paired with the signing of a more ready option at CB, like Marc Guéhi or Josip Šutalo. My purer RCB rankings are here.

• That’s the big question looming over this all, however: how many defenders are Arsenal likely to sign? If only one, as is looking increasingly likely, different rankings are in order. This is because the lone signing of a project full-back, particularly Fresneda, will unfairly tax the system — White serving both as primary right-back (ahead of a project RB), and primary RCB depth will turn he and Tomiyasu into the “single points of failure” we are hoping to avoid. Fresneda needs to be slow-cooked, and in truth, moving to an Arsenal 1-2 years early is a big risk from both his developmental perspective and ours.

• My “sign one defender” rankings, then, would lead with Mohamed Simakan (RB Leipzig), and probably followed with Anel Ahmedhodžić (Sheffield United), because I am a madlad who thinks the brilliant 6’4” pool noodle could be an excellent CB/RB hybrid. Sacha Boey wouldn’t be far behind, because he offers a skillset (speed and dueling) that could keep him in the lineup even as he adjusts otherwise. Šutalo, Guéhi, and Lutsharel Geertruida would probably work their ways in there, as well.

• In short, I’d prefer either a package like [Šutalo/Livramento] or [Guéhi/Fresneda], or, if only one player can be signed, a player from that grouping of Simakan/Ahmedhodžić/Boey/Šutalo/Guéhi, starting with Simakan. Good CB’s are generally FB depth, but not necessarily vice versa.

I was not enamored with the Castagne links, and he still wouldn’t be my choice. After better acquainting myself to the nuances of his game, I understand the logic.

By introducing a high-floor player who can capably deputize both flanks, do the White Thing on either side, and even do some wide-CB’ing in a pinch — the team gets depth, of course, but it also removes both White and Saliba as potential “single points of failure” in one fell swoop. Castagne and Tomiyasu could help form the triple, modular redundancy we were after earlier — allowing the team to invert on both sides, better sustain faults and injuries, and more. You’re not losing a lot of sleep over Castagne starting.

To go back to those computing terms, he would essentially be a load balancer:

Load balancing can optimize the response time and avoid unevenly overloading some compute nodes while other compute nodes are left idle.

Now, in truth, my appraisal of such a “savvy” and “practical” move would depend on what Arsenal does with the savings that are hopefully found (just £10m for Castagne, please); just as your appraisal of this article may depend on your subject matter expertise with servers (I may have gotten slightly ahead of my skis here).

Castagne is a pretty good player, and I can live with almost any ancillary signing if Caicedo and Rice are on the way. If it’s just one, I’d look to level up.