Jagten

Yea but we have Longstaff and Miley in midfield. It’s never going to happen

Ah, man. We nearly got good money for Almiron

Completely agree. Yes, that’s correct. I can personally confirm our new head of data doesn’t know anything about stats/ML/AI/whatever, but probably best I don’t expand on that. It’s the only info I do have on Ashworth’s job here and it’s a bad miss.

At West Brom he oversaw the appointments of Di Matteo, Hodgson and Michael Appleton. At England, he was involved in the appointments of Hodgson, Allardyce and Southgate. At Brighton, he was no doubt involved in Potter, which is his only obvious hit. Further, as someone who works in the space, I have found his data appointments here pretty disappointing. I don’t have enough information to properly assess him, but I don’t get what’s so obviously excellent about him.

I’m guessing many things here but I’ve been a little surprised at Staveley’s continued presence in operations

Welcome home Phil Giles.

Using the individual goal probs, based on Opta's model, I calculate there was a <1% chance we scored 3 or more goals.

I think he could have reduced the prob of the first-time shot going in by a similar % if his position wasn't so shit. Others were very involved in creating the opportunity for them though.

Dubravka at least 50% to blame

Burn often becomes a left-sided centre back in possession, with Trippier basically moving into midfield. Livramento a pretty different profile.

Tino for Burn is not tactically like-for-like, and it restricts Howe's preferred use of Trippier. I'd still do it, because Dan Burn is shit, but whatever.

I guess it turns out one good game wasn’t a low-variance predictor of our return to form

They've done a good job of fucking with our build up. That was pretty bad.

Phenomenal. Great performance

This has been really good

It's hard, for sure, but I don't agree with this. That said, it certainly doesn't lend itself well to the gen pop analysing a bunch of closed-source proprietary metrics. Inferring things like xG and its derivatives requires a model, which requires assumptions and skill. These are PhD-level questions, and not ones most people can answer, so I understand the frustration when they are beyond even most advocates.

Most free analysis is shit, most people interpreting it don't know what they're doing. I'd recommend feeling however you want, and I would probably be less inclined to engage if I didn't understand the basics of linear models. Enjoy the win, and don't use a spreadsheet.

Inferring latent quantities from noisy data is an extremely valuable skill, common to basically everywhere that matters, which can very definitely afford you at least your own basement. I think Joelinton is one of the worst starting PL central midfielders when it comes to passing, yes. His out-of-possession game is good and has suited us very well however.

Passing % isn’t a good way to evaluate this. Generally, you need some model of his passing “value”. For example, Statsbomb use their “On Ball Value” model (OBV), which aims to describe the downstream xG delta arising from an action (in this case, whenever Joelinton passes the ball). Ignoring the Onana overlay, his passing OBV is close to the worst in the league

This is probably a stretch, given he can’t really pass.

https://www.theguardian.com/football/2020/dec/19/top-football-clubs-relying-on-transfer-valuations-made-by-volunteers

His Transfermarkt value is 30m euros, which I believe is a legit point of reference commonly used in negotiations. We can justify that fee, which would be excellent business.

To sell Tripper for a fee that represents a positive risk/return would require a high-value buyer in need or a short-ish term fix, which Bayern are. There's no guarantees there's a market for that in the summer. Likewise, there appears to be no market right now for an Almiron, who I would also much rather sell. I agree there's obvious downsides, and I don't have the data to make a confident decision, but I tend towards doing it.

It's an expectation over the next ~5 years though. Of course, he cannot be replaced right now, but we can replace his impact over that duration, during which he stops playing.

His transfer record isn't actually that bad, even if it's difficult to know the level of involvement he's had in every deal. Clearly there was great talent ID in moving Bournemouth through the football league, even if this doesn't always scale well to PL. In his last two years at Bournemouth, he also spent circa 60m on non-UK players: Danjuma, Lerma, Rico (not these guys were a great use of 60m).