|
Post by tachyon on Mar 12, 2020 9:07:11 GMT
How much does his barnsley half way line goal skew the xg? I'm thinking that will have a very small xg so may have a statistically significant bearing on the total.[/quote] Barnsley goal doesn't skew the data. It was around a 6% chance (the keeper was awol). Take it entirely out of the dataset and Clucas is still an (unsustainably) hot finisher. Just a 8% chance an average player scores 9 or more goals from these revised figures.
|
|
|
Post by tachyon on Mar 12, 2020 9:09:18 GMT
I would suggest his age helps him as he now shows more composure re goals scoring opportunities than when he was younger
[/quote][/b] A reasonable theory, but it's not borne out by the data. If you include age as a parameter to an xG model in large datasets, it isn't a statistically significant factor. Player's don't become more (or less) clinical with age. They do show an age related improvement and decline in the amount of xG per 90 they generate. Physicality combines with experience to peak a player's xG/90 production around 27/28, then the latter declines and takes a player's xg/90 down with it....but conversion rates aren't player or age dependent.
|
|
|
Post by tachyon on Mar 12, 2020 9:13:51 GMT
Expected goals has a value in judging how well a team is performing but it’s less impactful as a guide to individual players.
Clucas is unusually good at shooting from range. Even when he doesn’t score he tends to at least work the keeper.
It’s easy to view one goal as a fluke but when a player consistently outperforms others in terms of shooting from range there comes a point when you have to just accept that he’s really good at that particular skill.
I'd disagree. XG does a very good job of projecting a player's future output. I took the 25 biggest overperformers from the 2014/15 season when it came to scoring with low probability chances. (<6% chance). They scored a combined 83 actual goals from a combined xG of only 33. Massive over-performance, but was it sustainable? Nope. I then rolled on to the subsequent season and the same players had a combined xG of 35 expected goals from low quality chances and scored exactly 35 goals. Overperformance in one season or one part of a season isn't sustainable going forward. Clucas himself had 36 attempts with an xG of <6% in Hull's 2015/16 season and didn't score once.
|
|
|
Post by walrus on Mar 12, 2020 9:20:12 GMT
Just a 8% chance an average player scores 9 or more goals from these revised figures. I like these expected goal metrics but I’m intrigued as to how an “average player” is calculated? Also, on empirical evidence alone, I’d put Clucas and Campbell as the best finishers in our squad, which must put them roughly in the top 10% of professional footballers. Why should we judge how likely they are to sustain their goalscoring form based on the “average player”? Labouring the point a little, Lionel Messi has scored 37, 34, and 36 league goals in the last three seasons. Would you say his goalscoring is unsustainable based on the average player too?
|
|
|
Post by walrus on Mar 12, 2020 9:23:18 GMT
Expected goals has a value in judging how well a team is performing but it’s less impactful as a guide to individual players.
Clucas is unusually good at shooting from range. Even when he doesn’t score he tends to at least work the keeper.
It’s easy to view one goal as a fluke but when a player consistently outperforms others in terms of shooting from range there comes a point when you have to just accept that he’s really good at that particular skill.
I'd disagree. XG does a very good job of projecting a player's future output. I took the 25 biggest overperformers from the 2014/15 season when it came to scoring with low probability chances. (<6% chance). They scored a combined 83 actual goals from a combined xG of only 33. Massive over-performance, but was it sustainable? Nope. I then rolled on to the subsequent season and the same players had a combined xG of 35 expected goals from low quality chances and scored exactly 35 goals. Overperformance in one season or one part of a season isn't sustainable going forward. Clucas himself had 36 attempts with an xG of <6% in Hull's 2015/16 season and didn't score once. I think what I’m struggling with is the implication that over time all players trend towards to average. See my above post re. Messi for elaboration. Aren’t some players just better than others?
|
|
|
Post by tachyon on Mar 12, 2020 9:45:12 GMT
Just a 8% chance an average player scores 9 or more goals from these revised figures. I like these expected goal metrics but I’m intrigued as to how an “average player” is calculated? Also, on empirical evidence alone, I’d put Clucas and Campbell as the best finishers in our squad, which must put them roughly in the top 10% of professional footballers. Why should we judge how likely they are to sustain their goalscoring form based on the “average player”? Labouring the point a little, Lionel Messi has scored 37, 34, and 36 league goals in the last three seasons. Would you say his goalscoring is unsustainable based on the average player too? It's based on large datasets of historical evidence from the competition in question. Uses mainly location, shot type and phase of play as the major predictors, so as not to overfit the past. Player identity is not a statistically significant predictor. The skill is in being able to get into scoring positions. It is this that differentiates strikers from non strikers. Hence players who are good at this can maintain or, within the confines of the general aging curve, improve or suffer realtive decline in the likely number of goals they score. Once they are in position to take an opportunity, the skill differential in getting the ball in the net is minute between players. In short, getting into good scoring position >>>>>> compared to then putting the ball in the net. Hence any over or under performance in actual gols scored compared to a player's xG is almost always attributed to statistical noise, rather than finishing prowess. Teams who buy players on their actual, rather xG record are anachronistic dinosaurs. Messi is the only exception.
|
|
|
Post by tachyon on Mar 12, 2020 9:53:06 GMT
I think what I’m struggling with is the implication that over time all players trend towards to average. See my above post re. Messi for elaboration. Aren’t some players just better than others? [/quote] Getting into position to score ( BIG SKILL DIFFERENTIAL) Scoring once in position to score ( NO SKILL DIFFERENTIAL) Average player is poor choice of words, but unfortunately it has stuck.
|
|
|
Post by walrus on Mar 12, 2020 10:48:17 GMT
I think what I’m struggling with is the implication that over time all players trend towards to average. See my above post re. Messi for elaboration. Aren’t some players just better than others? Getting into position to score ( BIG SKILL DIFFERENTIAL) Scoring once in position to score ( NO SKILL DIFFERENTIAL) Average player is poor choice of words, but unfortunately it has stuck. [/quote] Thanks for taking the time to explain. It makes more sense in that context.
|
|
|
Post by tachyon on Mar 12, 2020 10:53:55 GMT
I think what I’m struggling with is the implication that over time all players trend towards to average. See my above post re. Messi for elaboration. Aren’t some players just better than others? Getting into position to score ( BIG SKILL DIFFERENTIAL) Scoring once in position to score ( NO SKILL DIFFERENTIAL) Average player is poor choice of words, but unfortunately it has stuck. Thanks for taking the time to explain. It makes more sense in that context. [/quote] No problem. It's counter intuitive & counter to most media/manager/supporter narratives Some clubs operate fully by xG, many don't. Taking it on board is the biggest free gift they'll ever get.
|
|
|
Post by liathroid on Mar 12, 2020 13:34:28 GMT
|
|
|
Post by walrus on Mar 12, 2020 15:36:05 GMT
Teams who buy players on their actual, rather xG record are anachronistic dinosaurs. One last question. Wouldn’t this approach lead to teams overvaluing players like Diouf who is widely held to get in good positions but be a poor finisher? Is this why Leicester reportedly bid £15m for him a few years back?
|
|
|
Post by markby on Mar 12, 2020 18:13:54 GMT
Apologies if I'm dragging this thread off topic, but can this be broken down into different types of finish? I only ask, because NI had a striker called David Healy, a journeyman as a club player, but outstanding when it came to playing for his country eg 35 goals in a 70 cap run for an otherwise poor team. He had one particular skill at which he was especially adept, one-on-ones with the keeper. He rarely hit it first time and he NEVER tried to go round the keeper (no pace). Instead he invariably took it up to him, then stroked it home before the keeper had time to react to where Healy was placing it. Sometimes he put it to the keeper's right, sometimes his left, or often through his legs (nutmeg) as the keeper "made himself big". I think what makes this skill distinctive is that with most goals, the hard part is getting away from defenders and into a goalscoring position; thereafter it's usually a straightforward matter of passing the ball into the net. But with a one-on-one, the striker still has decisions to make (early/up to/round), which causes some players to freeze ("Theo Walcott Syndrome" as it's sometimes known), esp if the keeper doesn't commit himself early and effectively make the striker's decision for him (Peter Schmeichel was especially good at this). Or is this just my imagination/faulty memory?
Remember Healy well, Leeds? 1 on 1 tend to be high quality chances, so a relatively high proportion will be converted. I don't have Healy data, but shrinking your sample size to account for specific types of chances will inevitably lead you to find players who score more than you'd expect. But the act of reducing the sample size makes it more difficult to discount simple random variation, (rather than a skill differential) as being the cause of this over-performance. Thanks for that and yes, you're quite right about sample size etc.
But as a fan, I only remember Healy because with the type of game he/we played, he often found himself with one-on-one finishes. And when he got through, you could put your mortgage not just on his converting, but on exactly how he was going to do it.
Thing is, at least some of the opposition gk's must have figured it out from watching videos etc (he was our only danger man up front), yet he was so proficient in his technique that he still deadly.
|
|
|
Post by markby on Mar 12, 2020 18:24:13 GMT
The skill is in being able to get into scoring positions. It is this that differentiates strikers from non strikers. Hence players who are good at this can maintain or, within the confines of the general aging curve, improve or suffer realtive decline in the likely number of goals they score. Once they are in position to take an opportunity, the skill differential in getting the ball in the net is minute between players. In short, getting into good scoring position >>>>>> compared to then putting the ball in the net. I remember seeing an article about Gary Lineker where someone pointed out that at 31, he wasn't as quick as when he was 25, say, the implication being that as he lost his pace, his game must suffer. He replied: "Yes, but when I was 25, I wasn't as quick as I'd been at 18, but I was still a much better player."
The point being that he more than compensated for his declining pace by his improved knowledge and understanding of the game. In his case, knowing how to find space in the box and evade defenders when the chance came etc was what made him special.
|
|
|
Post by tachyon on Mar 12, 2020 19:07:49 GMT
Teams who buy players on their actual, rather xG record are anachronistic dinosaurs. One last question. Wouldn’t this approach lead to teams overvaluing players like Diouf who is widely held to get in good positions but be a poor finisher? Is this why Leicester reportedly bid £15m for him a few years back? We got Diouf at the start of the 2014/15 season. His average xG per 90 was around 0.26 over the four Premier League seasons. So he got on the end of a cumulative one expected goal every 3 and 3 quarters of a game. He was around his peak age when he joined us & his xG/90 was 0.4 during that season. He was just over 30 when we started in the Championship & Leicester made their final approach. By that time his xG per 90 had fallen to 0.23. His actual goal tally was on par with his xG for the first two seasons. He marginally under performed from 2016-18. Overall, his conversion rate compared to xG was as good or better than 33% of simulations. That's an under performance, but it's was almost certainly just noise. The quoted price tag was £15 million for a player outside his peak years, who was taking 0.23 xg per 90 in his previous season. For comparison Vardy's xG was 0.32 non penalty xG/90 that season and he was valued at around £20 million. Maybe Stoke used that as a yardstick, but MD was a declining asset, with little resale value in a season or two. £7-8 mill tops in my book, asking for £15 mill was way overpriced.
|
|
|
Post by tachyon on Mar 12, 2020 19:11:16 GMT
The skill is in being able to get into scoring positions. It is this that differentiates strikers from non strikers. Hence players who are good at this can maintain or, within the confines of the general aging curve, improve or suffer realtive decline in the likely number of goals they score. Once they are in position to take an opportunity, the skill differential in getting the ball in the net is minute between players. In short, getting into good scoring position >>>>>> compared to then putting the ball in the net. I remember seeing an article about Gary Lineker where someone pointed out that at 31, he wasn't as quick as when he was 25, say, the implication being that as he lost his pace, his game must suffer. He replied: "Yes, but when I was 25, I wasn't as quick as I'd been at 18, but I was still a much better player." The point being that he more than compensated for his declining pace by his improved knowledge and understanding of the game. In his case, knowing how to find space in the box and evade defenders when the chance came etc was what made him special.
Good point. Experience compensates for declining pace, but eventually experience flattens and pace continues to decline, taking xG/90 down with it.
|
|
|
Post by Veritas on Mar 12, 2020 19:43:45 GMT
I remember seeing an article about Gary Lineker where someone pointed out that at 31, he wasn't as quick as when he was 25, say, the implication being that as he lost his pace, his game must suffer. He replied: "Yes, but when I was 25, I wasn't as quick as I'd been at 18, but I was still a much better player." The point being that he more than compensated for his declining pace by his improved knowledge and understanding of the game. In his case, knowing how to find space in the box and evade defenders when the chance came etc was what made him special.
Good point. Experience compensates for declining pace, but eventually experience flattens and pace continues to decline, taking xG/90 down with it. Michael Owen and Des Walker are very good examples at both ends of the pitch of players whose game was excessively dependant on pace and saw massive decline in effectiveness when they lost it
|
|
|
Post by mikeoxlong on Mar 12, 2020 20:02:18 GMT
I am not really interested in stats, because they can be used to suit whatever you want them to. For instance, does this include the period at Hull he was playing at left back, yet still scored goals. At Swansea he was constantly played out of position. Last year he played while still recovering from injury with no pre-season. This season has played in a variety of positions on the left, including 10? games with a broken foot. Includes like for like. Championship games as a midfielder. 10 games with a broken foot?? Seems to have escaped the attention of the mainstream sports media. His injury was certainly no secret, did XG take it into account?
|
|
|
Post by Championship Potter on Mar 12, 2020 20:44:02 GMT
Good point. Experience compensates for declining pace, but eventually experience flattens and pace continues to decline, taking xG/90 down with it. Michael Owen and Des Walker are very good examples at both ends of the pitch of players whose game was excessively dependant on pace and saw massive decline in effectiveness when they lost it Owen is an interesting one - lost most of his pace with a hamstring injury at 19 and would presumably have smashed a load of records if he'd had that pace his whole career.
|
|
|
Post by tachyon on Mar 13, 2020 9:04:47 GMT
Michael Owen and Des Walker are very good examples at both ends of the pitch of players whose game was excessively dependant on pace and saw massive decline in effectiveness when they lost it Owen is an interesting one - lost most of his pace with a hamstring injury at 19 and would presumably have smashed a load of records if he'd had that pace his whole career. Ligament/tendon injuries are real production killers. Kane, pre all his ligaments pinging, non penalty xG per 90 average 0.55Games played post his ligaments pinging, non penalty xG per 90 average 0.32Peak age, with a huge red flag.
|
|
|
Post by Championship Potter on Mar 13, 2020 19:24:15 GMT
Owen is an interesting one - lost most of his pace with a hamstring injury at 19 and would presumably have smashed a load of records if he'd had that pace his whole career. Ligament/tendon injuries are real production killers. Kane, pre all his ligaments pinging, non penalty xG per 90 average 0.55Games played post his ligaments pinging, non penalty xG per 90 average 0.32Peak age, with a huge red flag. Wow didn’t realise Kane had dropped that much. That suggests he’s massively overrated in the current England pecking order.
|
|
|
Post by Miles Offside on Mar 14, 2020 10:52:35 GMT
Like every other player we've got, he's gone through some poor patches in the season, but taken overall Clucas must be a contender for Player of the Season (even if it ends now).
|
|
shimmer
Academy Starlet
Posts: 244
Location: Middle Earth
|
Post by shimmer on Mar 17, 2020 18:23:04 GMT
Proving a class act
|
|
|
Post by markby on Jul 6, 2020 16:56:20 GMT
Owen is an interesting one - lost most of his pace with a hamstring injury at 19 and would presumably have smashed a load of records if he'd had that pace his whole career. Ligament/tendon injuries are real production killers.Kane, pre all his ligaments pinging, non penalty xG per 90 average 0.55Games played post his ligaments pinging, non penalty xG per 90 average 0.32Peak age, with a huge red flag. Is that because it robs players of their pace/explosiveness?
If so, that might not inhibit Kane quite so much, because he has never had that much pace to lose in the first place.
I always think Kane is the closest English football has seen to Alan Shearer. And having suffered a bad ankle ligament injury in 1997 when he was 27, Shearer got injured again in 2000/01 (unsure of the nature), playing only 19 PL games (5 goals).
Yet he then went on to play 109 PL games in the next 3 seasons, scoring 62 goals when, of course, he was into his 30's.
What is Kane's sample size post-ligament damage?
For such is his Shearer-like professionalism that his xG decline to 0.32 might just prove to be a blip?
Time will tell.
|
|
|
Post by walrus on Jul 6, 2020 20:05:05 GMT
Ligament/tendon injuries are real production killers.Kane, pre all his ligaments pinging, non penalty xG per 90 average 0.55Games played post his ligaments pinging, non penalty xG per 90 average 0.32Peak age, with a huge red flag. Is that because it robs players of their pace/explosiveness? If so, that might not inhibit Kane quite so much, because he has never had that much pace to lose in the first place.
I always think Kane is the closest English football has seen to Alan Shearer. And having suffered a bad ankle ligament injury in 1997 when he was 27, Shearer got injured again in 2000/01 (unsure of the nature), playing only 19 PL games (5 goals).
Yet he then went on to play 109 PL games in the next 3 seasons, scoring 62 goals when, of course, he was into his 30's. What is Kane's sample size post-ligament damage? For such is his Shearer-like professionalism that his xG decline to 0.32 might just prove to be a blip? Time will tell.
No doubt it also somewhat depends on how good the service he’s had from teammates has been.
|
|
|
Post by markby on Aug 3, 2020 12:03:02 GMT
Owen is an interesting one - lost most of his pace with a hamstring injury at 19 and would presumably have smashed a load of records if he'd had that pace his whole career. Ligament/tendon injuries are real production killers. Kane, pre all his ligaments pinging, non penalty xG per 90 average 0.55Games played post his ligaments pinging, non penalty xG per 90 average 0.32Peak age, with a huge red flag. I don't know his xG stats but I note that HK eventually scored 34 goals in 40 games ths season (Spurs and England).
This was made up of 27 in 34 games before getting injured on 31st Dec. He came back after lockdown and scored 7 in 9 PL games, none of them penalties. More to the point, 5 of those goals came in the final 3 games i.e. after he'd shaken the rust off.
I know it's hardly scientific(!), but I watched those games and the way he finished the season, he looked to me to be as good as ever.
|
|
|
Post by Goonie on Aug 3, 2020 19:25:25 GMT
Pick those ones out!
|
|
|
Post by tachyon on Aug 4, 2020 7:13:26 GMT
I don't know his xG stats but I note that HK eventually scored 34 goals in 40 games ths season (Spurs and England).[/div] This was made up of 27 in 34 games before getting injured on 31st Dec. He came back after lockdown and scored 7 in 9 PL games, none of them penalties. More to the point, 5 of those goals came in the final 3 games i.e. after he'd shaken the rust off. I know it's hardly scientific(!), but I watched those games and the way he finished the season, he looked to me to be as good as ever. [/quote] This season Kane's non penalty xG per 90 is 30% down on his average for the previous 5 Premier League seasons. His shot volume per 90 is also down by 33%. Since lockdown his NPxG/90 has crept back up to his rolling average from 2014/15 to 2018/19, so you eyes aren't fooling you. But a couple of caveats. It's just 9 games (in unique circumstances) and his shot voume is still at the same lower levels, which would lead you to suspect he's been shooting from smarter locations post lockdown. Statistically, he's not the same player compared to his shooting/xG monster season of especially 2017/18.
|
|
|
Post by ethers26 on Aug 4, 2020 7:33:02 GMT
Regardless of this being a Clucas thread, he is different gravy.
|
|
|
Post by markby on Aug 4, 2020 13:19:59 GMT
This season Kane's non penalty xG per 90 is 30% down on his average for the previous 5 Premier League seasons. His shot volume per 90 is also down by 33%. Since lockdown his NPxG/90 has crept back up to his rolling average from 2014/15 to 2018/19, so you eyes aren't fooling you. But a couple of caveats. It's just 9 games (in unique circumstances) and his shot voume is still at the same lower levels, which would lead you to suspect he's been shooting from smarter locations post lockdown. Statistically, he's not the same player compared to his shooting/xG monster season of especially 2017/18.
Wouldn't argue with that analysis (a recent dip), but I still predict that it is temporary and he can get back to his (statistical) best, if not better.
For I think the "unique circumstances" actually back my theory, rather than contradict it, as follows.
First, that 5 month break (injury and lockdown) allowed HK to get properly fit for the first time in four or five years. Previously he was worked like a dog, both Spurs and England, getting only the briefest of breaks each summer when you count Finals tournaments and pre-season club tours to Asia etc;
Second, he is now benefiting from Mourinho's coaching. Now I'm most definitely NOT a Mourinho fan, and if he doesn't like you it doesn't matter how talented you are, you're out (eg Zlatan, De Bruyne, Salah etc). The flip side is that if he does like you - and he clearly likes HK - then he can definitely improve you - see eg the "shooting from smarter positions";
Third, as well as being dedicated, HK is also very intelligent. I suspect that he will adjust his game to cope with his changing physicality, just as eg Shearer did after injury, and eg Owen didn't.
I remember saying to someone 4 or 5 years ago that HK would be the first £100m footballer. I've now revised that to the first £200m footballer (Covid permitting)!
And yeehaw - Bairstow has just gone, England 14 for 2, c'mon Ireland!
|
|
|
Post by Goonie on Aug 4, 2020 17:54:13 GMT
Regardless of this being a Clucas thread, he is different gravy. The quality of his 10 is evident. Him and Campbell are the jewels in the crown
|
|