-
The Fighting Cock is a forum for fans of Tottenham Hotspur Football Club. Here you can discuss Spurs latest matches, our squad, tactics and any transfer news surrounding the club. Registration gives you access to all our forums (including 'Off Topic' discussion) and removes most of the adverts (you can remove them all via an account upgrade). You're here now, you might as well...
Tactics TFC's Tactical Autopsy Thread
- Thread starter bus-conductor
- Start date
Latest Spurs videos from Sky Sports
So just how inconsistent are we this season? Is it just confirmation bias (because actually everyone is consistent this year and no games are predictable any more)?
To put it to the test, I modelled teams' performance over the last 2 seasons in terms of creation of xG, prevention of xG against, and shot conversion. Then if we look match-by-match at this season's xG results compared to model, we can get some objective data for consistency, as well as level of improvement vs last season. So here's Tottenham's performance game-by-game, taking into account the quality (per model) of the opposition:
* before international break
Note that this is looking at xG, not actual scores, so it does not reflect shooting or goalkeeping performance. It's just looking at the underlying metric of chances created and chances conceded, which I think that is a better measure of performance in individual games. So looking at say the Everton game, model would predict an xG of 2.11-1.23. We did slightly better than that, creating 2.4 xG and conceding 1.0. Then some excellent shooting by us, and some poor shooting / good goalkeeping at the other end, resulted in a 4-0 scoreline. But for this analysis I'm ignoring the 4-0 and concentrating on the 2.4-1.0.
So looking at overall xGD performance (ie xG of chances created, minus xG of goals conceded) compared to model, the first 10 Spurs games (up to and including Villa) are reasonably consistent (within +/- 1 goal of model xGD), except for Brentford and United which are in the positive direction and an indication that we were really starting to click at that stage. The Brighton game looks reasonable if you look at the match overall, although obviously if you split the second half from the first it would tell a different story. Even the Palace game doesn't look too bad, which suggest that the model thinks Palace are a better team than what their results up until then had suggested (and results since then have tended to bear that out).
Also in the first 10 rounds our average xGD was 0.55 goals per game better than model predicted, indicating a significant improvement relative to last season.
So how does our overall consistency compare to other teams? To look at this, I have taken the standard deviation (ie variability) of the difference between model and actual xGD, and then divided this by the average total xG per game (xG and xGA):
So after round 10 we were 3rd-highest in terms of variability (with City, Wolves, Fulham and Newcastle close behind), and most improved in terms of average xGD. Man Utd were actually not doing too bad in terms of xGD at that stage, but their shooting had been terrible so it hadn't translated into results, and their defensive variability had been high.
Now if we wind the clock forwards 4 more rounds, we've had 3 very poor performances and one very good:
So now the comparison to other teams looks like this:
Up until the last 4 games we were looking significantly improved but fairly inconsistent (but not the most inconsistent). Now suddenly we look no better than last season, and are the most inconsistent team by some distance. Is it the European match load catching up with us? Injuries? Discontent? A bit of everything?
Regarding the table more generally, some observations:
I also split out the variability of xG creation vs xGA prevention over the 14 rounds:
To put it to the test, I modelled teams' performance over the last 2 seasons in terms of creation of xG, prevention of xG against, and shot conversion. Then if we look match-by-match at this season's xG results compared to model, we can get some objective data for consistency, as well as level of improvement vs last season. So here's Tottenham's performance game-by-game, taking into account the quality (per model) of the opposition:
| Match | xG | Model xG | xGA | Model xGA | xGD | Model xGD | Creation Perf | Prevention Perf | Overall xGD Perf |
| Leicester City (A) | 1.2 | 1.74 | 1 | 0.71 | 0.2 | 1.03 | -0.54 | -0.29 | -0.83 |
| Everton (H) | 2.4 | 2.11 | 1 | 1.23 | 1.4 | 0.88 | 0.29 | 0.23 | 0.52 |
| Newcastle Utd (A)* | 1.2 | 1.45 | 1.8 | 2.33 | -0.6 | -0.88 | -0.25 | 0.53 | 0.28 |
| Woolwich (H) | 0.7 | 1.02 | 0.7 | 1.72 | 0 | -0.70 | -0.32 | 1.02 | 0.70 |
| Brentford (H) | 3.5 | 1.80 | 0.8 | 1.40 | 2.7 | 0.39 | 1.70 | 0.60 | 2.31 |
| Manchester Utd (A) | 4.4 | 2.02 | 1 | 1.73 | 3.4 | 0.29 | 2.38 | 0.73 | 3.11 |
| Brighton (A)* | 1.3 | 1.59 | 1.8 | 1.60 | -0.5 | -0.01 | -0.29 | -0.20 | -0.49 |
| West Ham (H) | 1.9 | 2.45 | 0.8 | 1.40 | 1.1 | 1.05 | -0.55 | 0.60 | 0.05 |
| Crystal Palace (A) | 0.7 | 1.31 | 0.8 | 1.37 | -0.1 | -0.06 | -0.61 | 0.57 | -0.04 |
| Aston Villa (H) | 2.4 | 2.19 | 1.8 | 1.44 | 0.6 | 0.75 | 0.21 | -0.36 | -0.15 |
Note that this is looking at xG, not actual scores, so it does not reflect shooting or goalkeeping performance. It's just looking at the underlying metric of chances created and chances conceded, which I think that is a better measure of performance in individual games. So looking at say the Everton game, model would predict an xG of 2.11-1.23. We did slightly better than that, creating 2.4 xG and conceding 1.0. Then some excellent shooting by us, and some poor shooting / good goalkeeping at the other end, resulted in a 4-0 scoreline. But for this analysis I'm ignoring the 4-0 and concentrating on the 2.4-1.0.
So looking at overall xGD performance (ie xG of chances created, minus xG of goals conceded) compared to model, the first 10 Spurs games (up to and including Villa) are reasonably consistent (within +/- 1 goal of model xGD), except for Brentford and United which are in the positive direction and an indication that we were really starting to click at that stage. The Brighton game looks reasonable if you look at the match overall, although obviously if you split the second half from the first it would tell a different story. Even the Palace game doesn't look too bad, which suggest that the model thinks Palace are a better team than what their results up until then had suggested (and results since then have tended to bear that out).
Also in the first 10 rounds our average xGD was 0.55 goals per game better than model predicted, indicating a significant improvement relative to last season.
So how does our overall consistency compare to other teams? To look at this, I have taken the standard deviation (ie variability) of the difference between model and actual xGD, and then divided this by the average total xG per game (xG and xGA):
So after round 10 we were 3rd-highest in terms of variability (with City, Wolves, Fulham and Newcastle close behind), and most improved in terms of average xGD. Man Utd were actually not doing too bad in terms of xGD at that stage, but their shooting had been terrible so it hadn't translated into results, and their defensive variability had been high.
Now if we wind the clock forwards 4 more rounds, we've had 3 very poor performances and one very good:
| Match | xG | pxG | xGA | pxGA | xGD | pxGD | Creation Perf | Prevention Perf | Overall xGD Perf |
| Ipswich Town (H)* | 1.5 | 3.75 | 1.6 | 0.80 | -0.1 | 2.95 | -2.25 | -0.80 | -3.05 |
| Manchester City (A) | 2.5 | 0.93 | 2.1 | 2.14 | 0.4 | -1.21 | 1.57 | 0.04 | 1.61 |
| Fulham (H) | 0.8 | 2.12 | 1.5 | 1.31 | -0.7 | 0.81 | -1.32 | -0.19 | -1.51 |
| Bournemouth (A) | 0.9 | 1.39 | 3.5 | 1.54 | -2.6 | -0.15 | -0.49 | -1.96 | -2.45 |
So now the comparison to other teams looks like this:
Up until the last 4 games we were looking significantly improved but fairly inconsistent (but not the most inconsistent). Now suddenly we look no better than last season, and are the most inconsistent team by some distance. Is it the European match load catching up with us? Injuries? Discontent? A bit of everything?
Regarding the table more generally, some observations:
- Liverpool haven't actually improved. They've just been reasonably consistent at about the same level as last season.
- City and Woolwich have dropped off significantly, and with Chelsea improving it looks like a 3-way race for 2nd/3rd/4th.
- Significant improvements from Fulham and Bournemouth (and Spurs up until last couple of games) is adding to the log-jam for 5th/6th/7th.
- Bournemouth most improved so far this season in terms of xGD, although their results to date have been impacted by poor conversion of xG to goals (underperforming by 0.44 goals per game, missing someone maybe?).
- Villa have improved their underlying stats but this hasn't translated to results because they have conceded significantly more goals than their xGA (poor goal keeping?).
I also split out the variability of xG creation vs xGA prevention over the 14 rounds:
- Most consistent in terms of xG creation: Everton 1 (consistently low!), Man City 2, Wolves 3. Least consistent: Tottenham 1, West Ham 2, Brentford 3.
- Most consistent in terms of xGA prevention: Bournemouth 1, Liverpool 2, Fulham 3. Least consistent: City 1, Woolwich (I see red) 2, United 3. (Tottenham 5th least consistent).
Last edited:
Really great work and very interesting.
No coincidence that, just like last season, we start losing a bunch of players - some key, some not, and consistency drops. We just do not have the squad to cope yet, let alone do two games a week. Chelsea effectively have two first teams, I believe they haven’t even put players like Palmer in their Euro squad.
Added to that some of the selections and in game management of Ange lately - along with the omnipresent inherent risk/flaws of Angeball.
We’ve been top 3-4 for xpts most of the season until the last couple of weeks which has seen that go well south, now we are about 10th. Not good.
It's certainly a challenge. If we can come out the other side with most of the squad, staff and fanbase intact then we will all be stronger for it. Maybe it's the only way to effect real change in the mentality.
The stats don't show why anyone does anything. But it does show whether they do it. If they are skying all their shots then I would expect their post-shot xG (PSxG) to be below their xG. (And if we are blocking their good shots well then their goals (G) will be less than their PSxG).
So far this season our opponents have had 20.1 xG, 19.3 PSxG and 15 Goals. So our opponents' shooting has been slightly worse than average (19.3 vs 20.1, hardly significant). Our shot stopping (by keeper or cleared off the line etc) has been excellent and prevented ~4.3 goals compared to "average" keeping / blocking.
If you look at individual matches, we benefited from poor shooting by Bournemouth (3.5 xG but only 1.8 PSxG), Man City (2.1 xG but only 1.1 PSxG) and Villa (1.8 xG but only 1.0 PSxG). Since Bournemouth was our most recent game, "recency bias" will lead you to think that feel that it's happening more often that it is.
Ipswich actually shot well against us (2.4 PSxG from 1.6 xG) as did Palace (1.5 PSxG from 0.8 xG) for example. I can't be bother going back further than that but you can see the pattern: sometimes our opponents shoot well and sometimes not, and it tends to average out over the season. There's no real surprise there.
The biggest beneficiaries of poor shooting so far have been Brighton (22 xGA but only 19.2 PSxG). Bournemouth have had the worst luck so far in encountering teams that had their "shooting boots" on (16.7 xGA but 23.4 PSxG).
Absolute garbage performance but somehow we’ve had far better open play chances than them and but for two stupid pens might have got more than we deserved:
View: https://x.com/dataanalyticepl/status/1865826259449573615?s=46&t=IomICbZIqZC4FsH907qgfw
View: https://x.com/dataanalyticepl/status/1865827149426381205?s=46&t=IomICbZIqZC4FsH907qgfw
View: https://x.com/markstatsbot/status/1865829740721897608?s=46&t=IomICbZIqZC4FsH907qgfw
View: https://x.com/markstatsbot/status/1865829806367015371?s=46&t=IomICbZIqZC4FsH907qgfw
View: https://x.com/dataanalyticepl/status/1865826259449573615?s=46&t=IomICbZIqZC4FsH907qgfw
View: https://x.com/dataanalyticepl/status/1865827149426381205?s=46&t=IomICbZIqZC4FsH907qgfw
View: https://x.com/markstatsbot/status/1865829740721897608?s=46&t=IomICbZIqZC4FsH907qgfw
View: https://x.com/markstatsbot/status/1865829806367015371?s=46&t=IomICbZIqZC4FsH907qgfw
A bit painful updating these after a couple of bad results but that's how I cope...
Goals scored:
| Opp | Turnover | Turnover height | Passes | Scorer | Assist | Defenders in Box @T-3s | Defenders in Box | Category |
| Leicester | Vic | 15 | 12 | Porro | Maddison | 3 | 5 | Build |
| Everton | Throw-in | 80 | 21 | Bissouma | Kulusevski | 7 | 9 | Build |
| Everton | Son (Pickford) | 90 | 0 | Son | 0 | 0 | High TO | |
| Everton | Corner | 100 | 1 | Romero | Maddison | 10 | 10 | Set Piece |
| Everton | van de Ven | 20 | 1 | Son | van de Ven | 0 | 4 | Break |
| Newcastle | Bissouma | 70 | 3 | OG | Johnson | 5 | 7 | High TO |
| Brentford | Maddison | 75 | 0 | Solanke | Maddison | 1 | 4 | High TO |
| Brentford | Maddison | 40 | 2 | Johnson | Son | 1 | 4 | Break |
| Brentford | Bissouma | 20 | 3 | Maddison | Son | 0 | 3 | Break |
| Man United | van de Ven | 45 | 1 | Johnson | van de Ven | 0 | 5 | Break |
| Man United | Romero | 15 | 3 | Kulusevski | Johnson | 0 | 4 | Break |
| Man United | Corner | 100 | 2 | Solanke | Sarr | 9 | 9 | Set Piece |
| Brighton | Maddison | 65 | 2 | Johnson | Solanke | 0 | 2 | High TO |
| Brighton | Goal kick | 0 | 6 | Maddison | Werner | 2 | 5 | Build |
| West Ham | Solanke | 40 | 3 | Kulu | Maddison | 0 | 6 | Break |
| West Ham | Udogie | 45 | 13 | Bissouma | Udogie | 5 | 6 | Build |
| West Ham | Bissouma | 35 | 3 | OG | Son | 1 | 6 | Break |
| West Ham | Sarr | 30 | 1 | Son | Sarr | 0 | 4 | Break |
| Aston Villa | (Drop Ball) | 45 | 7 | Johnson | Son | 3 | 6 | Build |
| Aston Villa | Davies | 55 | 5 | Solanke | Kulusevski | 0 | 3 | High TO |
| Aston Villa | Sarr | 55 | 2 | Solanke | Richarlison | 0 | 5 | High TO |
| Aston Villa | Free Kick | 80 | 0 | Maddison | (Solanke) | 10 | 10 | Set Piece |
| Ipswich | Corner | 100 | 0 | Bentancur | Porro | 10 | 10 | Set Piece |
| Man City | Bissouma | 50 | 10 | Maddison | Kulusevski | 2 | 3 | Build |
| Man City | Maddison | 75 | 2 | Maddison | Son | 4 | 6 | High TO |
| Man City | Kulusevski | 25 | 4 | Porro | Solanke | 4 | 5 | Build |
| Man City | Werner | 50 | 1 | Johnson | Werner | 0 | 5 | Break |
| Fulham | Bissouma | 15 | 7 | Johnson | Werner | 3 | 7 | Break |
| Chelsea | Johnson (Cucurella) | 65 | 1 | Solanke | Johnson | 0 | 4 | High TO |
| Chelsea | Johnson (Cucurella) | 80 | 2 | Kulusevski | Porro | 2 | 5 | High TO |
| Chelsea | Corner | 100 | 3 | Maddison | Son | 10 | 10 | Set Piece |
9 from high turnovers, 10 on the break, 7 via build-up form the back, 5 from set pieces = 31
Goals Conceded:
| Opp | Turned Over by | Turnover Height (Opp) | Passes | Scorer | Defenders in Box @T-3s | Defenders in Box | Category |
| Leicester | Johnson | 25 | 3 | Vardy | 5 | 8 | Box Defence |
| Newcastle | (Throw in) | 75 | 2 | Barnes | 2 | 5 | Transitional Defence |
| Newcastle | Maddison | 25 | 2 | Isak | 0 | 3 | Break / High Line |
| Woolwich | Corner | 100 | 1 | Gabriel | 10 | 10 | Set Piece |
| Brentford | Kulu | 30 | 3 | Mbeumo | 4 | 4 | Transitional Defence |
| Brighton | Porro | 75 | 9 | Minteh | 3 | 5 | Transitional Defence |
| Brighton | Keeper throw | 10 | 7 | Rutter | 0 | 3 | Transitional Defence |
| Brighton | Throw in | 90 | 2 | Welbeck | 5 | 5 | Box Defence |
| West Ham | Keeper throw | 10 | 11 | Kudus | 5 | 6 | Box Defence |
| Palace | Van de Ven | 90 | 2 | Mateta | 2 | 6 | High TO |
| Villa | Corner | 100 | 1 | Rogers | 10 | 10 | Set Piece |
| Ipswich | Son | 30 | 12 | Szmodics | 8 | 8 | Box Defence |
| Ipswich | Bentancur | 45 | 3 | Delap | 3 | 7 | Transitional Defence |
| Fulham | Davies | 75 | 10 | Cairney | 8 | 8 | Box Defence |
| Bournemouth | Corner | 100 | 1 | Huijsen | 10 | 10 | Set Piece |
| Chelsea | Kulusevski | 70 | 9 | Sancho | 0 | 4 | Transitional Defence |
| Chelsea | Bissouma (Pen) | 90 | 0 | Palmer | 0 | 0 | Penalty |
| Chelsea | Throw in | 25 | 9 | Enzo | 7 | 7 | Box Defence |
| Chelsea | Sarr (Pen) | 90 | 0 | Palmer | 0 | 0 | Penalty |
6 from poor Box defence, 6 from what I'm calling "Transitional Defence"*, 3 from set pieces, 2 from penalties, 1 from a high turnover and 1 where our high line was broken = 19.
* I'm call it "Transitional Defence" when we have players behind the ball but they are back-pedalling, typically not out-numbered but maybe a 5-V-5 type scenario for example.
Last edited:
That box defending is a killer.