Tag Archives: soccer stats

Crunch game weekend – Statty previews for the Premier League gameweek 04/02/2017-05/02/2017

This is a week of crunch games. At the very least, 6 of the 10 Premier League games are of potentially vital importance, with the other 4 giving potential for wild swings in mid-table.

Three games – Hull vs Liverpool, Tottenham vs Middlesbrough, and Manchester City vs Swansea – involve one side tightly tied up in the top 4 race with the other more or less tightly involved in the relegation battle.

Hull have the home advantage, both over Liverpool and their relegation rivals, and although Marco Silva seems to have tightened things up a bit, they’ve still been conceding 5.7 shots on target a game under him. Through the season as a whole they’ve conceded 6. Continue reading

Advertisements

Defensive Correlations: The Return

Hello.

A couple of years ago I made some posts on correlations between defensive actions and how many shots were conceded while that central defender was playing. A helpful commenter aided me, but my stats know-how was very poor and I recently took the posts down to try and avoid someone stumbling on bad work in their search for solid stats.

However, I’ve since gone back to the data I collected (by hand, from the very excellent StatsZone app, in the days before I’d even heard of a data scraper, not that I know what to do with them now I’ve heard of them).

I went back to the data from the 2014/15 season, across the Premier League, Bundesliga, La Liga, and Ligue 1. I took the guys who’d played more than 900 minutes (141 of them) and did some correlations for defensive actions per 90 against shots conceded per 90. Continue reading

An introduction to clearance maps

When I was busily logging all of those defensive stats to try and do something with individual defensive statistics (which will come first, a breakthrough in that or the next Game of Thrones book, eh, eh, amirite?) I started to come to the conclusion that, while clearances weren’t a great measure of defensive ability, they were a decent measure of the pressure a team has been put under.

After I stopped doing CB stats and started getting stats cravings again, I turned to clearances. From observation, I think they can be useful.

Sweden England U21s

Largely, I think you can divide clearances into 2/3 groups. The first is basically just the penalty area, or Red Zone. The second is a kind of trapezium shaped second box, or Orange Zone. This is a trapezium instead of a straight rectangle because at each side of it is a fairly non-essential area that full-backs occupy. These clearances generally, in my mind, aren’t too important, and don’t happen too often anyway.

This clearance map is for the Sweden U21-England U21 match. Reading it, it looks to have been an even match, both have similar numbers of clearances in total, in the Red Zone, and in the Orange Zone. Perhaps England put Sweden under slightly more pressure, as there seem to be a higher concentration of clearances in Sweden’s Red Zone, and England’s clearances give the impression of a higher line than Sweden, which suggests that the balance of play was pushed into Sweden’s half.

I didn’t watch the match, nor look at possession or shot stats before writing the above. England did indeed outshoot Sweden 14-10, getting more shots in dangerous areas too. They also had 58% of possession, which slightly surprises me, given how similar the clearance map looks. Perhaps England largely had control of the match, but failed to make it count in really forcing the matter, pushing Sweden back and creating chances, choosing instead to keep the ball and pass it around perhaps somewhat aimlessly (I can say this, I’m English).

So basically, that’s how you read a clearance map. Despite trying to push them, I’m still a bit hesitant about them. For one thing, I’m not sure how much I am reading into them what I want to read into them. I may not have known much about the U21 game, but I did know from the score that it was probably a close game with England edging it, which is what I read. Red Zone clearances will also be affected by corners, which could, in some cases, skew how the clearance map looks. Clearance maps will also work less well with Continental teams than English club sides, as they clear the ball far less. For example, season long clearance maps for the 2014/15 season champions from the La Liga and the EPL:

Barca 14-15 just BarcaMan City 14-15 just City

Below the heatmaps are rough Red Zone, Orange Zone, and ‘5th line’ totals. The 5th line total, in the EPL, is a rough proxy for how high the defensive line is. Man City is actually a bad example of this, because it very unusually goes down in total from column 4 to 5, then back up again at 6, which might suggest a brand of ‘charge and hope’ CB defending. For reference, in the Premier League, RZ totals are on average 550, OZ totals vary a fair bit, but usually between 200-300, and 5th line total EPL average is around 110. City are obviously a fair way below all of these.

However, Barcelona are HUGELY below these EPL totals, particularly the Orange Zone and 5th line, which are virtually non-existent. So perhaps, for now, keep a bit of distance from La Liga clearance maps.

Thanks for reading. Below is some more clearance map stuff that is less public-aimed, more stats-y.

I haven’t done too much correlational stuff yet, as I need to collect averages of everything and put it all together, but I have done it for one season, the Premier League 14/15. Total clearances per game had a -0.59 correlation with point totals, and ‘Red Zone’ clearances per game a -0.67 correlation. This is probably similar to why the Red Zone totals for Barcelona were much more comparable with the Premier League clearance maps than Orange Zone and 5th line totals. It also makes some sense. No matter how much you might want to pass the ball out of the back, there will always come a point (in your own box) where you’ve just got to boot it out of play and away from danger.

What next? Do more data-y stuff, to see whether Red Zone clearances are a good link to points totals elsewhere. Look more at clearance maps for other leagues outside of the EPL, to see if/how they can be of worth. Try and decide what clearance maps are actually *for*.

Can they explain why things happen, why teams win or lose, and thus how they can improve? It seems unlikely. They look like they can be used to tell the story of a match, or a general look at a team’s season, but is there any call for that?

Anyway, as usual you can comment below or get me on Twitter @ETNAR_uk

Premier League predictions based on defensive performance (feat. tableau)

This is another piece that is more a vehicle for me to play around with and showcase tableau than a serious analytical or statistical thing. This time, I’ve plotted points per games of this season’s Premier League clubs against both their shots on target+shots blocked conceded, and just their shots on target conceded.

Team pts per game vs SoT+SB CIt’s a fairly simple graph, though might look a little crowded. I’ve no idea if shots on target+shots blocked conceded per90 is a good measure of a team, but I figure that if you’re giving away large numbers of shots on target (generally ‘good’ shots) and shots blocked (in my mind they signify some type of pressure) then you’re probably going to be losing matches. However, like we saw with Liverpool last season, you can give away a fair bit defensively but still do well as long as you blitz it offensively. For this, the main graph is the one on the left (SoT+SB), with the one on the right (SoT) as a kind of supplementary or comparison graph.

A few observations:

– Chelsea are phenomenally high up given how many SoT+SB they concede per match. They’re so far away from the average, they look like an outlier. Are they doing ‘a Liverpool’ (from last year)? Can their results up considering these shots conceded numbers?

– Southampton maybe deserve to be where they are. I say maybe because there are other better statistical methods of judging whether a team is ‘where the belong’ in the table, or whether their results are masking something deeper. They’re not too far away from the average in either graph, and within the confidence bands in both. It’s promising at least.

– West Ham will drop, Burnley will up their points per game. This might seem like an obvious thing to say because it seems like common sense, but these stats at least seem to back common sense up. West Ham are conceding shots on target+shots blocked that should get them around 1.3 but they’re getting 1.8, which over just six games is a difference of a full three points – a free win. Burnley, on the other hand, while they haven’t been great, their ppg of 0.44 seems very harsh, and on this graph they should be getting about 1 point per game instead.

– Newcastle (the pink dot between Palace and Stoke on the left graph) should pick up too. Using the same logic as for the other teams, they should have been getting nearer 1.5 ppg rather than the 1.11 that they’ve had so far. Michael Caley wrote an article a few days ago on this subject (http://www.sbnation.com/soccer/2014/10/28/7081825/newcastle-tottenham-2014-premier-league) essentially saying that their underlying stats say they’re better than their points total does. He also suggests that Newcastle has some underlying issue around scoring goals, which I agree with. It’d explain why they’re where they are on the graph, and also just agrees with what my eyes tell me – that they have a good team for who they are but that they ALWAYS look like they’re underperforming.

– Stoke are a huge outlier on just SoT but not SoT+SB. This probably just means that they’re throwing themselves in front of a lot of shots, but that they’re still giving away a fair number of opportunities – the SoT+SB graph has them roughly where they ‘should’ be.

Anyway, as I said there are more rigorous and accurate ways of looking at team performances, but defensive data is what I have available and I wanted an excuse to use tableau and write something. The link for this particular interactive tableau graph is here: https://public.tableausoftware.com/profile/mark6903#!/vizhome/Othercentre-backcomparisons/TeamptspergamevsSoTSBC and the interactive graph for all PL, Bundesliga, La Liga centre-backs judged on a couple of TICAD defensive scores is here: https://public.tableausoftware.com/profile/mark6903#!/vizhome/Somecentre-backsSpainGermanyandEngland21-10-14/PLLaLigaBundesligaCBsafterGW8

Thanks for reading, you can comment below or get me on Twitter @ETNAR_uk