Stats and projections for Spurs-Newcastle

Newcastle didn't just win home to Chelsea last weekend, they beat the Blues pretty comprehensively. With four more shots on target inside the box and one more from close/central positions, the Magpies had the better of the match statistically. The manner of their victory should be concerning for Spurs. Newcastle sat back and absorbed pressure, allowing little of real danger. But they weren't bunkered, they were actively playing on the counter. In the second half, Alan Pardew freed Yoan Gouffran, Loic Remy and Moussa Sissoko to blast forward when the opportunity came, and Yohan Cabaye looked to pick out the killer pass.

Tactically, that's basically exactly what I'd try to get my club to do against Tottenham Hotspur. So Newcastle are fully capable of doing what it takes to beat us. That said, Newcastle really have not been that strong a defensive side. They've allowed 30 shots on target inside the box and 15 big chances. (Spurs by contract have allowed 21 and 8). So the shutout of Chelsea, while impressive, was not fully in character. Newcastle have also been shockingly on-and-off all season, following as they have a loss to Sunderland with a win over Chelsea. So one can hope that the bad version of Newcastle shows up. If not, however, this could be a very difficult match.

As I did last week, I've run projections not just for Spurs-Newcastle but also for the two matches to follow. What do Spurs need to get out of November's matches in order to remain in Premier League contention? I've found that Spurs kind of need two wins out of the next three, or at least a win and two draws, to hold position. With fewer than five points out of nine, we won't fall out of top four contention, but we'll be back down in 50/50 Liverpool territory and nearly out of title contention entirely. Obviously we do not want to be going into the Manchester games looking for six points out of six, so getting three from Newcastle looks significant in this context.

Spurs Projections for November

Do remember that because of rounding, not all the numbers necessarily add up quite right.

• Expected Goals methodology
• Season Simulation methodology
• Points is the number of points, 0-9 with 8 excluded as impossible, that Spurs might take over the next three matches.
• %Chance is the odds of Spurs taking that many points from the next three matches in my simulator.
• %Top 4 is the odds of Spurs making top four after taking X points from the next three games, and %Title is the same but for the EPL title.

You see the dip at five points from nine because draws are generally more unlikely events than wins and losses. The top four% doesn't seriously spike or crater at any point, though being under 1-in-3 this early in the season would suck a bit. The title% demands five points, and really demands seven if Spurs are to remain in good position to make a run.

Odds for Spurs-Newcastle

I do also have the numbers for just this game. Despite Newcastle's reasonably strong attack, our defense still rates an impressively high chance at a clean sheet. My numbers love Spurs defense, and while they could take or leave the attack, that defensive quality is what keeps the projections running at a good rate. I have the most likely scorelines as 1-0 (13%), 2-0 (12%), 2-1 (10%), 1-1 (10%), and 0-0 (8%).

I hope we win.

Excursus on Attack and Defense

I had an interesting conversation in the comments recently about the relative values of attacking and defending. Conventional wisdom holds that while both are important, attacking is what you need to win titles while defending is what keeps you from getting relegated. In their quite good book, The Numbers Game, Chris Anderson and David Sally ran a series of regressions on data from the Premier League, 2001-2011. They found that while it is true that improvements in attacking quality results in slightly more wins than improvement in defensive quality, the difference is not large. In fact, the added value of prevented losses from improvements in defense, on average, slightly trump the value of added wins. It's close, and the gaps are too small to say that defense wins championships or something like that. But my take away is that both defense and attack win championships. You just have to be good enough, it doesn't matter much exactly how your goodness is divided. These are the numbers from their regressions:

+10 G = +2.3 Wins, -1.76 Losses
-10 GA = +2.16 Wins, -2.35 Losses

That is, adding ten goals scored resulted in, on average, an extra 2.3 wins and 1.76 fewer losses. So likewise for 10 more goals prevented. Either way of getting better is good. Preventing more goals is ever so slightly better.