The Upgrade

With preseason right around the corner, Timbers fans can finally start to look forward to some football to fill the time instead of (or more likely, as well as) internet drama and trademark disputes.

Much of the speculation among fans has centered on who will form the midfield, and who will lead the attack under our new head coach, Caleb Porter, but a more intriguing question may be who will be between the sticks.

The front office won few fans last year when they moved Troy Perkins to Montreal in exchange for Donovan Ricketts. It was, on the face of it, a strange decision and was compounded by the rather unfortunate framing of the move as an “upgrade”. While, clearly, no club would make such a big move without thinking they would benefit from it, the perceived slap in the face to a firm fans favourite and loyal servant rankled with some, and still does for many.

So before we look at who may be first choice in 2013, it may be worth looking back at the move and seeing if the Timbers really did call it right.

The Figures

I’m going to throw a lot of figures and terms at you, so if you’re not a fan of soccer analytics, it’s probably best you skip this post. If you want to hang around, then let me frame some of what follows.

The MLS site gives figures for goalkeepers, in terms of shots and saves, but I’ve always felt it was rather clumsy. In that system a 40 yard daisycutter carries the same weight as a point black volley, so I’ve gone back through the data and sorted it into In-Box chances and Long Range chance. I’ve also discarded own goals as these are largely outwith the goalkeepers control and such freak occurrences shouldn’t be used to judge a keepers effectiveness, in my opinion.

From doing this one fact leapt out: a team is 4 times as likely to score from an In-Box chance than a Long Range one. Of shots of target, 40% of In-Box chances resulted in a goal, as opposed to 13% from distance; of total shots, 16% of In-Box shots were scored, 4% of Long Range shots.

This confirms to me my suspicion that the basic figures provided on the official site are flawed. Two keepers could maintain an identical save percentage, but the keeper with the greater In-Box save percentage would be of much greater value than the guy who stops everything from 30 yards.

When I talk about “shots faced”, what I mean are shots on target – ie, shots that make the keeper work. Some teams have better block percentages than others, or are better are pressuring teams into missing the target, but these are more measures of defensive strength than goalkeeping ability, and I may come back to these in a future article.

So, ground set, lets move on.

The Trade

The Timbers conceded fewer goals with Ricketts in goal, dropping from 1.68 goals against per game (GaPG) to 1.58 GaPG (a difference of around 3 goals across a whole season) and this is borne out by looking at both keepers save percentages.


In each metric Ricketts outperforms his predecessor, and the difference in In-Box Saves (IBSv) jumps out the most – from 50.9% with Perkins to 60.7% with Ricketts. Given the importance of IBSv as stated above, this should lead to a marked difference, but the improvement, or “upgrade” if you will, was worth only 3 fewer goals conceded per season. Why?

Well, there was actually very little between the keepers in “bottom line terms”, as both conceded a goal from a shot in the box every 70 minutes or so (70.6 for Ricketts, 69.1 for Perkins – less than 1 goal difference across a season) and the reason for that was the Timbers actually allowed more In-Box Shots (IBSh) late in the season.

inboxshotsThe 10% improvement in IBSv was offset by the 20% increase in IBSh. There’s unlikely to be one single reason for the increase in chances, but the change of system and a new goalkeeper (who the defenders weren’t familiar with) could reasonably have contributed.

Despite this increase in IBSh against, the Timbers actually allowed fewer shots in total at goal with 44% fewer Long Range Shots (LRSh) against. There’s little surprise here as moving from a 4-4-2 to a 4-3-3/4-5-1 would give the team an extra body in midfield to block and pressurize players on the edge of the box.


Though Ricketts faced more shots in Portland than he did in Montreal, it was still a good deal fewer than Perkins faced. Perkins meanwhile had to face more shots in Montreal, though it would be pretty much on a par with his figures as a Timber.

As shown above, Ricketts brought an increase in shots saved to Portland. The same was also true of Troy Perkins in Montreal.


Again we see improvements in every metric, and much more markedly so than with Ricketts at Portland. The overall save percentage leapt from 61.4% to 81.4%, with IBSv rising from 50% to 71.4%. So, even though Perkins was facing more shots, he was delivering an overall improvement, contributing to a drop from 1.72 GaPG, to 0.89 GaPG, and 4 shut-outs in his 9 matches. Across a whole season, only Sporting Kansas City would’ve had a better defense than that with 0.79 GaPG.

Both clubs can certainly make a case that they got the best of the deal. From a Timbers perspective, the numbers certainly seem to indicate that the club were better off with Ricketts, but one would have to questions whether the modest improvements with a veteran keeper on a higher salary represent good value.

Enter Kocic

The Timbers went out and added Milos Kocic to the roster in the off-season. The Serbian keeper comes to Portland from Toronto which, if you were to just look at the goals against column, wouldn’t be very encouraging as Toronto were one of only 2 clubs to have a worse defensive record than the Cascadia Cup™ Champions. In fact, Toronto were rock bottom, shipping 62 goals, with Kocic picking 47 of those out of his net in only 27 starts.

Yet, things could’ve been much worse for Toronto if not for Kocic.


Kocic actually performable reasonably, if not spectacularly, well. In terms of LRSv, he performed about average for the league, maintaining a 86.4% save ratio. His IBSv was pretty similar to Ricketts’ overall ratio (52.4% for Kocic, 53.3% for Ricketts). Considering his salary is a fraction of that of Ricketts or Perkins, then it certainly seems like Kocic is a solid pick-up and someone who could, on performance, legitimately challenge Ricketts for the starting spot.

In trading away Perkins, the Timbers traded someone who had been on a run of 51 consecutive MLS starts after missing the start of the 2011 through injury. In that time Perkins missed 23 minutes after taking a boot to the face, in Montreal of all places. He’s since gone on to stretch that run to 60 consecutive starts, and if you look at the keepers who’ve had long careers, they tend to be the durable guys. I have my doubts about Ricketts’ durability.

I’d suspect, given Ricketts’ hefty salary, that the Jamaican will start the season as first choice, but I have no doubt that Kocic will get his chance at some point. At 35 years old, there’s a sense that Ricketts is increasingly prone to injuries and strains. While it’s true than most keepers can play on deep into their 30s and even 40s, once the niggling injuries start to bite, it’s usually a sure sign that the end is near. Though they may not need the cardiovascular fitness of a midfield dynamo, the goalkeeper position is high impact and takes it’s own unique toll on the body.


Another name that pops up now and then among Timbers fans is that of Dan Kennedy. Chivas USA have made a lot of noise about building an overtly Mexican identity and the Californian-born Kennedy doesn’t fit that aesthetic. Rumours have swirled practically all off-season about a potential move for Kennedy, and it’s easy to see why he would be such a prized asset for most clubs.

Chivas were the second of the two clubs to concede more than the Timbers in 2012, but once more you would say that things could’ve been much worse for The Other LA Team FC.


Kennedy maintained a 57.8% IBSv ratio and a 84.8% LRSv ratio – both very respectable numbers. He was badly let down by a defense that allowed 102 IBSh on target. To put that into perspective, it’s 3.2 per game compared to 3.1 for Kocic, and 2.8 for Ricketts and Perkins. That difference between Ricketts and Kennedy would result in around 14 more IBSh per season, or 5 goals (based on average scoring rates). Essentially, even if Kennedy performed as well as his peers, he’d still be expected to concede at least 5 more goals over a season thanks to an abysmal Chivas USA defense.

Kennedy’s 2012 guaranteed salary was $100,000 less than Ricketts’, which would make him an absolute bargain, especially with Kocic in reserve. However, last time I checked Donovan Ricketts’ Mexican credentials were even more tenuous than Kennedy’s, so any trade would be unlikely at best, rendering a potential deal to get Kennedy reliant on putting Ricketts elsewhere.


It’s galling to see Perkins’ figures improve so markedly on leaving Portland. It could be because he found himself behind a better defense in Montreal, or he was fired up by the way he was pushed out the door by the Timbers, or 9 games aren’t enough to draw any serious conclusions. Also, given the narrow focus of this analysis – ignoring defensive performance for one – it can only give a glimpse at the answer, if there even is one.

In Ricketts, as unpopular as it may be to say in some circles, the numbers suggest the Timbers did indeed get an upgrade. I can’t speak to what Ricketts also brings to the locker room, but in on-pitch terms, there was an improvement.

Kocic had a torrid 2012 in Toronto, eventually losing losing his spot in the team, and will undoubtedly want to prove himself. The figures suggest he’s a capable keeper and though I suspect he may have to be patient in waiting his chance, I think it will come. Last year saw Joe Bendik get a short run in the team, and I thought he performed reasonably well, but as soon as Ricketts was fit, Bendik was out. I’d hope that Porter has the constitution to sit the high earner if the guy in place is doing the job.

I would be shocked, but delighted, if Dan Kennedy ended up on the Timbers roster in 2013. If Chivas USA truly are determined to attain some kind of relevance by out-clusterfucking Toronto, then someone will pick up a very solid keeper, but I don’t think it’ll be the Timbers.

21 thoughts on “The Upgrade

  1. Good stuff Kevin. I appreciate the added analysis on Kocic. I suspected he played much better than the standings would indicate and am glad there is some verification from the stats. I think with his salary and potential for being the #1 starter (Ricketts is going to get hurt – it’s an evitability I think) this will turn out to be a very good move from the FO.

    1. Yeah, it’s incomplete without a bit more analysis of defense, and pairing the two together, but I’m only about 60-70% done with my defensive breakdown, and didn’t want to wait.

      I think the takeaway is that Ricketts is a decent enough keeper. He’s not the same guy who won Goalie of the Year, but for a team aiming for midtable, he’s perfectly servicable and hits the right numbers. My concern, I share with you, is his ability to stay injury-free. Kocic’s figures are pretty similar, and with the bargain salary, I think he could be a hidden gem in reserve, ready to step up if, or when, the time comes. Of course, this leaves Jake Gleeson as the forgotten man…

      I started working all this out because, in my gut, I felt that Troy was the better shot stopper, so I was surprised by some of the outcomes, but if you average it all across the year, both keepers come out pretty even. We’re just paying more, and have a keeper whose less durable (cue: long term Troy Perkins injury news breaking).

  2. It would be amazing to see Dan Kennedy in Portland. But I think he’s sticking around Chivas as Chelis recently named him team captain. Great analytics. I’m rooting for Kocic to be the guy this year.

    1. Yeah, as crazy as Chivas have been this off-season, it would be a whole other level of cuckoo to push Kennedy out the door. In fact, it was only late in the season that his figures dipped. Through the first 22 matches, he had an overall save percentage of 76% (way above the mean) and conceded 1 goal from 38 shots on target from distance.

  3. I know this was not in your brief for this (excellent) analysis, but I believe that Ricketts also outperformed Perkins in distribution. This was (and remained in Montreal) Troy’s Achilles heel. I don’t know if stats exist, but I would guess that “turnovers within two touches of distribution” would be easily double with Perkins v. Ricketts. This may account for why he faced more shots – he was giving the ball back far too often and frequently in our half. I cannot recall Perkins ever springing a fast break, while I clearly remember Ricketts doing it twice (once with an incredibly long and accurate throw). If this is true, then the “upgrade” is even more clear.

    PS – Glad you’re back!

    1. Of course, Ricketts injured himself on that long throw, which brings us back to your other point about durability….

    2. Funnily enough I do have incomplete figures on distribution. I broke it down to kicks (from hand, dead balls and live balls) and throws, where the passes/clearances were aimed and “successful delivery”, but it fell by the wayside as I concentrated on shots and goals against. I may go back to it, but I’m concentrating more on outfield players now.

      I didn’t have enough data to draw any conclusions yet (I had majority of Perkins’ and only a handful of Ricketts’, none of Kocic’s) but the early figures suggest very marginal difference though I agree, my “gut sense” is that Ricketts was better, at least at long distance.

      This raises an interesting point re: perception and the numbers. I started this to “prove” to myself that Perkins was the better shotstopper, but the figures don’t tell us that (and yes, I accept that you can’t make final judgements from such numbers, but still, I think I can inform the debate). Thinking on it, as I did, the reason I believe I had the perception was Perkins made better saves. Saves of the Week. Hollywood saves. Those are the ones you remember most. Ricketts made, clearly, as many if not more saves, but most were of “bread and butter” variety and so they don’t stand out so much.

      Those big “Hollywood” throws certainly stand out. I remember the first one, and thinking “Holy shi…”! But when you give it the same weight as a routine ball forward, the figures tend to even out between the two. Perkins may give you the big-time save now and then, and Ricketts may give you the blockbuster ball forward, but in general, there’s really not much difference between the two, I think.

      1. I was quite incensed by GW’s statements after the trade and spent the remainder of the season watching Ricketts with a gimlet eye (without the gimlets, unfortunately). Even my hyper-critical self could find little to fault. Granted I had no facts, but it seemed at the time that Ricketts really was a quality keeper. Much though I loved Perkins’ personality, his keeping fell in the good-but-not-great category. Ricketts could be classed slightly-better-but-still-not-great. My benchmark here is Van der Sar or Cech.

        A save is a save, regardless of the degree of difficulty. Steven Smith probably had the “best” save of the year for the Timbers. But effective distribution solves two problems: it drives the offense and can lead to goals; and it prevents further attack, thus changing (at least for a while) momentum. If you say that any keeper worth his salary will make the routine saves and a few spectacular ones, then distribution is the difference.

        After watching FA Cup games on Sunday, I would take ANY of our keepers over Chelsea’s #2. But I digress…

        I cannot wait to see your take on the outfield players. Perhaps you will prove it makes sense that Alexander is still a Timber 🙂

    3. Excellent point about distribution. I’ve been frustrated by Ricketts giving up rebounds but I hadn’t compared it to Perkins. It seemed like Perkins would be more likely to kick out of bounds or to the opposition, whereas Ricketts would deflect shots back into 18-yard box. My perception may be completely at odds with reality though.

  4. Are there metrics used to analyze overall team defense? You’ve broken down GK’s nicely, but is there any way to tell if their success/failure was due to having a bunch of idiots in front of them? We can SAY that Portland’s, Chivas’s, and Toronto’s defenses sucked, but can we put numbers to this? Numbers that separate the goalkeeper’s ability from the back four’s?

  5. Kevin – I’m glad your back bringing the sophistication. If we’re going to compare Ricketts vs. Perkins on distribution – the length of typical long range kicks has to be taken into account. A relative Ricketts strength vs. one of Perkins biggest deficits. Perkins has a very weak leg for a top flight keeper – resulting in 50-50 balls just past mid-field – not a good thing for a team. This is part of why I always thought Perkins was especially weak in terms of distribution. His strengths are of course well known here.

  6. From watching many matches the past two seasons I felt that Perkins was often out of position on long range shots, so opposing players were much more willing to take those shots on Perkins than on Ricketts. This may show why there is a larger discrepancy in that stat above (not just because of formation). As a former keeper I feel Perkins was not good at judging long range shots and being in the right place to make a good stop. From my perspective (based on observation and not stats) Ricketts was an upgrade because we saw less long range tests against him.

  7. Not that I disagree with any of the analysis but there is also the fact that Perkins was traded just after Spencer was fired and after a couple of miserable team performances like the 5-0 loss in Dallas and the whipping in Utah that really affected Perkins’ statistics. So while statistically I think this will always be viewed as an upgrade it’s impossible for me to believe Perkins would have much improved stats once Wilkinson worked on the defense as that was what he was known for.

    1. I’m not convinced those two results had anything to do with the trade. There were bigger failures in the back line and midfield than anything expressly wrong with Perkins’ game. It was probably already set up before hand. What I’d like to know is what the Merrit/Gavin/Porter brain trust saw that would make that such a beneficial trade. There must have been a lot of early plotting and planning and early strong suggestions from a man who was not even announced as hired until August.

  8. Nice analysis on the keepers. It will be great to see the synthesis of keepers/back line at some point. It’s so hard to get an accurate view without it, yet your analysis does bring out some sense of why the trade was made. While it’s not vital at all, I’m interested in PK stats. Is there an easy comparison on that? I felt like Perkins didn’t stop one during his time here. But I didn’t care. His heart and work ethic endeared him to me and most.

  9. The other piece of the puzzle that didn’t quite get mentioned regarding salaries was contract length. Yes Ricketts’s pay rate is higher, but his contract is up this year (2013), as opposed to Perkins’s going on for another year (2014). And while having the highest paid goalkeeper on your books for 1 year as you work to develop cheaper, younger options may not seem like the ultimate “value” decision, in the end it is probably cheaper and less logistically challenging than having the 4th highest paid keeper ( for 2 years. Add in the cost of labor for the FO to renegotiate a new Perkins contract, or shop him around, etc, and it does make financial sense if you look at it over a couple years.
    It allows the team to be confident in the product the have on hand, focus on the future development of players in the roster and academies, and if Ricketts is not a part of the future plans after this year, they can simply let his contract end.

Leave a Reply to chilepoblano (@chilepoblano) Cancel reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s