Briefly, on 2017 ATLUTD Finishing

Every so often Ted Knutson (@mixedknuts) who runs StatsBomb and StatsBomb Services will take public requests for analyticsy data viz type things his company normally only produces for clubs who employ them. I was lucky enough to have my wish for an Atlanta United xG histogram fulfilled, and I thought I’d post that here and talk about it for a minute. First, here’s what Ted posted on twitter:

The way to read this is that every shot Atlanta United took in 2017 has been stacked into a bin based on the underlying quality of the chance according to StatsBomb’s xG model. Each unit (horizontal bar) represents a shot, and each dark unit represents a goal. Pink-shaded shots and goals are those of Tito Villalba who Ted has highlighted.

It’s curiously bi-modal, in that Atlanta took a ton of low quality chances, but also created a fair amount of chances in that 40-50% range (or more broadly, 30-60% range). I would love to see a histogram like this for the league as a whole to see what of this is unique to Atlanta and what is par for the course. But my instinct is to say that the league as a whole would look more like a normalish distribution around the average xG/shot of 10.5 (converted 10-11% of the time) and so this Atlanta look is interesting. ATL has a ton of shots taken that are worse chances than an average one, and a nice pile of above average chances as well with a crater in between. It’s interesting.

I looked closely at Ted’s graphic and decided to deconstruct it so to speak. Here’s the graphic with the actual conversion rates for Atlanta United (and Tito in pink) typed out and hovering above the histograms themselves. Compare these rates to the xG bins themselves to see where the Five Stripes over-performed and where they under-performed:DN9lhE2X4AAoW9O (1)

Atlanta finishes at or above average conversion rates in all bins except for that 0.15-0.20 xG bin where curiously they also have not taken a ton of shots. I wonder if these are traditionally where corner kick type chances land on this chart since the team’s trouble with corner kicks and/or preference for short corners was well documented. Atlanta’s overall xG per shot of 0.12 is slightly above average for the league and it’s conversion rate of 14% is definitely above average. For chances registered as more likely to score than to not (xG ratings above 0.5), Atlanta finished at a rate of 76%.

As for Tito, he performed above average on the lower quality end of the scale, which is an unsurprising result given what we saw this season with those laser beams from outside the box. He finished poorly across the broad spectrum of “good chances” though, taking 16 shots in the 0.15 to 0.5 xG range (where one might expect to finish 30%+ of the time) and only finishing 2 of them (for a pedestrian 12.5%). Tito did finish the exceedingly good chances well though, putting away all 4 of the chances that registered as more likely to be scored than not.

As discussed on this blog throughout the 2017 season, finishing is a tricky subject. And generally speaking high conversion rates and goal returns in excess of expected goals do not persist over long periods. Further, there are qualitative reasons we’ve discussed as to why Atlanta may have overperformed xG this year that do not relate to their ability to shoot better than others. I think breaking down the chances into views like this one are helpful, although it doesn’t necessarily crack the case for Atlanta in 2017. Seeing a league-wide histogram distribution might help to tell a better story.

I would say the team as a whole (and definitely Villalba) could work on passing up long range shots in favor of working the ball around the box to find an entry into the dangerous areas to create higher quality chances. Even though they scored at higher rates from those distances than historical figures would suggest, I’d say they might move from good to dominant or at least hold off a sophomore slump if they could improve in this area in 2018.

For reference, here are some other end of season finishing-related charts just to round out our previous discussions on the topic:

AccuracyOpenPlayShotConvSOTConvTotalShots

And lastly, ATLUTD’s place in history using AmericanSoccerAnalysis‘ data for goals minus expected goals. Definitely stands out, and this is without much love from set pieces (which TFC were excellent at) and penalties.G-xG History

Advertisements

Squad Construction & Competitive Edges in MLS Roster Rules: A Manual

“It depends” is a frustrating answer to any question. I hate it. Even if it’s the answer that makes the most sense. In today’s post, which I hope will live on as an evergreen document, I’m not going to use that answer.

The question at hand: What is the best way to build an MLS squad using the myriad of complicated MLS roster rules: salary cap, designated players, GAM, TAM, Allocation Order, Loans, Superdraft, etc?

The answer: There are right ways and wrong ways to use each of the complicated MLS roster mechanics. Period. It does not dependWelcome to the MLS Squad Construction 2018 User Manual. I advise you to put on a fresh pot of coffee, and perhaps, bookmark this page.

While I will touch on some of the major rules, generally speaking, this post is not an explainer of the MLS Roster rules (like you can find here). Instead, it provides pointers on how best to take advantage of each of the key roster requirements, rules, and mechanisms to create a superior squad structure. I should note that this analysis is based not as much in empirical method but in one of intuitive theory. I believe it is a sound underlying logic but please chime in by commenting on the post if you have questions or concerns.

Get ready for a long list of “Do”s and “Don’t”s for any MLS Front Office. Nashville SC, this one is free, the next one will cost you. LAFC, I still haven’t received that check you promised. Here is a table of contents for what is to follow:

MLS Squad Construction Manual

  1. Introduction to Framework
  2. Allocation Money as Increased Budget (not as tradable asset)
  3. Designated Players and Young DP Savings
  4. The Option Value of Loans and Subsidized Wages
  5. Monetizing Allocation Order Ranking
  6. Differences between GAM and TAM
  7. International Slots & Investing in Scouting/Analytics
  8. Limited value of SuperDraft picks
  9. The real benefit of Homegrown Player Contracts

1. Introduction to Framework

Major League Soccer is a league of relative parity compared to other leagues around the world. This is by design as it is a single entity and the parity enforced via a complex set of roster construction rules orbiting mostly around a “salary cap” or “salary budget” depending on how technical you want to be. The broad strokes are that 20 senior players per team carry with them a “charge” that for most of them equals the wages they are paid by the league as a whole (single entity), no single player can have a charge higher than $480K (players who earn more than $480K are only allowed on a roster through the use of various other mechanisms like “Designated Players” and “Allocation money”), and the total “budget charges” on a team cannot exceed $3.8M (2017 amounts).

There’s a lot I want to discuss here, and I have given this a go before, but the core of what I’m going to talk about in this post is the general idea of maximizing a team’s talent given these roster constraints. As a general rule, I’m going to roll with the idea that wages (not transfer fees) are the better measure of a player’s competitive value on the field (outside of what his value as an asset might be in a cap or non cap environment). I first read about this in Soccernomics, but it’s a basically intuitive concept that on the whole, better players demand better wages, and so a team with higher wages generally has better players. I don’t want to get hung up on this point – we’re just going to need to accept it as a general fact – even though as I Google it, it appears some do take exception to the idea. And of course, teams make mistakes in signings. Regardless, it’s helpful to understand how to fit more wages into the constraints of the MLS Salary Budget given that most wages are funded centrally by the league and so with the exception of designated players, maximizing league-funded wages relative to a club’s competition does not come at an incremental cost to the individual club. And salary information is publicly available for MLS players.

Total Wages & League Parity

So I started by saying that there’s relative parity, and that all the teams have this “salary cap” thing of $3.8M, right? Yea, sort of. Here are the actual player wages paid by each MLS team in 2017 courtesy of Steve Fenn‘s excellent Tableau Viz:

Dashboard 1

These are the total team wages (players 1 – 29), not just the Senior players (1-20) whose charges must sum to the $3.8M budget cap. But the point should be clear, that while there is a salary budget/cap in place, teams pay vastly different total wage bills. Most of this variance is tied up in the optional signing of designated players whose wages can be as high as the individual owners are willing to fund them with the team taking only the league-funded $480K charge per DP against the $3.8M budget. What happens if we exclude DPs? While I’m working with a slightly different data source, here is a graph I put together of just the non-DP wages by team in 2017 (the league-funded wages):

League-funded Wages
This chart has the non-DP wages plus the DP and young DP roster charges (but not the DP wages in excess of the budget charges, which the individual club owners fund). The first black line approximates the nominal MLS Salary cap and the second line is a proxy for the cap plus an average year’s worth of distributed allocation money.

What you can see is that on a total dollar measure, the disparity between the teams is greatly reduced once you take away the excess wages supplied by individual owners directly for the designated players. On a low to high basis, we go from a range of $5M-$23M to $4M-$7M. There’s some timing noise in my numbers above, and I’ve done my best to add back the “league funded portion” of each team’s DP wages (200K for a young DP, 480K for any other). But an important reality is that once you take away all excess owner funding for designated players (a max of 3) and you’re just staring at the allocation of central league-funded wages (the salary budget), there’s still a not insignificant amount of room to maneuver if your goal is to maximize your team’s league funded wages relative to the rest of the league (as a proxy for maximizing your team’s talent relative to the rest of the league).

I should mention that while most of this article talks about the importance of maximizing league-funded wage bill, it is absolutely important to get your signings right also. All things considered, Orlando City popping up atop that graph above shouldn’t dissuade you from believing that more league-funded wages is better. They just signed some bad expensive players over their life (and to be fair, they only paid half of Dwyer’s wage this year and it’s counting his annual salary in the above).

2. Allocation Money as Increased Budget (not as a tradable asset)

The disparities in league funded team wages originate in several places, but the largest driver is Targeted Allocation Money (TAM). Each team received $500K in 2015, $800K in 2016, and $1.2M of TAM in 2017 (and there are reports of a fresh batch of $2M+ coming in 2018). This pool of league funds has several uses. Its primary use is that it allows teams to pay players that make more than the maximum budget charge and still comply with the roster rules without having to tag them as designated player (using TAM to pay down the wages of high earners to the max charge). As an example, this might allow a team to pay a total of $5.2M in non-DP wages even though  the cap is $3.8M. TAM can also be applied to a transfer fee which otherwise would blow up a team’s budget number or require the player to be tagged as a designated player. And, it can be traded between teams. So already, we’ve got 1 way to use TAM that increases your team’s wages relative to other teams, and 2 ways that do not.

If you could choose between 1) signing a player for free who is worthy of a $700K wage by applying TAM to his wage and 2) paying a $500K transfer fee for a player worthy of a $200K wage and applying TAM to the transfer fee, it’s clear which one is the more efficient choice when building a squad in this rule-intensive environment. Find the players who are out of contract that can help your squad (seems obvious) and you’ll be able to pay higher wages than if you’d spent TAM to pay a transfer fee. And all else equal, the third option is even worse: trading your TAM to another MLS team for another MLS player’s contract (think of it like an intra-MLS transfer fee). Instead of spending your TAM money to purchase a player like in #2, you do that PLUS you give another MLS team more TAM –TAM that they can use to increase the wages of their roster relative to yours! Transactions like these are veritable “six-pointers.” If you trade someone $200K of TAM, they now have exactly $400K of potential wage budget more than you had before the transactions. If the primary framework of this post is to max out league-funded wages to build a stronger roster than your opponents, trading allocation money is bad.

Orlando City Examples (Good and Bad)

  1. As an example of what a team should do, Orlando City traded Kevin Molino (2017 guaranteed comp of $400K) to Minnesota United for $650K of allocation money. If Orlando City thought they could find a suitable replacement player on the international market demanding somewhere around $650K of transfer fee and $400K in wages, this move makes sense as they’d be extracting team wages from Minnesota, on net. Alternatively, if there was an international player out of contract somewhere that fit their needs, they’d potentially have $1M of cap room to play with to find a replacement TAM player (a real potential upgrade over Molino).
  2. As an example of what not to do, instead of using the allocation money they’d just received from Minnesota, halfway through the season Orlando City traded $1.6M of allocation money to acquire Dom Dwyer’s contract (2017 guaranteed comp of $670K) from SKC. Dwyer’s current contract requires its team to feed it a minimum of $180K of TAM in addition to the max budget charge of $480K. By making this trade Orlando has lowered its team’s league-funded wage bill relative to its peers while still taking on a big contract that requires continuous league assets to stay compliant. On top of this, it sounds like Dwyer’s contract is nearing its end and his next one will need to be DP level for him to re-sign, so it’s possible Orlando’s best case scenario and worst case scenarios for Dwyer are both difficult to swallow. Worst case (i think), he leaves when his contract is up, and the team effectively paid $1.6M for a season and a half of no playoffs. Best case, he signs and becomes a DP, and now he’s taking up one of those precious DP spots for $480K max cap charge.
  • Rule: “Use” TAM to reduce a player’s cap hit to your team’s salary budget, don’t trade it to other teams. At least not for another MLS player contract.
    • Sub-rule: If a team offers you TAM for a player that you can find a replacement for from outside MLS for similar money, go for it. You’re lowering their total potential league-funded wage bill. You will either use the received TAM to replace the player via transfer fee, or if you can find a similar player on a free, you can apply that TAM to those wages or increase your wage bill elsewhere while decreasing your competition’s overall roster strength.
    • Sub-rule: These same principles should work for General Allocation Money (GAM) as well. GAM is like TAM but it has even greater utility in its flexibility.

3. Designated Players and Young DP Savings

Recall the chart above that showed most team wage inequality comes from designated players. This is obvious, but it is critical that a team “hit” on their 3 designated player signings. There are two different lenses we’ll view this from. First, the fan lens, where we can assume that our team’s owner has virtually infinite resources and does not consider transfer fees for designated players to be an investment. This is a simple thought exercise: simply sign the best players that are available and that you can convince to play at your club: most likely attacking players as goals and assists are hard to find. But there are simple ways to maximize your overall league-funded wages with designated players also: namely signing young designated players (aged 23 and younger) who only count $200K against the cap instead of the standard $480K. If you were to sign 3 stud young DPs instead of 3 DPs in their prime, you would have $840K of additional cap room to spend on strengthening the rest of your squad. This is a significant chunk of change: Tito Villalba, Victor Vazquez and Sascha Klestjan make around $650K each. However, I have to mention that if this is strictly the fan perspective first, you may not want to spend all 3 of your DP slots on young (potentially risky) players when you could buy 3 David Villas instead that are proven.

However, from an owner’s perspective — specifically an owner who cares both about having a winning team AND about internally financially prudent things like ROI — the 3 designated player slots should be thought of as investments to the extent that the players signed command significant transfer fees. If this is the case, the Young DP route is absolutely the way to go. First, we already mentioned the competitive advantage to the rest of the team’s budget structure ($200K vs $480K cap hits for each DP). Secondly, if you care about ROI, then any one-time transfer fee paid to another club around the world should be thought of as a prepaid asset and not as an expense. Simply put, you should be paying a transfer fee that you believe you can recoup when the player departs the club. And really, the only way to do that is to sign a young player whose value is still on the rise. The goal doesn’t even need to be to receive more than you paid, simply to recoup. You can’t do that with an established player but you can with a young player. The cap room you save by signing young DPs will allow you to strengthen elsewhere with veterans.

Good examples of this are FC Dallas (who filled all 3 2017 DP slots with young DPs, successfully generating that extra $840K budget amount) and Atlanta United whose Almiron and Villalba counted as young DPs (a combined $400K against the cap instead of $960K), allowing them to spend $325K on an expensive veteran CB like Michael Parkhurst to captain the expansion side. Below is a breakdown of the 2017 DP wages.

DP Wages
Someone will probably call out Montreal. For whatever reason Piatti still shows up in the Players Union Salary releases as making $450K (under the DP threshold). I don’t think this is true, but the data is the data.
  • Rule: Sign young DPs instead of old ones to a) gain an $840K budget advantage with which to strengthen the rest of your squad and b) to increase the chance of eventually recouping transfer fees paid to sign your DPs, and c) increase the chance of turning a profit on the player’s departure, which could add up to $650K of GAM to your team’s overall budget (or more, if they change the rule).

4. The option value of loans and subsidized wages

If we maintain our goal of increasing the squad’s wage bill relative to the competition, there’s another tool to add to the kit. It is customary in global football for clubs to loan young players to one another for half seasons or seasons at a time. Acquiring players on loan is helpful in MLS for a few reasons. The first is that it is not uncommon for a home club to pay a portion of the loanee’s wages. Having a player on your roster in MLS but not having to pay 100% of that players wages is immense as it allows you to once again strengthen the total wage bill of your club (and it’s overall roster strength) relative to the rest of the league. Secondly, loan agreements often have purchase options and options are always good. Specifically in MLS, the ability to not bring a player back next season if he’s not working out this season allows you to free up the cap space needed in the following season to find the player’s replacement. In MLS with strict salary cap rules, a bad player on an expensive contract that you can’t get rid of can cripple your team’s overall budget strategy in a way that doesn’t exist in other league’s around the world.

In 2017, Atlanta United acquired 4 players on loan: Josef Martinez, Yamil Asad, Greg Garza, and Anton Walkes. Martinez was a DP whose option was exercised after just 3 games because it turns out he’s Thierry Henry but good. Asad and Garza were paid $150K each by Atlanta, suggesting to many of us that their home clubs footed the bill for the rest. And Anton Walkes was a league minimum $53K charge for Atlanta, suggesting Tottenham paid some portion in addition to this. Atlanta was able to allocate these additional funds to strengthen the squad in other ways. But more importantly, nothing is ever a sure thing. It’s possible all 4 of these guys might’ve been failures and had the deals not been set up as loans, ATL could’ve been heading into 2018 with no roster flexibility (like Orlando City). The roaring success of all 4 players in 2017 has left Atlanta with much better problems to have (which I’ve discussed in other posts).

  • Rule: Roll the dice with 3 or 4 high upside loanees with purchase options while if possible having their home clubs subsidize the wages. If 2 are winners, exercise their options and call it a day. The rest go back to their clubs next year and you have cap space to play with. Use wage subsidies on loanees to strengthen other areas of the team, increasing your effective team wages relative to the rest of the league.

5. Monetizing Allocation Order Ranking

OK, this one is just weird. Basically there’s a ranking/priority order for MLS teams chasing USMNT players or other good players that used to play in MLS but who’ve left and want to return to the league. Basically if you want to bring one of these players into the league, everyone in front of you in the ranking order has dibs before you can sign the player. This order resets each season in a worst to first ordering (giving bad teams the first crack at utilizing this mechanism in the offseason) with expansion teams automatically getting the first pick. So a couple features to this thing that are really important. First, it appears to me that the only spot in this ranking that has value is the #1 pick as the #1 pick has right of first refusal whenever there’s a player trying to reenter the league. Second, as the summer transfer window closes (halfway through the MLS season), the market value should start to disappear because if the team in the #1 spot doesn’t use the mechanism it doesn’t get to hold onto it: the order will reset the following year and they’ll likely lose that spot. Third, the timing of when players decide they want to come back to MLS and whether these players meet the needs of MLS teams matters and is somewhat random. Fourth, just from looking at recent transactions and the valuations inherent in them, this thing matters, and clubs place significant value on it. Here are the 2017 allocation order uses and trades:

  • Chicago trades GAM (undisclosed) + 1st round pick + 3rd spot for Minnesota’s #2 spot (Juninho)
  • Atlanta signs Brad Guzan (free transfer)
  • VAN trades $225K + International Slot to Minnesota’s #1 Spot (Freddy Montero loan)
  • DC trades $175K + 9th spot in the order for Dynamo’s #1 spot. (Deshorn Brown)
  • NE trades $175K + 5th spot in the order for San Jose’s #2 spot.
  • NE trades $400K + 2 years of intl slots to CLB for rights to Kriztan Nemeth through CLB’s #1 spot in the order

Just eyeballing it, that looks like basically if a decent player is on the line, we’re talking about the #1 ranking having a trade value somewhere in the range of $300K – $600K in allocation money. Additionally, similar to the principles we discussed earlier around TAM, the type of player you bring in through this mechanism is either a high value TAM player or a DP — my point being if you trade for the #1 pick, you’re trading a bunch of allocation money and then needing to use more allocation money to pay the wages of the player you’ve acquired. While I don’t know for sure, I can’t imagine any team would trade league assets to another league team in order for the rights to then pay a hefty transfer fee to a foreign club for the rights to pay a hefty wage to the incoming player. But, I’m sure this has happened. What’s probably much more common is that teams monitor the allocation list for players whose contracts are expiring. As an example, Atlanta United were able to use their #1 ranking in the allocation order to sign Brad Guzan on a free transfer (i suspect he’s on $700K wages or around there).

The question though is, what’s the best thing a team can do with a high allocation order ranking (what’s our next “Rule”)? It appears to me that if you’re going to exercise your privilege from the top spot of the allocation order, you have to strike it big and you have to find a player who is out of contract (Brad Guzan is a good example). But there are generally, more efficient plays: namely, trade the allocation ranking down to another team who has their hopes set on an available player. Collect allocation money and other assets from that team, thereby necessarily increasing your team’s potential wage bill relative to your competition’s (all at zero actual additional cost). Because the #1 ranking is the only one with real value at any given moment, an ideal string of transactions would be finding yourself at the #1 spot multiple times in a given season, by trading the #1 spot down for the #3 spot let’s say + allocation money. Then after one more team uses their position, you’re back in #1 and ready to fleece the next team for allocation money should a big name be available. All the while, you would have to keep an eye on the clock and realize that your allocation order will not carry over to next season.

So, while Atlanta’s Brad Guzan signing was certainly a success, what Minnesota did was also very interesting and in my mind, clever. They started the season in the #2 spot, but on Dec 23 traded the #2 spot to Chicago for the #3 spot + allocation money (undisclosed) + a 1st round draft pick . Then After Chicago took Juninho and Atlanta took Guzan, Minnesota had the #1 spot and traded it to Vancouver for $225K + a 2017 international slot. So the basic concept is there, so long as you hold teams near the top of the order hostage for the highest possible ranking spots, you can put the league through the windmill so to speak, repeatedly landing yourself back atop the order and demanding more assets to let other teams go after their guys, all the while increasing your league-funded team wages relative to your competition by buying TAM not selling it. There’s some nuanced game theory in there somewhere, but at least as a framework this should be the goal. LAFC, read this paragraph again.

  • Rule #4: Trade allocation ranking order down a few spots to collect allocation money, then do it again once your team is back up in the top spot. Remember not to let the summer transfer window close on you in the top spot.

6. Differences between GAM and TAM

In principle, general allocation money and targeted allocation money do the same thing: they allow a team to have a wage bill that exceeds the nominal $3.8M salary budget cap. But there are key differences that should have an impact on front office behavior. First of all, targeted allocation money can be applied only to players who make more than the $480K max budget charge but less than $1M. And importantly, GAM and TAM cannot be combined and applied to a single player’s budget charge. Because of this, if a team is trying to sign a player right around that max budget charge and also comply with the league’s $3.8M budget cap, whether his wages are ultimately above or below the max charge may depend on which allocation money resource is more plentiful for the club. You may want to pay a player $500K instead of $475K if you have enough TAM (but not enough GAM) to reduce his wage low enough to slide under the overall budget cap. In order to increase your club’s league-funded wage efficiency relative to your competition, it may involve paying a player more than his market value, as counter intuitive as that may sound. Sam Steejskal first reported on these sorts of odd incentives.

Secondly, understand that while General Allocation Money is indefinite-lived, Targeted Allocation Money does in fact expire after 4 transfer windows. As an example, TAM issued in 2017 will expire at the end of the 2018 season. This presents interesting and complex scenarios. Take for example a situation where Team A may have TAM that’s about to expire. The value of this expiring TAM is quite low to most teams, but perhaps there’s a team out there (Team B) that is trying to pay a one-time transfer fee before the window shuts to secure a player. Assume also there are other teams with varying common needs of TAM and stocks of TAM. Team A’s goal should be to identify the appropriate trading partner (Team B) who has a need for TAM for the purpose of using it in the very short term (before it expires). Offers from other teams in the league will surely be low-ball in nature because the TAM is about to expire. Team B’s goal should be to find a cut rate deal for TAM given the presence of TAM that is about to expire. If Team B deals with any other team than Team A they’ll likely have to pay more for the same amount of TAM (different vintages). Something to think about. Less of a direct takeaway.

7. International Spots

I’ve mentioned over and over again that when teams are considering trading for a player currently under contract with another MLS team, that they should evaluate whether there are comparable players available on the global market because it makes sense to use TAM towards a fee outbound from MLS rather than trade another MLS team an asset that increases their wages relative to your own for a similar player. In order for this process to work, you need International Spots/slots. Each team starts with 8 but these are freely traded in 1 or 2 season increments. Based on my research, the market value seems to be between $50K and $75K of allocation money, but they’re also acquired by trading other assets like draft picks or players. I’ve also noticed that this market normally clears. Some teams like to have a lot of IS slots and others only use a handful, so teams don’t seem to have trouble acquiring the international slots they need, and I’m yet to see a team really be held over the barrel for one.

Most importantly though, if you can get your international players Green cards after they’ve been around for a year, then you no longer have to use an IS on them. This is a free lunch, and every team should pursue it. As IS spots free up, you can either monetize them or use them to sign players on the much more liquid global player market (compared to the unionized and single entity-contracted MLS player pool)

Example: Atlanta getting Green Cards for Kenwyne Jones, Chris McCann, and Kevin Kratz. Not only does it free up IS spots, but it theoretically increases the intra-MLS trade value of these players slightly.

  • Rule: Acquire international spots in order to ensure the ability to compare MLS trade opportunities to global player market. Work international players through the US Customs Office Green Card process to free up international spots and increase player trade value.

This is a good time to mention that I fully realize that the directive to sign comparable internationals instead of trade for MLS players in an effort to maximize your league-funded wages relative to your peers comes at a real cost (not a TAM/GAM asset cost). This is because while trading assets for MLS players is not an efficient use of league assets and cedes inherent competitive advantage, it costs less “real life dollars” to gather information on MLS players than it does to hire scouts and analysts to generate information on players abroad. The costs of setting up a quality scouting and analytics department do not count against the salary cap. They are simply real costs to be incurred by the owner of a club. So one way to think about this is a tradeoff between the following two models:

  1. Spend more money on scouting and analytics. In doing so, spend allocation money on wages and transfer fees for international players that you otherwise would trade to an MLS team for a comparable league player. In doing so, you increase your team’s league-funded wages relative to your peers, increasing your theoretical advantage.
  2. Spend less money on scouting and analytics. Instead, give up league assets to other MLS teams in order to acquire league players (the cost of this information is not the real life dollars you would spend on scouting/analytics but instead the ceding of competitive advantage to other MLS clubs).

One of these seems like MLS 3.0, the other MLS 2.0, at best.

8. The limited value of a SuperDraft pick.

I’ve placed a high value on allocation money in all of the above rules/arguments, so I should mention an asset that should be given a much lower priority: SuperDraft picks. Each year players are drafted out of college via the 4 rounds of the MLS SuperDraft. Before the draft there’s a combine where players work out and scrimmage in front of all the teams. Generally speaking, players drafted in the SuperDraft either make the senior minimum wage or the reserve minimum wage, or if higher than that, they are tagged as Generation Adidas players (normally the top prospects) and they are automatically eligible to be placed on the supplemental roster (and therefore not count against the cap). Kevin Minkus at AmericanSoccerAnalysis did a study to understand the value of SuperDraft picks based on their ordering. I’ve included his chart below:

SuperDraftMinutes
Kevin Minkus at AmericanSoccerAnalysis pulled together this helpful chart looking at empirical success rates as measured by minutes contribution across the 4 rounds of the draft.

Basically each pick in the first round is expected to contribute fewer and fewer minutes over the first 2 years a player is in the league, with anything outside of the first round being mostly valueless on average. When you consider that a healthy MLS academy is pumping out homegrown players who are competing for the same roster spots, there’s a very real argument that an MLS team should not want to make any pick outside of the first round (anomalous draft classes not withstanding). This makes me thinking that whenever a team needs to trade for a player or an asset, they should first start by offering SuperDraft picks. If you need an international slot, try a SuperDraft pick, what about a nominal amount of GAM or TAM, try a SuperDraft pick. Want to move up a few places in the allocation order? (say from 7 to 5 and then hope 4 teams utilize their rankings allowing you to spend some amount of time in the top spot) Try trading a SuperDraft pick or two.

Something you should not do is trade allocation money for a better SuperDraft pick. I simply do not understand it. As an example New York City FC traded $250K of GAM to Chicago for the rights to draft Jonathan Lewis, who is certainly a prospect with some upside. But I don’t follow why drafting Lewis and hoping he develops into a good player who if he becomes a good player will demand a higher salary is better than spending $250K of GAM (effectively $250K of salary budget) on a player of the $250K caliber. Players who make around $250K include Leandro Gonzalez Pirez, AJ DelaGarza, Steven Beitashour, Drew Moor, Sebastian Lletget. Point is that’s the caliber of player you can have on your team at that wage amount – now acquiring one of those players may cost something, or it may not. Older players like Jeff Larentowicz ($175K) was a free agent around this time.

In contrast, Atlanta United have done a decent job trading SuperDraft picks for assets, including: 2017 #24 pick for 2 seasons worth of international slot, 2020 4th rounder for Kevin Kratz, 2019 fourth rounder for Harrison Heath, 3rd rounder or conditional 2nd for Romario Williams, 2018 2nd rounder for the discovery rights to Greg Garza.

They did trade an expansion drafted player for the #8 pick and turned that into Julian Gressel, which has worked out swimmingly, but I don’t think this sort of move works out very often.

  • Rule: Trade your Superdraft picks away for assets you can use to increase your team’s total potential league-funded wages relative to your competition.

9. Homegrown Players

The league (and surely in partnership with USSF) has created special rules around homegrown players that are supposed to create incentives for clubs to invest in youth development. I want to clarify specifically what the incentives accomplish here, because I think there’s some confusion about the way people talk about this. First the rule: basically, if a player qualifies as a homegrown (he’s played in your development system for a year), he isn’t subject to the Superdraft and instead you can sign him to your team as a homegrown player. He will occupy a supplemental or reserve spot on your roster so he won’t count against the cap, so long as he makes $125K or less (they start out making much less: Tyler Adams makes $75K).

So I want to be clear, that the primary benefit of the homegrown player designation is *not* that a homegrown player doesn’t count against your cap. After all, no supplemental or reserve roster spot (21-29) counts against the cap (basically the low earners in MLS). If we play the scenario out where a club successfully develops a star out of their youth system, while he’s still developing in his early years in MLS he’ll be on the supplemental/reserve roster, making $75K or so and not counting against the cap. But if the player truly becomes a star, at some point it will be time to renegotiate a contract and his wage demands will exceed $125K (perhaps at this point there are clubs overseas offering him much more than this). It is at this point that an MLS club can either a) sign him to a fair market value wage, at which point he (like all other good players) counts against the cap and the homegrown player tag goes away, or b) sell him overseas and take the profits from the sale (which is basically the entire fee) after MLS takes its cut (or doesn’t if they change the rules as reported), and convert $650K of this profit into general allocation money to increase the strength of the squad. If you’re still reading at this point, I think you know where I’m going with this. The primary purpose of the homegrown player rule, is not for MLS teams to create homegrown superstars and keep them in MLS (if it was, then the rules would be much more generous to teams who do this – maybe the player’s wages would be exempt from the cap hit regardless of amount). The rule is designed to create an incentive for MLS teams to develop youth players in the area into stars and then sell them overseas. One important note, which I may explore further in another post, is that in order to sell a player oversees for a significant transfer fee you need to resign the player to a long term contract. So the act of turning a non-cap hitting homegrown into a transfer fee may involve a transition period, where the player is briefly making something closer to a fair market value wage before a foreign club comes in for him.

  • Rule: Develop homegrown players by stashing them in spots 21-29 on your bench and getting them minutes, and then when they have developed into stars, sell them to generate profits which you can convert into general allocation money to further strengthen your team (increase your team’s league-funded wage bill relative to your competition).
    • Alternatively, you can sign a homegrown player out of college who would otherwise have entered the SuperDraft if he has fulfilled the 1 year of development within your academy. This 22-23 year old is likely more ready to see significant minutes in MLS than a 16 year old phenom prospect. Accordingly, there’s a benefit to paying the 22-23 year old an entry-level homegrown wage and have him not count against the cap.

That’s it for now in terms of the “Do”s and “Don’t”s.

I should mention Jared Young at AmericanSoccerAnalysis has a similar full breakdown of Roster Construction here. He comes to some different conclusions about what portions of the roster a club should focus on. It’s an interesting read and a very helpful breakdown of the key rules.

Let me know if you have questions or ideas on where to take this from here. I’ll likely be publishing something like this in chapter form over at DirtySouthSoccer. I’d also like to looks specifically at some of the nuances of how an expansion team can find advantages – with the expansion draft and other types of things.

I also have vague plans to publish some MLS Trade Price Guides as the league becomes more and more transparent with the amounts being traded around.

 

2017 MLS Playoff Opposition / Matchup Analysis: Columbus Crew

This is exciting…

The return of match previews / opposition analyses, whatever you want to call them. Hello wordpress. It has been a while.

Atlanta United (#4) plays Columbus Crew (#5) at 7pm Thursday October 26, 2017 with #Narrative abounding. Quickly just to set the narrative table, before we get into the useful stuff:

  • Columbus Crew are probably getting moved to Austin, TX after the 2018 season
  • ATL were inches away (multiple times) vs TFC from earning a first round bye
  • Atlanta United fans boo’d the most recognizable US National Team players vs TFC
  • Both teams played Sunday afternoon/evening after a week of rest.

I am going to preview this matchup / prepare a stats-based opposition analysis.

First, to relive the home match at Bobby Dodd Stadium earlier this season, check out my old post here. If I’m honest, the content was better then. That was a fun match.

Goal Keepers (we’re going to go back to front, sorta):

Atlanta United have one of the best goal keepers in the league. Possibly the best. AmericanSoccerAnalysis has a model for measuring the average number of goals a keeper will concede on average for a given profile of shots on target he faces. If you compare the goals Guzan has conceded to his expected goals based on this method, he leads all keepers but one in MLS with -0.28 per game. Only Tim Melia’s -0.36 per game tops it. Crew SC’s Zack Steffen, on the other hand, is 18th out of 22 starters in the league by this metric, conceding 0.13 more goals per game than the model would suggest. See below for a visual. Red dots are playoff teams.GK Season Review

In terms of distribution, Guzan leads the league in pass completion rate, largely because he plays the highest percentage of short passes in MLS. Steffen is 7th in MLS and not so far off substantively. Columbus likes to play it short out of the back as well. See below:

GK Distribution Season Review

Style of Play

One metric to use for style of play comparisons is Cross to ThroughBall ratio. If I look on whoscored I can get this for key passes only (couldn’t find throughballs that don’t result in shots – probably because they are rare). Columbus looks similar to Atlanta in this regard:

CrosstoKP Season Review

Another metric we can look at is the portion of a team’s total passes that are long passes. Columbus plays the second fewest long balls in MLS (behind NYC). They are technically proficient and like to play the short passing game.

Long Passes Season Review

When Columbus visited Atlanta earlier in the season, they upped the percentage of long balls to 18% (16% in the first half). Even at home in Columbus, they passed long slightly more often – likely to combat Atlanta’s high pressure and bypass the midfield (to minimize turnovers). Here were the pressing/tidiness stats from those matches.CLB1 reviewCLB2 review

Across the two games, Atlanta turned Columbus over in its own half 6 times per own-half pass on average, which is a slightly less furious pace than Atlanta’s home average of pressing its opposition into errors every 4-5 passes. On top of this, the Five Stripes are disrupting opponents’ own-half buildup at a lighter pace in Mercedes Benz stadium, whether due to fitness conservation or the size of the pitch, or another reason. giveaways season reviewThis all adds up to make me wonder if we are going to see the same sort of high pressure success on Thursday. This worries me as an ATLUTD supporter because I know the team is comfortable simply releasing the hounds of pressure.

Centre Back Distribution

Since we’ve established both teams like to play from the back, let’s see how the centre backs stack up in terms of distribution. Below is a graphic of pass completion % on the Y and Long Ball % on the X, with Crew SC’s starting defenders colored yellow and Atlanta’s red:CB Distribution YE Review

Both teams appear to be above average passers based on this simple metric, and surprisingly, Atlanta’s CB’s actually play long balls slightly more often than the average centre back with over 15 appearances this year. If I had to pick a Crew CB to focus on turning over it would be Mensah, just slightly. ASA has a handy tool that’s more sophisticated than what I graphed above to measure passing skill, which they call xPassing.

xPass CB Season Review

According to this metric, all 4 starting CBs pass at a more successful rate than one would expected for the types of passes they attempt. Columbus’ duo shows up slightly better here. For anyone worried about Parkhurst not being fit for Thursday, these metrics show Larentowicz as slightly worse than LGP on this metric (but still positive), and Walkes as a worse distributor (-5.3%).

Atlanta created a goal in the first match when Asad stole the ball from the back line and found Villalba in the danger zone for an easy-looking put-away. But based on the above data, I wouldn’t label this as a particular opportunity for Atlanta to exploit. Columbus is pretty good at keeping the ball and playing it on the floor from back to front.

In the first match (BDS) Columbus did not press exceedingly high as Atlanta advanced, waiting until the ball reached the top of the center circle to engage. And this worked generally well for them in the first half, creating a few good transition chances from picking off passes from the ATL centre backs at midfield. If Boswell or Walkes play in place of Parkhurst, expect this to change (with a more aggressive pressing approach), but in general, I think this is a fine strategy for an away side at Mercedes Benz Stadium. Below is a poorly drawn formations graphic showing expected lines of engagement for pressing. These will change as the game state changes, but I expect Columbus to press in that middle third more often to spring transition, while Atlanta hunts the ball no matter where it is (all the way to the Columbus back line). This isn’t to say that Columbus won’t press all the way back to the keeper when Atlanta are in a vulnerable moment (facing their own goal in defense etc).

493px-Soccer_field_-_empty
Expect Columbus to invite Atlanta’s press and then try to bypass midfield in transition with one touch passing or balls over the top.

Shapes

I haven’t watched much Crew lately which is why this post is mainly stats-type analysis. But I’ll trust Matt Doyle that the Crew have stabilized into a 4-2-3-1 with Kamara leading the line and Higuain, Meram and Santos forming the band of 3. Trapp and Abu in the middle with Afful/ Mensah/ Williams/ Raitala in a back 4.

While Martino has played with various formations in the runup to this game (4-4-2s against New England and Red Bulls, and 3-4-3 vs Toronto), I expect him to return to a traditional back 4 (with 2 CB’s marking kamara) and the deepest midfielder dropping back between the CBs while ATL have the ball. A 3 man back line isn’t optimal against a lone striker.

Creative Influence

You can probably guess who leads chance creation numbers as measured by expected goals + expected assists per 90. Kamara, Higuain, and Meram are contributing 0.66, 0.49, and 0.47 xG+xA per 90, respectively, with Kamara’s output primarily accounted for with an expected goal every 2 games, and Higuain’s and Meram’s outputs split more evenly between expected goals and expected assists. There’s no doubting that this is a dangerous front 3. And honestly, the answer to how to defend against these guys does not lie in the simple kind of analytics I’m capable of. But there are other things of course.

Interestingly, if we look at xBuildup (a metric published by ASA, helpful for assigning credit to players for their role in the buildup or creation of a chances that does not include the player taking the final shot or shot assist), an interesting name pops up: Mohammed Abu, registering 0.4 XG per 90 -> suggesting he has a hand in some form of the possession sequences leading up to (but not including shots or shot assists) about 0.4 expected goals created per game. Not bad since the Columbus are creating 1.4 xG per game on average. It would appear as though Tata Martino should be focusing some tactics to disrupt Abu in possession. This might mean placing an athletic/motor player centrall in the #10 spot to harrass and deny Abu time and the ball. Break the link most commonly chaining possessions into scoring chances and you will have accomplished something important.

Attacking and Defensive Trends

If we look at expected goals for and against trends for Atlanta home matches and Columbus away matches, it’s a little uncomfortable for my liking. Both teams hit an electric attacking run of form in September and have cooled off some, and both club’s defensive averages have inched up in October.

Crewe xGfa 3game

ATLtrendxgfa

Atlanta’s latest 3 game home form on the attack side is 1.6 xG, which Columbus has matched away from home. And Atlanta’s home form on the defensive side is most recently a 3 game average of 1.3 xG conceded, with Columbus holding their opponents to 1.2 xG on the road. There are always #reasons. The Minnesota game is weighs heavy on the Atlanta averages given the red card to Reynish and the injuries to Almiron and Garza and callups of Martinez and Guzan. And also, as I’ve discussed previously on the blog, big chances that Atlanta creates on the break are often not captured well by the expected goals model. The overall defensive trend before these last couple of games was very, very solid for the Five Stripes.

Other Concerns

Kevin Minkus yesterday pointed out that Atlanta United are giving up high quality shots off of set pieces and Columbus may in fact create good shots off of set pieces, even if it has been relatively few to date:

Ideas

This is scary. Lots to think about. I have this general fear of going down a goal early and then chasing the game against an awful game state (DCU type stuff), and so I’m hesitant to suggest really pushing numbers forward early to strangle the visitors — the thing is, this is exactly what the other part of my brain is screaming. Columbus will be travelling on short rest, with non soccer baggage lingering, and so it’s tempting to suggest a full on raiding party in the opening 10 minutes – kill them off early.

The thing is, the underlying numbers (to me at least) suggest a slowing of the high press at MBS. Maybe I’m not capturing what’s really going on, but I can think of reasons why the press works to less effect at MBS than BDS. And I also recognize that Atlanta have the home crowd advantage for this entire match, whether it be 90 minutes or 120 minute or all the way to PKs (shudders). I’d hate to spoil that by playing 75 minutes down a goal against a bunkering Crew who nicked a counter goal early. So I think my overall recommendation would be to exercise caution early on (I know you’re probably shaking your head), save some legs for what might be 120 minutes. The other recommendation is to task someone in the central attacking channel to press Mohammed Abu relentlessly, even at the expense of showing for the ball in transition. I’m not sure if this should be Almiron or Asad or even Gressel(!) who could be in for a starting role if Parkhurst isn’t ready (and Jeff slides back). And I hate to over think it, but something tells me we won’t succeed with Tito’s pace out wide in this one like we have in recent matches. We didn’t create much from wide in the first Columbus match (admittedly the field was narrower), and our goals came over the top in the second. I think I am calling for an inverted Tito on the left, a centrally located Asad, and Almiron inverted on the right. With players rotating to conserve energy and sanity as the central most player bothers Abu all night.

God speed. Hopefully, I get to do this again before next year.

 

Atlanta United Venue Analysis: Bobby Dodd Stadium vs Mercedes Benz Stadium vs Opponents Stadia

Josh over at Dirty South Soccer wrote a really nice piece on some of the observable differences between Atlanta United playing at Bobby Dodd and Mercedes Benz. Click here to check that out. I’m not as eloquent or as wise as him, nor as versed in soccer tactics, but I wanted to just drop some data / graphs in here that I’d been tweeting about in recent weeks to add to this comparison.

First up, here’s a table I keep to track some of the Atlanta United tidiness stats.

VenueTidinessCompare

Takeaway1: Atlanta are tidier on the ball at home, and more tidy on the ball at the Benz compared to Bobby Dodd. This shows up in the own-half passes per giveaways and own-half passing % metrics.

Takeaway2: Atlanta are suppressing opponents’ shots (and shots on target) at home, and most impressively at the Benz. The result is improved expected goals against figures as well as improved goals conceded figures.

Tweets that support this:

Next up, the pressing stats for Bobby Dodd, Mercedes Benz, and away matches:VenuePressingCompare

 

Takeaway1: Interestingly, while Atlanta United pressed more furiously and to greater effect at Bobby Dodd than on the road, the pressing numbers come down quite a bit at MBS – could be schedule congestion and fitness conservation – could be larger pitch makes it harder to press opponents into mistakes.

Takeaway 2: ATL Goals and Expected goals have trended up furiously at MBS compared to Bobby Dodd, both being higher than the average away matches output. Expected goal data comes from AmericanSoccerAnalysis.com as always.

 

Game by Game Pressing/Tidiness Game Stats (I hadn’t posted these here since the FCD match.

Finally, I wrote something about the inevitable sales of Atlanta United players and how this works out with the current MLS rules.

https://www.dirtysouthsoccer.com/atlanta-united-fc/2017/10/19/16465762/mls-transfer-rules-regulations-fees-percentages

God speed to everyone in the postseason.

Updated table math and squad rotation and what’s left to play for

I wrote a piece for DirtySouthSoccer, which you can find here, where I update us all on the PACE standings as of Monday 9-25-17. Then I ponder some difficult questions around whether one can rotate the squad when there’s still so much left to play for. #4 comes with homefield advantage in the play-in round, #2 comes with a first round bye. Atlanta are favored to achieve both of these, but the Almiron injury…

Atlanta United FC 3 – 0 FC Dallas

Chance Dominance & Efficiency in the Final Third

In the first ever soccer match at Mercedes Benz Stadium, Atlanta dominated the score sheet with a 3-0 victory and created their highest single game tally of expected goals (2.8 per ASA) this season. The team held FC Dallas to 0 goals and 1.2 expected goals (this spread is partly luck and partly Guzan). And the resulting 1.7 xGD is the highest winning expected goal difference for the club this season. All this points to dominance by the Five Stripes, and I think that’s probably fair. The box score suggests dominance as well:

Game Stats

However, I would say as a fan, I only felt comfortable in the second half. FC Dallas’ only two shots on target were two very good chances that Guzan had to make big saves for.

FCD on target and blocked
Green (SOT) early in 2H. Yellow (blocked) all in 1H.

And as dominant as Atlanta were in the end, those types of games will occasionally go south depending on the ordering of events and the direction of some bounces. I’ve included a map of all 5 of FC Dallas’ shots on target and blocked shots. They’re all very good chances, and would’ve altered events.

 

On the Atlanta side of things, 12 of 19 shots being put on frame is a high number (63% vs the league average 36%). The home side were exceedingly efficient at turning shots into #problems for Jesse Gonzalez. But as I recall, they were also much more efficient at turning attacks into shots (contrast this with away at Orlando where many counters and other attacking moves ended up just shy of registered shots). Touches Around Box - ATLWhoscored has Atlanta at 27 touches inside the box, with an insanely high 8 firing within the 6 yard box. While a few of the 19 were from range, 19 shots on 27 touches inside the box feels pretty high to me – it feels *very* efficient. Because of all of this, I would not describe Atlanta’s 25% SOT conversion in this game as wasteful or as evidence of “sloppiness in the final third.” I would say the team were mostly lights out efficient in turning attacks into threats and threats into huge threats. Jesse Gonzalez stood on his head some, but at the same time, even while putting on massive shots and xG numbers, Atlanta still bested the 2.8 expected goals with 3 actual goals. I would say finishing was not a problem in this one.

That all being said, while it’s not surprising to hear very bullish talk of the team challenging for a top 2 spot in the east, along with talk that the wider and deeper playing surface has offered us a glimpse at a new normal for Atlanta where they will dominate every away side with pace and tenacity, (inhales) I expect we might see the team be slightly less efficient going forward at turning attacks into shots, slightly less efficient at turning shots into shots on target, and slightly more efficient at turning shots on target into goals (as has been the trend so far this year). I don’t feel too strongly about it. I’m just hesitant to think that what we saw against FC Dallas is what we’ll see going forward because of a bigger field and a slicker surface.

Pressing & Tidiness

Pressure FCD Table.PNG

These pressing numbers seem fairly standard for Atlanta with the exception that Dallas was slightly tidier than a typical Atlanta opponent. This may have been impacted by the severe game states of Atlanta leading for most of the game and increasing its lead as the game went on. Also both teams played more own-half passes than I’ve been seeing in these numbers. Atlanta’s 225 higher than its season average of 195, and FC Dallas’ 163 higher than Atlanta’s opponents’ average of 126. Perhaps this is the bigger pitch showing up, hard to tell. I thought Atlanta’s press was somewhat more ruthless than these numbers show with several balls won back quickly after a turnover, and quickly converted into through balls and other transition moves.

Atlanta’s defensive actions in the first half (left) and second half (right):

This is somewhat surprising given what I thought I observed which was a press that slowed somewhat in the second. I think in the second what you’re seeing is more counter-pressing — pressing to win the ball shortly after losing it to keep Dallas from countering.

Shape

We saw a return to the more pronounced back-3 shape with Larentowicz eager to join the other centre backs when in possession (LGP and Parkhurst in the left graphic, Jeff in the right).

Asad and Almiron seemed to connect over and over again with great effectiveness on the left. I can’t tell just yet if we should credit the larger pitch or not. It very well could have been the case that they each felt more comfortable on the ball with the defense stretched both laterally and vertically by the speed of Martinez and Villalba (and also Almiron). Jason Poon suggested that Atlanta and Almiron specifically targeted Dallas’ right back Grana. And that might very well have been the case. On the left below is the chalkboard for Garza + Asad, and on the right the map for Almiron, who does seem awfully left-leaning for a central playmaker. Asad drifts inward some but no more than usual. Anyhow, the key passes coming in from that side from these 3 players alone is … it’s high.

Walkes and VillalbaOn the right side, there has been some criticism recently that the Garza/Asad dynamic of the fullback stretching the defense wide and the attacking midfielder tucking in wasn’t being replicated to the same degree of success. In this game however, I noticed Villalba tucking in often nicely into pockets between the lines to receive the ball. We don’t see the same degree of advancement from Walkes that we might see from Garza game to game, but the variety in attacking width on the right seemed to work well enough.

 

 

One last look at the attacking passes for each team. You can really see a contrast in style between the two. The first half below, Atlanta on the left and Dallas on the right:

FC Dallas obsessed with width. Atlanta just killing them from the most dangerous areas. In the second, it’s a little bit more varied as Dallas chased the game.

I’ll have to cut this one off now. Too many games.

In short, this game was a lot of fun. It’s possible the larger pitch is super-charging the team, but I want more evidence before we call that one. Atlanta was extremely efficient at turning attacks into shots, but Gonzalez was also a disruption. Guzan was very good. Atlanta good. Dallas bad.

 

 

Atlanta United and Finishing Chances

Finishing. This is not the fist time I’ve approached this topic on this here blog. But as we drive deeper into the season, it’s an increasingly difficult topic to ignore for Atlanta. While its easy to lament missed chances and their impact on points dropped in the table (think of DC home match), the truth is that Atlanta United are a statistical outlier, finishing an exceptionally high percentage of their chances relative to the rest of the league, and outperforming their expected goals by the widest margins we’ve seen in MLS. It begs the question…

When it comes to converting shots into goals, are Atlanta luckier than most or better than most? Stick around. I have charts!

Background:

Evidence has shown that in soccer, consistently creating good chances and limiting your opponent’s chances is a key success factor. The rate at which your chances are converted into goals may come and go, but whatever it is that you do that creates more and better chances and limits the quantity and quality of your opponent’s chances is the secret recipe, the valuable IP that makes a team good. In fact when trying to predict a team’s future performance, the number and quality of its past goal scoring chances compared to those of its opponents is more often a better predictor than its past goal differential.

And it’s because of this that when you see a team with an exceptionally high shot conversion rate, you are supposed to be worried that the team will perform worse in the future than it has in the past — that its past results are due in large part to something that won’t be repeated in the future (chance). This fear could be comforted if the team were taking higher probability shots than the rest of the field (the sort of thing you might measure by looking at the team’s expected goals per shot – EDIT: STOP WHAT YOU’RE DOING AND WATCH THIS VIDEO), but generally speaking high conversion percentages are red flags.

Atlanta United’s historically high shot conversion and exp. goals over-performance

Atlanta United has an exceptionally high shot conversion rate. And it has achieved this all season long. Atlanta are on top of almost every flavor of shot conversion chart (shots, shots on target, open play shots, etc). As an example, immediately below is the shot on target conversion metric through the 21 games or so in 2017:SOTConv

And here are the shots on target conversion rates for all teams from 2012 – 2017 YTD:SOT conversion

So it’s not just the highest SOT conversion rate in MLS at the moment, but it is historically high (only 2014 Dallas and 2013 Red Bulls top Atlanta’s 2017 figure), although curiously a few teams this year are also very high . Does Atlanta have a high shot quality per shot (xG/shot) number to prop up this high conversion percentage? No, they don’t. It’s middling at best. See a helpful graphic of MLS shot quality and volumes from AmericanSoccerAnalysis here. Another way of looking at this is to look at Atlanta’s goals scored compared to their expected goals. Atlanta are averaging 1.9 goals per game against an expected goals per game figure of 1.3. They’ve scored 15 more goals (41 vs 26) than MLS teams historically have based on the volume and location and other attributes of the shots they have taken (based on publicly available data). And again to prove the point, this will look familiar, but here’s how Atlanta looks historically in terms of their goals scored above expected goals per game (it is the highest recorded):G-xG all years

In Search of an Answer

So what do we say about this? As we’ve discussed before on the blog, the possibilities seem to be as follows, and a combination of these is likely:

  1. Atlanta are luckier than most.
  2. Atlanta are more skilled at shooting than most.
  3. Atlanta are taking better shots than most, which is propping up the high conversion number; however, the models we have for measuring shot quality (expected goals) are struggling to properly value the particular shots Atlanta are taking.

Shot Accuracy (also historically high)

Quickly, on #2: Skill, I’d like to point out that not only are Atlanta putting a higher percentage of their on-target shots past the goalkeeper, they are also putting a higher percentage of their shots … on target — testing the keeper at a higher rate than anyone else. And while, like shot conversion percentages, this isn’t a perfectly repeatable stat (repeatability suggests skill is involved), it is more repeatable than conversion. First, here’s the shot accuracy of 2017 MLS teams (shots on target / shots). Atlanta kills it.ShotAccSome pretty good teams at the top, but some also dispersed throughout if I’m honest. And below are the shot accuracy rates by team from 2012 to 2017 YTD:Historic Acc

Once again, that’s historically high. Only the 2015 Sounders put a higher percentage of their shots on target. And again curiously, another 2017 team is right there with Atlanta, this time the Fire.

I’m tempted to suggest this very high accuracy number is evidence of either #2 (shooting skill) or #3 (Atlanta’s chances being better than the models suggest). I can’t really prove that first idea but it seems plausible. The second one — I don’t know — I think I remember something about shots on target including some bit of embedded information about the quality of the chances (since its easier to put a high quality chance on target than a low quality one). To my eye, watching the team, it doesn’t feel like they are lucking into a high percentage of shots on target. The team really seems to prefer to not shoot until they’ve set up an open look. Many good opportunities end before a shot is generated as the team works to find a better shot by passing or dribbling — this sounds a little bit like *gaming xG* but I can’t be too sure.

Shot Openness: Dissecting all the goals

I haven’t looked at every single shot taken this year by the Five Stripes, but if we look at all the goals (yep all of them, many of them here: thanks Whitecaps), a very high percentage of these are 1v1 against the keeper (and several are empty net opportunities), or there are very few opposition outfield players between the shooter and the keeper. This type of “openness” of a shot has a big impact on its chance conversion, but most expected goals models can only “guess” at the the openness of a shot based on other event data context. They don’t “see” the openness, but they might see that the shot was assisted by a through ball or that the buildup was defined as a “fast break.” I tried to record a bunch of admittedly judgmental shot quality attributes for all these goals. See below for a summary of what I compiled watching every goal, looking out for certain “openness” attributes and then I’ll compare it to another team:

Shots through 22

I haven’t done this for all MLS teams because of time, but what jumps out to me are the 26 1v1 chances, and the 35 chances with 1 defender or fewer between the shooter and the keeper/goal. My gut is that these figures are high compared to the league average (which I do not know). For comparisons sake, I picked another team, New York Red Bulls and gave them the similar treatment of watching every goal scored and capturing “openness data.” I chose them for two reasons, first they are currently running neck and neck with Atlanta for the 4th seed in the playoffs (and a first round home match), and secondly (and more importantly), they are an example of a very good team that’s not crushing its expected goal numbers like Atlanta is. In fact they are slightly under-performing. So I was interested to see if there’s a significant difference in the composition of the goals scored. There is.

NYRB Goals

Across the board NYRB’s goals feature slightly fewer “openness” attributes, except they do a great job of finding BWP in the 6 yard box. If we compare them side by side it’s somewhat easier to see.

ATLNYRB

Is this the difference between a team overperforming its expected goals by historic margins (+0.6/game) and a team slightly underperforming its expected goals (-0.11/game)? Here are the distributions of defenders in between the shooter and the keeper for Atlanta vs Red Bulls.

 

Again, we can feel confident it’s either 1) chance/luck and Atlanta really aren’t as good in attack as the results show, 2) Shooting skill with Atlanta being better at finishing than most other teams, or 3) the quality of Atlanta’s chances (primarily related to their “openness”) not being picked up in the event data based expected goals models.

I would suggest it is a mixture of all three with the above “goal quality attributes” hinting gently towards #3. And perhaps Atlanta’s three designated players being signed into the attacking front 4 hinting gently towards #2.

Let me know your thoughts. What am I missing?

I should note that watching all of a team’s goals is a fairly weak attempt at trying to identify other shooting attributes that might be contributing to outliers in shooting efficiency and that looking at every single shot would be a much better exercise (and even still a flawed one). But I don’t know how to get the footage and honestly I don’t know if I could do it. This is at least something. I think it’s better than throwing one’s hands up and saying “welp, Atlanta are lucky” or “welp, Atlanta are master finishers.”

Expected goals data comes from AmericanSoccerAnalysis. If you read this regularly you probably know that by now.

Post credits Scene hinting at more shared universe titles to come

While I was digging through finishing metrics (primarily goals minus expected goals figures), I noticed something interesting that I’m not sure how to interpret. There’s reason to believe its further proof that there’s significant “openness” of shots not being appropriately valued by expected goals models. Basically, if you look at G-xG distributions between home and away matches you see notable differences. The home distributions stretch out more positively with large totals of significant overperformance games showing up whereas the away matches look more like a normal distribution but tilt slightly in the direction of under-performance.

 

 

If you take all these distributions as a percentage of the total and chart the differences in percentage share by bin between home and away you get the below image, which more clearly shows the lean towards an over performance in finishing for the home team relative to the away team.2017 home away bins 10

 

This is 2017 data for all MLS teams over 22 or so weeks of the MLS season, so it’s a good number of games and it runs the full spectrum of good and bad teams. It would suggest that either 1) teams shoot better at home, or 2) when playing at home teams get better chances than the models are able to capture in xG. The first should be dubious to anyone that thinks finishing generally averages out across all shots. The second one makes more sense with what we know about home field advantage in MLS, that it definitely exists. It doesn’t seem too wild to think that teams create better chances at home, and perhaps even chances that are better than the expected goals models give them credit for. But that’s just 2017 data. What does the 2011-2017 data show? Well, it’s both a) much less pronounced (smaller differences between bins between home and away), and also b) home teams seem to both significantly over-perform and significantly under-perform more often than away teams. 2011-2017 home away bins 10

So that leads us back to something strange going on in 2017, which could very well be a skew driven by Atlanta United, or a shift in the landscape of finishing for MLS in 2017, or an increase in the types of good chances that are undervalued by models? Or… variance working its way out over the course of a full season.

Appendix: Assorted resources on finishing skill in football

https://cartilagefreecaptain.sbnation.com/2014/4/25/5652640/player-finishing-skill-is-real

http://www.optasportspro.com/about/optapro-blog/posts/2014/on-the-topic-of-expected-goals-and-the-repeatability-of-finishing-skill.aspx

http://statsbomb.com/2014/03/thinking-about-finishing-skill/

http://www.optasportspro.com/about/optapro-blog/posts/2017/blog-re-examining-finishing-skill/

http://statsbomb.com/2017/07/quantifying-finishing-skill/

http://www.americansocceranalysis.com/home/2016/4/4/does-finishing-skill-matter-in-mls?rq=finishing

http://www.americansocceranalysis.com/home/2016/10/18/is-finishing-real-heading-towards-a-conclusion?rq=finishing

There are many others out there, plus I’ve discussed it occasionally in ATL match previews and recaps on this blog.