Yesterday, I received a fantastic gift. For years, I've been trying to get the various outlets that I've written for to give me a data dump on the fantasy season that just concluded. I'm not talking about the basics like total points and matches/minutes played (which are a great place to start) but the sorts of splits - home vs road and vs top teams vs relegation fodder - that we always assume work in specific ways. I'll probably be finding different ways to look at this data for most of the summer but as I get started, here are some initial thoughts as you start reading news about new players arriving in the Premier League (welcome Naby Keita and Fabinho) and start evaluating how attractive they might be in draft formats. As it turns out, the same logic also applies to returning players.
There are two questions related to opportunity that are important for you to consider about a player. Those two questions are "How many minutes are they likely to get?" and "How productive has that position traditionally been?" Sometimes the answer is simple, sometimes not so much.
How Many Minutes
The logic is pretty obvious. It doesn't matter how good a player is in a vacuum, production can only come with minutes. The difference between a player you draft with the intention of keeping them vs. someone you pick up for a week or two when injury, suspension, or rotation dictates it is likely they are going to start is all about the opportunity to rack up big minutes.
Eden Hazard and Willian accumulate fantasy points in the Premier League's fantasy format at about the same rate per minute (0.071/minute for Hazard and 0.069 for Willian). The difference? Hazard racked up 2417 minutes over 27 starts and 7 substitute appearances while Willian only managed 1874 minutes over 19 starts and 17 substitute appearances. The disparity would have been even greater had Hazard not been recovering from an injury when the season started. Add in the details of the scoring system that rewards playing 60+ minutes twice as highly as merely appearing in the match at all and you can see the value of opportunity over and above merely presenting more chances to score/assist/accumulate bonus points.
There are only 990 minutes per match per team to divide up and 38 Premier League matches per season so when looking at any player you might consider drafting you have to consider how many of those minutes a player might play as much as how objectively "good" they are.
Riyad Mahrez's rumored move to Manchester City doesn't change his quality as a player, what it does do is put him in competition with an incredibly talented and deep group of wide attacking players for minutes. Last season he played 2948 minutes despite missing a couple of matches due to being unfocused after his January move fell through. How many Manchester City midfielders do you think played that many minutes in City's title-winning season? Well, Kevin De Bruyne did (he played 3073). That's it, that's the list. The genius playmaker David Silva who some consider City's most important attacking player only managed 2431 minutes while Raheem Sterling played 2584, Leroy Sane played 2415, and Bernardo Silva played 1514. Oh, and Gabriel Jesus (1660) and Sergio Aguero (1960) were closer to playing half the Premier League minutes available to them than the full 3420 that represents the maximum a player could play across 38 matches.
Want to know how many of those minutes will be "up for grabs" because a player is known to be either injured or leaving the club? Exactly zero based on what we know now. So, without getting any worse as a player, Riyad Mahrez has gone from playing almost every minute available to having to fight it out with the upside of his expectation being what Sterling/Sane did last season at around 2400 minutes and the more likely scenario looking like Bernardo Silva's 1500 minutes.
It isn't just the number of minutes that a player is likely to play that helps dictate his likely fantasy value, context matters. 2500 minutes playing an attacking role at Manchester City has proven to be more productive than playing 3300 minutes for, say, Stoke City. Raheem Sterling (229 points in 2584 minutes) plays a similar attacking role to Xherdan Shaqiri but the Swiss international could only manage 155 points in 3039 minutes despite his season being a pleasant surprise in terms of both productivity and health. Some of that is down to Sterling's quality but the rest is down to the exceptional City attack compared to the anemic one at the Bet365 Stadium.
The occasional Gylfi Sigurdsson at Swansea City can overcome the context of the club that they are at to become a truly exceptional fantasy performers by contributing to almost every point-making action for their team rather than being on a team that provides a lot of opportunities overall. Pascal Gross and Wilfried Zaha are current examples of players rising above their club's station but they are rare and hard to pick out ahead of time in most cases as Gross was well behind Knockaert, Ince, Mooy, and likely others as we tried to pick the value coming up with the newly promoted teams.
The thing that looking at the context of the expectations for Opportunity (minutes played) and Context (historical production from a team at a given position) won't get you is the outlier. If you looked at Adam Lallana's production under Jurgen Klopp at Liverpool and projected Mo Salah as a full-time, healthy starter in that spot you wouldn't have predicted anything close to what happened. Even if Salah had only produced a peak Theo Walcott season in his debut campaign at Liverpool (that's 14 goals/10 assists if you're wondering and that's spectacular for a wide attacker) we would have underestimated him by using Lallana projected over a full season as the point of comparison.
This is where you have to exercise your brain rather than a spreadsheet to make your predictions. What teams are going to make a significant leap forward and create additional opportunity for a number of their players to produce unexpectedly high fantasy point totals? What players are going to raise the level of the position they play at a given club in the way Romelu Lukaku did when he was at West Brom and, to a slightly lesser extent, Everton?
Very few clubs outside of the top few have proven adept at replacing star players. Sam Clucas "replaced" Gylfi Sigurdsson but couldn't come close to approximating his impact. Likewise the earlier vintage Yohan Cabaye at Newcastle. Lanzini seems to have the talent to do an adequte job of replacing Dimitri Payet but he couldn't put an entire season together like Payet did before shoving off for Ligue Un.
When you're drafting a team, you can take a couple of swings hoping that a Gross, Lanzini, or Mooy is going to be that next Sigurdsson or Lukaku producing big club results at a small club. What you're much better off doing is assuming that the players who come into a reliable system like Burnley's will produce at a similar level to the people they are replacing. Nick Pope and James Tarkowski and Chris Wood were far easier to predict than Gross maintaining his Bundesliga success in the Premier League or Marcos Alonso coming to Chelsea and producing goals as well as the expected clean sheets.
This is just the kick off to a summer of looking at some of the fundamentals of evaluating players for the draft fantasy format.
One parting thought at the World Cup approaches and we get ready to be giddy about a bunch of players we've never heard of. Of the 32 teams that will compete, only about a quarter of that total feature the depth in quality to compete in a league like the Premier League. When a no-name forward scores an unexpected hat trick against Senegal or Croatia or Iran we shouldn't assume that means that they can do the business against Manchester City or Everton or even Wolves. The World Cup is a fantastic event and we should all enjoy it but it is a rotten opportunity to evaluate talent due to widely disparate talent and small sample sizes. Your club may do some great PR business by signing a World Cup star but they are unlikely to help the club progress.
More to come as I get into the big data set sitting on my desktop.