As part of our 2018 season kick-off, I wanted to collect the best of LacrosseReference’s articles from last year in one place.  Consider this the cliff-notes for new readers.

Lacrosse Talent is Heading West (and South)

This analysis shed some light on changes in the number of players coming from traditional hotbeds vs the newer lacrosse hot spots.  It was actually the 2nd post in a series of posts that we did about demographic and geographic changes in the player pool.

The central crux was just what the title would suggest: more and more players are coming from the western US and the southern states.  And there are some pretty maps to look at too.

Albany has developed overlooked talent better than anyone

Albany was probably the most electric team in Division 1 last year.  They played one of the best games of 2017 against the Terps during the regular season, they had (and have) a Tewaaraton finalist in Connor Fields, and they enter 2018 as the darling of Lax-ELO.

But focusing on last year ignores the rest of the Great Danes’ illustrious recent past.  They’ve had extreme star power with the Thompsons and now Fields, but the story that we highlighted in this post is all about how much they’ve done with a relatively under-appreciated set of players.

Top Individual Performances of the Year (so far…)

Sadly for all fans of college lacrosse, Dylan Molloy graduated.  The same goes for dozens of top senior stars from last year.  But no one can say that we didn’t appreciate them while they were here.  This post was the first in what became a weekly feature highlighting some of the top individual statistical performances.  Spoiler: Molloy’s 8.6 expected goals added against Holy Cross did not hold up as the top performance of the year.

This post also highlights one of the central challenges of lacrosse analytics.  If you check out the leaderboard as well as the subsequent weeks, you’ll notice lots and lots of offensive players.  We are not saying that defenders did not put in some of the best performances during these weeks, but we simply don’t have enough counting stats recorded consistently to encapsulate their contributions.

This is one of the things I’d like to try and resolve this year.

Johns Hopkins Part 2: Turning into a pumpkin after 60 seconds

I’ll admit that our work from last season was primarily about quantifying what happened.  We didn’t do a whole lot that really asked deeper questions about why teams were or were not successful.

This post was one of the early exceptions.  We looked at all the possessions that the Blue Jays had on offense and split them up into categories based on the number of seconds between possession being gained and a first shot being taken.  The intent was to try and assess the effect of pacing on the Hopkins offense.

What we found was that Hopkins was a much more effective offense when they attacked (i.e. had a first shot) within 60 seconds of gaining possession.  When they didn’t, their efficiency dropped fairly precipitously relative to the other top offenses.  I also added some very speculative analysis that perhaps the Hopkins staff was intentionally forcing a slowdown on offense and that when the offense was artificially slowed down, it suffered.

I’ll leave the x’s and o’s to the experts for the most part, but this was a neat analysis because it let us peek under the covers and move beyond the basic performance-based counting and rate stats.

Predicting 4-year value from Freshman stats

After Championship Weekend last year, we took some time to step away from the on-field stats and analysis and focus a bit more broadly on what it means to be good in college lacrosse.

As part of that work, we needed a way to normalize stats for underclassmen so that we could compare them apples to apples with the graduating seniors.  Obviously, freshmen have fewer games worth of stats to look at, so we needed some way to estimate a freshman’s career value from a single season.  But you can’t just multiply by 4 because freshmen play a different role in most cases.

So, we built a model that attempts to give a best guess of career contributions using only freshman year stats.  Pioneers fans rejoice, because it projected Ethan Walker to have the most productive career of any of last year’s freshmen.  Of course, three years is a long time, and things will change, but it’s interesting to see whose freshman seasons project out best.

MLL & NCAA teams have overlapping Twitter followings, but not uniformly

Most of our work last year was focused on wins or losses, efficiency rates, and leaderboards.  But as I said above, the end of the season gave me a chance to step back, take a bigger picture view, and do some off-field stuff as well.

This post was our first foray into the world of social media.  We grabbed all the followers for every college, MLL, and NLL team using the Twitter APIs.  Those lists were the basis for an analysis on social media followings and more importantly, social media overlap.  The idea was to compare the college and MLL followings for each pair of teams that share a geographic footprint.

In theory, the more that two teams share followers, the better they are doing in taking advantage of the local fan base.  Since MLL is less popular than college, I wouldn’t argue if you framed it as an analysis of which MLL teams had the best social strategy (although that might be putting too much stock in follower lists).

There were a few surprises, but none more than the fact that only 16% of the twitter accounts that follow Maryland lacrosse also follow the Chesapeake Bayhawks.  For contrast, 19% of the accounts that follow the Denver Pioneers follow the Chesapeake Bayhawks.  It would seem that there is an awful lot of local lacrosse excitement that the Bayhawks are not capitalizing on.

Estimating the goal value of each type of play

This was one of our first posts, but it describes a very foundational piece of this site.  Our player comparisons typically rely on a metric called “expected goals added”.  The idea is that every play that ends up in the play book has a positive or negative effect on a team’s likelihood of winning.  By looking at every play recorded and determining whether a goal was scored in the following sequence, we can estimate the “value” of each play type.

(If there are 100 ground balls picked up, and the player’s team scores a goal shotly thereafter 18 times, then we would say that a ground ball is worth .18 expected goals.  But don’t get too hung up on the exact definition; it’s more of a relative thing between plays.)

This post introduced the idea, shared some stats, and generally laid the foundation for a lot of our later stuff.

Introducing ELO for College Lacrosse

For those not familiar, ELO is a very common way to compare the relative strength of two teams.  It was initially developed for chess, but has since been implemented in lots of different systems for various sports.

As far as I know, this post describes the first ever implementation of ELO for lacrosse, hence the name Lax-ELO.  (Creative, right?)

Notre Dame has been riding its stars so far

You may have watched Denver put Notre Dame out of their misery in last year’s NCAA tournament.  I’ll be very generous to myself and say that if we didn’t call that, we at least identified an early season trend that was very much in line with that result.

Early on, we developed a metric called “play shares” which attempts to quantify whether a team has a deep bench of contributors or whether they rely heavily on their top line stars.  Play shares relates to the percentage of a team’s recorded plays that each player makes.  It’s a bit different than value because a player can have high play shares while making a bunch of terrible plays.

In theory, the distribution of play shares won’t tell you much about how good they are, but it should tell you something about depth.  The hypothesis that I laid out at the time was that a team with a flatter distribution (i.e. deep bench) is less likely to wear down.  A team with an uneven distribution may start strong, but is more likely to fade down the stretch.  Coaches have a decision to make between winning games and saving their players for the stretch run.

This analysis posited that Notre Dame was relying too much on their stars at the expense of depth and rest.

How to get a job in sports analytics

This was the least lacrosse-relevant post of last year.  It was also the one that got the most attention.  The article laid out my three top tips for someone who is interested in getting into the sports analytics world.

Of course, the reason that it was more popular is that it was targeted at a much larger population than the rest of my content.  And since it was the least lacrosse-relevant, it is getting stuck at the end of the list

If you found this interesting/helpful…

There is a link in the upper right hand corner of this page, which will allow you to enter your email address and subscribe to my newsletter.  I send out something about once a month during the season with extra perspective, content, and links.  If you enjoy this kind of stuff, you’ll get a bit more benefit from signing up for the newsletter too.