Scatterplot showing all players drafted in first two years of the draft. Draft order is on the x-axis, and minutes played by the player is on the y-axis. Players' dots are also colour-coded by the team on which they played. There is no obvious pattern to the data - lots of zeros at most draft positions, and quite a few draftees playing high minutes.

CPL-U Sports Draft Part 2: Does Draft Position Matter?

“The best players in the CPL for the money, for the quality, and for all those variables considered, have come out of U Sports.”

Carmine Isacco, Former York9 Assistant Coach (source: The Northern Tribune).

About the series

This is the second of a five-part series analyzing the Canadian Premier League-U Sports Draft.  The draft, like the league itself, is young.  But even with such a small sample size, some trends are evident-ish.  And numbers are fun.  The five parts are:

  1. What is the CPL-U Sports Draft?
  2. Does draft position matter? *YOU ARE HERE*
  3. Which clubs succeed at the draft (and which one fails)?
  4. Which player positions get drafted (and which ones succeed)?
  5. How can the draft be improved? 7 ideas from people smarter than me

How does draft position relate to value in other leagues?

Many pro sports leagues in North America have a draft. And in all of them, of course higher-drafted players should be better than those drafted later.  But how much better?

In some drafts, there’s a steep drop-off in player quality from the first few players picked to everyone else.  Like the NHL’s draft (see graph below).

Scatterplot of players' value on the y axis versus draft pick order on the x axis. Player value is defined as game score value added over the first seven seasons. There is a steep drop off in value from the first few picks to everyone else.
In the NHL draft, the highest-drafted players (left side of graph) are worth way more than later-drafted players (right side). Based on drafts from 2000-2019. Used with permission from Dom Luszczyszyn who published this graph in this article for The Athletic. Visit that article for details on how he calculated player value (game score value added over first seven seasons).

Other leagues’ drafts can have less of a drop-off between early- and late-drafted players.  Like in MLS (see graph below).

Scatterplot showing MLS draftees, with each player as a dot. Draft position is on the x-axis, with higher-drafted players on the right side. Minutes played is on the y-axis. There is a pattern of higher-drafted players playing more minutes than lower-drafted players.
In the MLS SuperDraft, higher-drafted players (on the right side in this graph – sorry it’s opposite of the first graph) tend to play more minutes than later-drafted players. But the drop-off isn’t steep. Used with permission from Scott Knuth who published this graph in this post on American Soccer Analysis. See that article for details on how the y axis value (percent of minutes played) was calculated.

Ok just give me the result for CPL

So, for the CPL-U Sports Draft, do players drafted early play more minutes than those drafted later:

  1. By a lot, like in the NHL?
  2. By a little, like in MLS?
    or
  3. NOPE. Like in CPL.

Scatterplot showing all players drafted in first two years of the draft. Draft order is on the x-axis, and minutes played by the player is on the y-axis. Players' dots are also colour-coded by the team on which they played. There is no obvious pattern to the data - lots of zeros at most draft positions, and quite a few draftees playing high minutes.
In the CPL-U Sports Draft, early-drafted players (on left side) tend to play about the same minutes as later-drafted players (right side). The y-axis is CPL minutes played in a draftee’s post-draft season, for every player drafted in 2018 and 2019.
For the statistically-inclined, the correlation is Pearson’s r = -0.07, p = 0.68 (i.e., no relationship). Minutes played data are freely available from the league’s Centre Circle Data.

Wut?

If you’re looking at that graph and seeing a pretty random scatter of dots, that’s exactly the point. Higher-drafted players do not play more minutes than lower-drafted players.  That’s weird.

So the next question, after wut, is why.

Why don’t higher-drafted players play more minutes than lower-drafted players?

There are data reasons.  And pandemic reasons.  And some soccer reasons too.

Data reasons for the non-relationship between draft order and minutes played

  1. Small sample size – just two seasons
    There are only one and a sort-of-half seasons of data to go on.  Weird stuff happens with small sample sizes. For example, injuries and life events happening to a few higher-drafted players can skew things.

  2. And the second season was barely a season
    The Island Games were great, but they were only a small number of games. So even if a draftee played in most of their team’s games at the Island Games, their total minutes would be only a fraction of those played by 2018-drafted starters.  By combining 2018- and 2019-drafted players in one graph, we’re not comparing them fairly.

    I considered using some measure of the percent of a team’s possible minutes a draftee played in. e.g., Cory Bent was drafted first overall in 2019 and played quite a bit in the Island Games – 500 of the team’s 900 possible minutes (56% of possible minutes).  But I didn’t do this, for a couple of reasons. First, so few draftees played at all in the Island Games that it probably wouldn’t matter; weighting zero minutes by anything is still zero minutes.  Second, it still wouldn’t fully account for possible playing minutes, because some 2018-drafted players signed Developmental Contracts and could only play until mid-August. I don’t have access to all draftees’ contract statuses, so I can’t correct for this. So both years’ drafts are combined here, and that contributes noise/messiness to the relationship.

Pandemic reason for the non-relationship between draft order and minutes played

  1. Draftees never got a chance in the Island Games
    Most teams did not play their draftees at the 2020 Island Games. Like, at all. Most teams gave their draftees zero minutes of playing time.

    The 2020 season had 1/3rd as many games as 2019. But its draftees played a mere 1/15th as many minutes as did draftees in 2019. Only 4 of 14 draftees played at all in 2020. Take a bow, HFX Wanderers, for playing both of their draftees more than any other team played a draftee (Cory Bent with the 1st pick and Jake Ruby with the last pick).

    Why didn’t teams play draftees at the Island Games?  I don’t know.  Maybe some draftees couldn’t play due to pandemic restrictions impacting their lives (e.g., Thomas Gardner apparently couldn’t play for Pacific FC due to travel restrictions).  Maybe coaches were less likely to play new players in a shortened season, not wanting to risk “rookie mistakes” when every game matters.  For whatever reason, most draftees didn’t play at all in 2020 regardless of their draft position – that helps produce a non-relationship between draft position and minutes overall.

    For stats nerds, I ran a General Linear Model to see which of the following factors predict minutes played by draftees:
    • Year (2019 or 2020)
    • Draft position
    • Team played for
    • Playing position (defender, midfielder, forward, or goalkeeper)

The only factor that came out as significant was Year. 2020 sucked for draftees (and the world).

Soccer reasons for the non-relationship between draft order and minutes played

I suspect none of the following are as influential as the first three reasons covered.

  1. Team effects
    Some teams play draft picks a lot, and some barely play their draftees at all. These team effects might detract from an overall draft position pattern. But, as revealed in the statistical model described above, team effects are overwhelmed by the impact of 2020 being a bad year for draftees.

    I’ll get more into team effects in the next article in this series. Forge fans might want to avoid it.

  2. The league is new and scouting is hard
    Other leagues (NHL, MLS, etc.) have spent years developing scouting networks and analytical approaches to draft the best players first. CPL, on the other hand, is new and might not rate U Sports players as reliably. Player evaluation is tough, especially for U Sports with no advanced stats readily available. That’s not a criticism of the clubs’ and league’s scouts – it’s just a big new job with limited resources.

  3. Location, location, location
    Many teams preferentially draft local players. There are good reasons for this, which I’ll cover in the next story in this series. But for the purposes of this analysis, it adds noise to the relationship by not fully aligning player quality with draft order. Good players might slide down the draft order because of where they live.

Conclusion

I expect that as more (non-pandemic) seasons happen, a relationship will emerge where players drafted earlier play more minutes than those drafted later. I was surprised that wasn’t the case already. But at present the relationship is weak to non-existent.

So if you’re Jackson Farmer, and you’ve been drafted with the final pick of the 2021 draft, you can be optimistic that you have about as much chance of playing big minutes this season as the guy who was drafted first overall.


Next in the series

Some clubs play their draftees a lot and have found gems (HFX). Some clubs are allergic to draftees (Forge). And then there’s FC Edmonton’s obsession with drafting local players, which I’ll cover in detail.

  1. What is the CPL-U Sports Draft?
  2. Does draft position matter? *YOU ARE HERE*
  3. Which clubs succeed at the draft (and which one fails)?
  4. Which player positions get drafted (and which ones succeed)?
  5. How can the draft be improved? 7 ideas from people smarter than me


Let me know your thoughts on the article @JayFitzSoccer.