SSL Fantasy - Mock Drafting - The Finale

[AustinP0027]

SSL Fantasy Series:
SSL Fantasy - An Intro
SSL Fantasy - Rosters
SSL Fantasy - Scoring
SSL Fantasy - Roster Build

Note: Any data for the current season is two weeks old at this point as that's when I wrote this article.

We’ve done it. We’ve walked through what fantasy would look like, created roster positions, nailed down the scoring, and tested out the potential scoring against roster positions to land on a decent setup for our fantasy league.

Now it’s time to do the fun part, draft some actual rosters and see how they performed!

The setup:
To try and make this seem real, I’ve crafted 5 personas who will be drafting from the pool of players. They all have different motivations/styles for drafting, and obviously they will be limited in those by other managers drafting players out of the pool. It’s important that you know that these are completely made up people, who in no way exist, and any similarities to real people are made up in your mind. Quit it. Quit thinking about it.

---

Persona 1: Proque
Proque is picking Best Player Available, regardless of roster slot. Think of Proque as autodraft. They’ll fill every roster slot, but if the best player available is a Defending Back twice in a row, they are picking them twice in a row even if they have other unfilled spots.

Persona 2: CamSilverGun
CamSilverGun is worried about position scarcity. That means they’ll be focused on drafting positions in order of how valuable/scare that position is. For example, even though Attacking Midfielders score the most point, CamSilverGun recognizes that you pick 3 of them and instead focuses on positions you choose fewer of to try and get the most elite choice.

Persona 3: Defino
Defino is focused on Tier Cliffs. Meaning, Defino doesn’t want to be stuck drafting a tier down at a position because they were stuck at the end of a position run. Defino is more likely to start the position run if they realize that there are fewer elite players left in a position they still need.

Persona 4: Summersat
Summersat is focused on balance. They want the best player available, but they refuse to overload one position, so they spread their selections out across the roster positions. Summersat will be seen grabbing the highest tier possible, but for positions where there are multiple roster spots, they’ll grab a different position even if a higher tier is still available for the position they just drafted.

Persona 5: McHumbermack
McHumbermack is the stats analysis god. They’re using previous season’s scoring and stats to try and aggressively draft players who have the highest ceiling. They ignore the safe weekly floor of players and instead push to fill a roster of players who have a chance week to week to put up really gaudy numbers. There’s a lot of risk to this approach as players may not perform as well season to season, or teams may shift them around and really hurt their production.

---

After creating the personas, I built the code to make them all draft. This way, it wasn’t me attempting to pick as if I were each person and potentially skewing the results, instead, it was based on formulas.

The code built, I put them into a snake draft, and had them pick players. However, a pretty big issue jumped out immediately with the first pass at a formula. It was heavily leaning into TPE for the “best player”. That works, right up until it doesn’t and you get a player who barely scored any fantasy points last season yet has TPE that fits them into the Impact category.

The next step after creating the persona logic, was to then try and create some base logic that every persona would use as a bit of a player projection. Meaning, in a regular fantasy draft, everyone would know which player was a huge disappointment last season and avoid them, so the personas need to do this as well. This was accomplished by applying some weighting. The model looks at last season’s points heavily, then looks at current TPE secondary. Finally, it applies a “bust penalty” if it finds high TPE but low points. This isn’t meant to be a perfect prediction, in fact, for the purposes of a mock fantasy draft I want the opposite. What it does do  is a good job of dropping the obvious players who wouldn’t be selected down so that the automated personas aren’t picking them.

The Results:
Now the part that’s the most fun. The draft is done, and we’ll talk about each team.

The Draft:
Round 1 - 4

Round 5 - 8

Round 9 - 11


Let’s look at each team individually.


Proque is our Best Player Available drafter. With their first pick, and first overall pick, they took Rigby Emerson, an AM with 1544 TPE. You’ll notice right away, this isn’t a “Superstar”. If Proque is drafting the best player available, why didn’t they grab one of the two Superstar AMs that are on the list?

The answer lies in the season 24 stats. Emerson was the highest fantasy scoring AM in season 24. That puts the projection score higher for Emerson than either of the Superstar players, who both were MUCH lower in fantasy points from season 24. Emerson projects as the top AM, so Proque selects them.

You see that pattern continue through the rest of Proque’s picks. Rashford II is selected at SK, picked after McHumberback grabbed a SK, but is the highest projected SK left. Same with the remaining picks. Proque went through each round and grabbed the player whose Projection score was the highest. Remember that Projection score is a combo of previous season stats plus TPE. Proque didn’t target any position, and instead just took the highest projection scoring player. You see that in the selection of everything but their final Mid taken before filling their AM spots, even though AM is a high scoring position.


CamSilverGun is our Scarcity BPA drafter. Cam focused on positions, drafting players for positions that had less choices left due to both the number of roster spots, as well as what other drafters had picked. With that focus, Cam behaved similarly to Proque, choosing the best player available based on projection score, but there was a heavy shift into positions.

Cam’s first pick was Alejandro Fuecoco, an Impact level AM. Again, this is a player who had a good season 24 in fantasy scoring, having the second highest fantasy score (behind Emerson who was selected 1OA).

Cam’s second pick was Thomas Jimotheus, an Elite level AM. At this point, the scarcity logic has kicked in, as 3 AMs have been taken before this pick. Now, the CamSilverGun persona is identifying AM as a potential problem position. They have 3 roster spots to fill, and 3 players have been taken before this pick. The third pick follows this exact logic, another AM, Santos Neymarinho, as the CamSilverGun persona is basically worried that they aren’t going to get a good AM anymore.

The rest of Cam’s picks follow the same logic, Midfielder gets picked next due to projection scoring, which triggers the persona to worry that they won’t get a good second one. They pick their Goalkeeper, since there is only 1 roster spot, but only 13 players available. Then they move down through DM, SK, and finally DB positions to fill out the roster.


Defino is up next, our Tier Cliff drafter. Tier Cliff looks at the positions to identify when they believe that the pool of highly ranked players at that position might run out, so they never get caught taking a lower ranked player.

Defino’s first pick is a Goalkeeper, Shingo Takechi. It’s an interesting choice, as no one else was looking at GK this early, but Defino has already identified that they want to get the best possible GK and grabbing early will prevent the pool from getting worse. Picking a Superstar GK falls right into this pattern.

That exact same logic is used for their second pick, as they grab a SK in Nerve De Groot. Teams only get one SK, and Defino wants to fill theirs quickly.

From there, Defino fills AMs, which makes sense as a number have been taken at this point and it’s the biggest roster spot to fill. Mids, DMs, then DBs are how Defino wraps up their draft, having identified DBs as their lowest impact spot so willing to let it be filled last.


We have Summersat in the 4th draft spot, who is our Balanced Best Pick Available drafter. Similar to Proque, they want to get the player with the highest projection. Unlike Proque, they are unwilling to pick the same position repeatedly if it can be avoided.

With the 4th overall pick, Summersat grabs Roquefort Cotswold, a Superstar DM. Roquefort has a high projection score due to very high TPE, and relatively high season 24 stats. I’m not sure I’d agree with this pick as a human observer, but Summersat is just following the logic they’ve been provided, so the projection score makes it the right selection.

Their second pick is another Superstar, Bernardo Fry at Mid. Again, projection score might be hurting Summersat here as they are emphasizing position balance over taking the very top players available. This is less a knock on the code, and more on the actual approach of Balanced BPA itself. In selecting with as much balance as you can, you’re actually sacrificing points.

Summersat continues down their roster needs, SK, DM, DB, and GK before finally selecting their first AM. Here’s where things get tricky for the balanced drafter, however. They have 3 AM spots to fill,and the quality is dropping as everyone else has been selecting AMs already. Summersat gets put in an awkward position as they now need 2 more AMs and only have 4 picks left. That leads to the first duplicate position pick as another AM is chosen next, before Summersat fills out their roster with DB, AM, then Mid.


Our final drafter is McHumbermack, who is our Ceiling drafter. McHumbermack has poured over the stats from the previous season, and has done their own modeling to try and find the players they think will have the highest possible offensive upside at each pick. The emphasis is on Goals, Assists, and then Shots on Goal because that means the player has the best chance each match of returning high points.

McHumbermack is the last draft, but that means they get the double pick at the turn. Their first collection of picks: Julian Rubio, a Superstar SK and Giacomo Santori, an Elite AM. Both players have a high amount of shots and shots on goal from season 24, so McHumberback is putting weight on them being able to score more in season 25.

McHumbermack leans heavily into the offense by finishing out their AM position spots with their next two picks, and exposes a possible issue with this approach. Dobra is a fine pick, they were towards the top of season 24 stats, but Miedema is all the way down around 50th for season 24 points. The emphasis on AMs and their offensive chances meant that McHumbermack had passed on other positions to grab another offensive position and it may not work out.

Next up is the DM position, followed by DB, Mid, then finally GK. Obviously McHumbermack doesn’t expect much out of defenders, but it’s interesting that Midfielders ranks that low for this persona.


How have they all done so far?
Now that everyone has a team, let’s take their team and apply all the stats and scoring to date in season 25.

Persona	Total Score
CamSilveGun	1513
Defino	1286.5
Summersat	1281.25
Proque	1200.5
McHumbermack	1186

Very interesting results. I had reported above that I had concerns about two strategies, Summersat’s balanced BPA forcing positions to not double up, and McHumbermack’s Ceiling drafting focusing on offense.

Well, one of those was a legit concern, as the Ceiling approach that McHumbermack used proved to have the worst results with this season’s data. The high risk/high reward style of grabbing offensive players first didn’t pan out. It seems that Summersat’s balance didn’t hurt them too badly, as it took 3rd place in the group.

Let's look at each team individually to see what happened.



Best Player Available no matter the position ended up putting Proque in 4th place. The problem with the approach? Inconsistent results. Rigby Emerson was the top scorer in season 24 with 204 points. This season, 29th overall, 13th just amongst AMs and a middling 118 points. That being Proque’s first pick really sunk their team. The rest of their picks weren’t too terrible, Hughes-Galloway was a steal at pick 41 and placed 4th overall. But the heavy reliance of last season's perception ended up costing Proque as too many of their picks ended up 30+ in final ranking.



Scarcity Best Player Available is your champion, easily beating the other personas. The problem with it as a real life choice…. Fantasy in SSL looks incredibly volatile! The top picked AMs (1OA and 2OA) both finished outside the top 25. Cam managed to grab Jimotheus who ranked 7th overall, but then the 3rd AM is still only 18th.  What actually stole the win was Cam managing to get some good luck in the positions they didn’t value at first. Having a Top 5 Midfielder (their second chosen Mid, no less), Striker, and Defensive Back don’t look to speak to skill, as they were all late picks, and others have taken those positions in front of Cam. Still, it’s clear that focusing only on the best player is not the right choice, position has to matter as well.



Our Tier Cliff drafter managed to take second place overall, even if it was a distant second. Defino grabbed the best GK to start their draft, unfortunately the best GK is only good enough for 17th place overall. The first 3 picks for Defino were fairly good but their second pick of AM went off the rails. Still, the 3rd AM pick righted the ship. Things slipped again, though, when drafting their second DM and onward. Their final five picks averaged 39th place overall. What could have been for Defino is they had managed to snag some of those positions earlier.



Balanced BPA had me worried as I watched it ignore AM until it’s 7th pick. Seeing it’s 1st pick go 30th overall in points didn’t help. Balanced didn’t have a major hole outside of a single AM pick that was 46th overall, but none of the picks really wowed either. Balanced BPA feels like the type of drafter who wants to make sure they don’t end up in last but also won’t ever end up in first because they don’t take the risks needed to put up big points.



I had higher hopes for the Ceiling drafter. Pulling stats of the previous season and trying to find the players who had the higher chances for offense really seemed like it might pay off. After all, there was a crystal ball type pick in Carolien Miedema as the 3rd AM who took 1st overall out of all players. The problem was most of the other picks. The DM and DB picks averaged 51st overall in scoring. The SK was only 25th (as McHumbermack’s first pick). Miedema really lifted the AM position, but the other two were outside the top 10 and were chosen before Miedema. Takahashi made an impact as a late Mid pick, but their counterpart on the roster in Jenkins was outside the top 50. Maybe most painful was the fact that GK Walrusson is the 2nd best GK, but still only 27th best player in scoring overall.


Overall, looking at the drafts and scoring results, it screams chaos. SSL Fantasy is going to be fun, but it’s going to be fun because you can’t predict what is going to happen. Maybe 4 or 5 seasons in someone will figure out what drags a player up or down from the previous season, but especially to start, it’s going to be a complete unknown of what players will actually help.

Well that’s it, that’s the entire series. I hope you’ve enjoyed the mind exercise of exploring SSL fantasy with me as much as I have using it to understand the league APIs and data better.

[norththegreatestshowman]

No Hohenhahn ?

[AustinP0027]

No Hohenhahn ?

TPE levels mark you as a Superstar. Unfortunately season 24 fantasy point total is only 76 so when they pick, that docks you enough to not be selected.

Based on stats at the time of pulling them for fantasy purposes, you ended up 9th for Strikers in season 25.

[DivogoJota]

Andre Caetano 51st overall, it’s not about where you start it’s about where you end baby