As we end our most recent season and start looking towards the upcoming draft, I realized I had let something slip past me. I never went back to revist my THICC rating system to compare it to the outcome of the actual draft, and see how it panned out in its first shaky iteration.
So I’ve gone back and reviewed the outcomes. Spoiler alert – don’t use this to mock the upcoming draft! However, while the THICC system’s limitations are evident in the early part of the draft, it appears to have merit the later in the draft you go. My interpretation of this is that the system does not take into account team needs, which is obviously extremely paramount in the first round, whereas the later you go in the draft the more teams are trying to find the best available/most likely to stick around players. This is much more in line with what the THICC system is shooting for. So that’s…encouraging?
Well without further ado, below is the outcome of the Season 26 draft, along with the THICC system selections. Below each round is the average difference between where the player was drafted v their ranking on THICC.
So the 1st round has an average 5.5 difference between the two lists. That’s terrible. BUT. What makes it terrible rather than just ‘meh,’ is it’s complete swing and miss on both Richard Littlewood and Jaja Ding-Dong. Littlewood was taken 25 picks later and Dong 17 picks later than THICC predicted. This is explained largely by position/team need. 16 of the first 28 picks were either a WR, LB, or S. Nearly 60% of the players taken in the first two rounds. While THICC nailed the Alexander Franklin OL pick at 8th, no other OL was drafted until picks 30 and 35, and then none after that for the rest of the draft. Clearly offensive line was not a priority for the league. So while Littlewood had a similar THICC score to Franklin, there was just no room for another OL that high on the board due to team needs. Similarly, with Ding-Dong, only four TEs were drafted in total, and none until 18th overall. While I believe he was a victim of teem/league need as well, I will admit the THICC system may have over scored him in part due to how long the user had been around.
So while I still don’t love the THICC 1st round predictive abilities, if we exclude those two players, the average difference between the system rank and the player’s draft slot is only 2.3. Much more palatable.
The second round showed a lot more promise, as 3 players were taken with one 1 spot of their THICC ranking, with 1 player picked exactly. A problem with THICC presents itself here as Jonathan Shuffleboard, ranked 17th, fell to 33rd in the draft. Shuffleboard was the newest creation of all eligible players, meaning his total TPE was low; however, his activity in a short period gained him multiplier points and vaulted him into the top-20 on THICC. Clearly, league managers place less weight on that measurement. While the avg diff for round 2 is 5.1, if we remove Shuffleboard (every round gets a mulligan!), the THICCraft slot difference is only 4.3. To me that is darn good for players picked 15th through 28th. A lot of uncertainty.
The last two rounds were a mess, lots of players taken earlier or later, though the system did get 8 players selected within 7 slots of their ranking. That’s no easy task when talking about players 29 through 49.
The outliers here were Redbeard McFredbeard who was drafted 13th overall but ranked 37th, and the big one, Derek Wildstar, who barely made the THICC list at 49, but was selected 11th! These two players really emphasize the limitations of the first THICC system – it does not vaule user joined date nearly enough. McFredbeard and Wildstar were the 12th and 13th “oldest” players out of the pool of draft-eligible players. This alone seems to indicate they are top-20 picks. Despite the fact that they had not separated themselves from their peers in activity or earnings, their mere presence in the league for a longer duration shot them up the charts.
While the limitations of this system are evident, I think that with some minor tweaks here and there, this system could beome a pretty predictive draft module. And if I dug a bit deeper into individual team needs then I think it would look even better.
I may see if I can make these adjustments in time for the next draft, so here’s to hoping another double media event is on the horizon so I can get some motivation.
So I’ve gone back and reviewed the outcomes. Spoiler alert – don’t use this to mock the upcoming draft! However, while the THICC system’s limitations are evident in the early part of the draft, it appears to have merit the later in the draft you go. My interpretation of this is that the system does not take into account team needs, which is obviously extremely paramount in the first round, whereas the later you go in the draft the more teams are trying to find the best available/most likely to stick around players. This is much more in line with what the THICC system is shooting for. So that’s…encouraging?
Well without further ado, below is the outcome of the Season 26 draft, along with the THICC system selections. Below each round is the average difference between where the player was drafted v their ranking on THICC.
So the 1st round has an average 5.5 difference between the two lists. That’s terrible. BUT. What makes it terrible rather than just ‘meh,’ is it’s complete swing and miss on both Richard Littlewood and Jaja Ding-Dong. Littlewood was taken 25 picks later and Dong 17 picks later than THICC predicted. This is explained largely by position/team need. 16 of the first 28 picks were either a WR, LB, or S. Nearly 60% of the players taken in the first two rounds. While THICC nailed the Alexander Franklin OL pick at 8th, no other OL was drafted until picks 30 and 35, and then none after that for the rest of the draft. Clearly offensive line was not a priority for the league. So while Littlewood had a similar THICC score to Franklin, there was just no room for another OL that high on the board due to team needs. Similarly, with Ding-Dong, only four TEs were drafted in total, and none until 18th overall. While I believe he was a victim of teem/league need as well, I will admit the THICC system may have over scored him in part due to how long the user had been around.
So while I still don’t love the THICC 1st round predictive abilities, if we exclude those two players, the average difference between the system rank and the player’s draft slot is only 2.3. Much more palatable.
The second round showed a lot more promise, as 3 players were taken with one 1 spot of their THICC ranking, with 1 player picked exactly. A problem with THICC presents itself here as Jonathan Shuffleboard, ranked 17th, fell to 33rd in the draft. Shuffleboard was the newest creation of all eligible players, meaning his total TPE was low; however, his activity in a short period gained him multiplier points and vaulted him into the top-20 on THICC. Clearly, league managers place less weight on that measurement. While the avg diff for round 2 is 5.1, if we remove Shuffleboard (every round gets a mulligan!), the THICCraft slot difference is only 4.3. To me that is darn good for players picked 15th through 28th. A lot of uncertainty.
The last two rounds were a mess, lots of players taken earlier or later, though the system did get 8 players selected within 7 slots of their ranking. That’s no easy task when talking about players 29 through 49.
The outliers here were Redbeard McFredbeard who was drafted 13th overall but ranked 37th, and the big one, Derek Wildstar, who barely made the THICC list at 49, but was selected 11th! These two players really emphasize the limitations of the first THICC system – it does not vaule user joined date nearly enough. McFredbeard and Wildstar were the 12th and 13th “oldest” players out of the pool of draft-eligible players. This alone seems to indicate they are top-20 picks. Despite the fact that they had not separated themselves from their peers in activity or earnings, their mere presence in the league for a longer duration shot them up the charts.
While the limitations of this system are evident, I think that with some minor tweaks here and there, this system could beome a pretty predictive draft module. And if I dug a bit deeper into individual team needs then I think it would look even better.
I may see if I can make these adjustments in time for the next draft, so here’s to hoping another double media event is on the horizon so I can get some motivation.
Quote:1,195 words