Parity Finals Predictions

Hadrian Mertins-Kirkwood's picture
Submitted by Hadrian Mertins... on Wed, 2019-01-30 11:53

Forums:

After another thrilling week of playoffs (a tie!) the stage is set for the Session 1 Championship. How did last week's predictions shake out? And who are the favourites heading into the final games? Behold!

---

Week 11 (Jan. 28): Semifinals

[SF] prediction: Big Love (19) vs. An Andréa's Fault (21)
Result: Big Love (18) vs. An Andréa's Fault (18) *Big Love advances on technicality*

[SF] prediction: Flaccid Member(ship) (18) vs. cRyan till the throws come home (17) *universe*
Result: Flaccid Member(ship) (21) vs. cRyan till the throws come home (17)

[5th SF] prediction: Uncorked (15) vs. Hot or Not (23)
Result: Uncorked (16) vs. Hot or Not (23)

[5th SF] prediction: #goals (22) vs. ikneebriated (21)
Result: #goals (18) vs. ikneebriated (17)

[7th play-in] prediction: Moneyball (25) vs. Badgers (16)
Result: Moneyball (18) vs. Badgers (20)

Through two rounds of playoffs the Effectiveness Rating-based predictions are now 6-4. Three of those missed calls ended on universe points (or ties) that could have gone the other way, so overall not bad!

Kudos to Big Love's big line of Jessie, Travis, Liza and Jim, who spearheaded a remarkable comeback to tie their game. In the other semi, Flaccid took an early lead and never looked back. John Haig had a monstrous game with 44 catches and 10 assists. Who was playing D on that guy?

Week 12 (Feb. 4): Finals

[1st] Big Love (20) vs. Flaccid Member(ship) (15)
[3rd] An Andréa's Fault (25) vs. cRyan till the throws come home (15)
[5th] Hot or Not (21) vs. #goals (16)
[7th] Uncorked (26) vs. Badgers (22)
[9th] ikneebriated (20) vs. Moneyball (19)

With the exception of a toss-up for last place, my methodology projects clear winners in every game next week. Big Love is slated to breeze to the championship after facing their toughest test in the semis. An Andréa's Fault is projected to rebound and take third while Hot or Not closes out the consolation bracket. Uncorked should take 7th.

---

All eyes will be on Flaccid Member(ship) in Week 12. Can the #7 seed continue to defy expectations and orchestrate their greatest upset yet? Tune in at 8:20pm, no excuses!

 

Hadrian Mertins-Kirkwood's picture

Hadrian Mertins...

Tue, 2019-02-05 09:43

The Effectiveness Rating-based predictions go 4-1 in the final week. The margins of victory were generally smaller than I called, but every game was still decided by multiple goals. The incorrect prediction was in a game that I called a "toss up" so hard to feel bad about that one.

Overall, the predictions were 10-5 during playoffs. Since there are a number of new players in session two I won't be able to make new predictions for a while, but expect them again for the playoffs!

---

[1st] prediction: Big Love (20) vs. Flaccid Member(ship) (15)
Result: Big Love (14) vs. Flaccid Member(ship) (11)

[3rd] prediction: An Andréa's Fault (25) vs. cRyan till the throws come home (15)
Result: An Andréa's Fault (22) vs. cRyan till the throws come home (19)

[5th] prediction: Hot or Not (21) vs. #goals (16)
Result: Hot or Not (23) vs. #goals (21)

[7th] prediction: Uncorked (26) vs. Badgers (22)
Result: Uncorked (25) vs. Badgers (17)

[9th] prediction: ikneebriated (20) vs. Moneyball (19)
Result: ikneebriated (18) vs. Moneyball (21)

Jon Rowe

Tue, 2019-02-05 10:06

I wonder how does your model account for pace of play? Why are some games predicted as lower scoring affiars than others - does it take team defense into account or just average offensive outputs?

Hadrian Mertins-Kirkwood's picture

Hadrian Mertins...

Tue, 2019-02-05 11:37

The score predictions have nothing to do with the team stats per se (i.e. average goals for/against a team don't matter). I take the average offensive ratings of the *individual players* on a team and compare them to the average defensive ratings of the players on the other team and then I multiply the difference by the average number of goals scored in all Parity games this season (about 19.7 per team).

So if two teams have good offensive ratings and bad defensive ratings on average then the model predicts a high-scoring game (more than 39 combined goals). Weak O and strong D ratings predict a low-scoring game (fewer than 39 combined goals). And when a team is strong on both O and D vs a team that's weaker in both ratings you get really lopsided score predictions (e.g. Andrea's vs. cRyan).

To answer your question directly: yes it does take team D into account. Teams with stronger defensive players overall predict lower-scoring games.

Matthew Cole

Tue, 2019-02-05 15:05

How did your model predictions compare to purely betting on the higher seed. If it was better that seems like good validation.

Christopher Keates

Tue, 2019-02-05 17:23

Ties go to the higher seed, they inherently have the advantage.

Matthew Cole

Wed, 2019-02-06 16:03

Haha, very clever  .... https://media.giphy.com/media/OPU6wzx8JrHna/giphy.gif

Hadrian Mertins-Kirkwood's picture

Hadrian Mertins...

Wed, 2019-02-06 14:00

Good question! I had to look up the answer...

In the 10 games where I called the higher seed to win, my record was 7-3. In the 5 games where I called the lower seed to win, my record was 3-2. My overall record was 10-5.

If you had just taken the higher seed in every playoff matchup, your overall record would be 9-6. So my methodology was a slightly better bet than just taking the higher seed in this particular playoff (and definitely better than a coin flip).

This is still a relatively small sample, so I'm looking forward to putting it to the test over a longer period of time.

edit: Keates' point is actually really relevant here because I called the lower seed (An Andréa's Fault) to win the game that was awarded to the higher seed by default. If we remove the tied semi from our sample then my record is 10-4 and the higher seeds approach is 8-6.

Justine Price

Tue, 2019-02-05 10:08

Much like the Super Bowl. the finals was a surprisingly low scoring game that could have benefitted from some chili and nachos for the keen fans. But Geofford had doritos and I spotted a beverage or two. Parity Party Lite. Good game and all players noticeably gave it their all. Thanks for the show, folks.

Brian Perry

Tue, 2019-02-05 13:59

I recycled at least 3 cans on the way out ;) 

Trevor Stocki

Tue, 2019-02-12 15:40

maybe I shouldn't ask it.. but whatever happened to the lemon cake of sadness??? :-)