Best hockey pool guide/list for the 2020-21 season

[hkyresearch]

Hello poolers,

Want to know which hockey pool list of projections was best for a single-season, points only draft in the 2020-21 season?

I did, so I went and ran 39 different guides, draft kits, cheat sheets and lists through more than 2.9 million simulations to find out.

If you are interested to know, you can check out my YouTube videos to see the results.
-The Intro (I recommend you start with this video)
-The Repechage
-The Finals (for 2020-21 season... corrected link)

-The Finals (apologies... this goes to the 2015-16 Finals)


Demetri.

Click to read more...

 

[DatsyukToZetterberg]

Thanks for doing this, I found it quite interesting especially as someone who's model was participating. I was particularly interested in the pick findings - that being in the bottom 3rd is essentially a death sentence in a perfect world and the advantage you have for picking early.

In terms of the models themselves I'm happy with how both the Marcels faired as well as my own model relative to the rest of the guides. I think one thing that may have hurt both the Marcel's and my own model is that they assume that all players will play 82 GP when we know that isn't the case. I'm hoping to be able to incorporate a better GP prediction model in this years iteration of the models. As well, It seems like in 2020-2021 my own model performed quite poorly in the 801-1000 range and then in the battles it seemed to really poorly in the larger style drafts.

I do have a question about the missed opportunity as I'm currently rebuilding the DtZ model itself right now and I'd like to compare the results from the old to the new model. The missed opportunity scores are:

the absolute differential of the real points at the projected rank minus the real points at the real rank for that player.

I was trying to follow along in the hypothetical section, but wanted to make sure I've got it right. If a player is projected to be 10th and they finish in 10th that would be TMO of 0, even if the projected points didn't match the actual points? For example my model had Brad Marchand projected to be 10th with 62 points; he finished in 3rd with 69 points, so I would have TMO score of 7.

 

[hkyresearch]

You are welcome. I'm glad you found it interesting.

I'm still working on other past seasons so it should help better define the percentages of winning when picking early vs late.

To answer your question, yes you are correct when a player is ranked 10th and ends up in 10th they have a missed opportunity score of 0 regardless of what the point projection was. And in the second part of your example, you had Marchand's projected ranked as 10th -> the truly 10th ranked player in the NHL by points just so happened to have 62 points (Crosby). Marchand actually ended up in 3rd with 69 points, therefore, 69-62 = 7 (abs). So your math was correct but it was a coincidence.

If someone had him ranked 20th in their projections, the missed op score would be 69-53 (Backstrom the real NHL 20th ranked player by points) = 16 abs.


I'm already looking forward to next seasons predictions.

 

[hkyresearch]

I have corrected the Finals link so that it now truly goes to the 2020-21 Finals.

Sorry about that. Was working on a past season at the same time and got the links mixed up.