Free season model for your league (1 Viewer)

[NorCal8]

Quick background: i have a regular job developing financial models for investment banking clients. Over the past couple seasons, i've been working on a system that will model fantasy football leagues to better evaluate teams, and trades, with the basic idea is that everything comes down to wins (not points), and it is a zero sum game. The model uses a combination of actual results and standard pre-season projections and plays out the entire season against your actual schedule thousands of times to project returns. So far i think i've been fairly successful in creating a system that works, but want to extend some trial versions outside of my own leagues.

Offer: I'll input your league into my system and model out your league and post the results (projected # of wins for each team in your league). You can fire off a few trade proposals, and I'll tell you how they affect the league, again in wins & losses.

Catch: You have to fill out a sheet on your league - including everyone's roster, scheduling, and scoring. It should be simple, but will take a few minutes. I can only do 2 or 3 - first come first served. I won't give you the model - just the results.

PM me if you are interested. Thanks.

 

[Ignoratio Elenchi]

How will you know if it works?

 

[NorCal8]

How will you know if it works?

Thanks - I'm hoping the people who I model for will tell me at end of year.Couple qualifiers I should have pointed out earlier:- ten team leagues only for right now- No IDPs

 

[Ignoratio Elenchi]

How will you know if it works?

Thanks - I'm hoping the people who I model for will tell me at end of year.

But there are at least dozens of transactions in my league every year. The team I have now will certainly look a lot different eight weeks from now, due to injuries, trades, add/drops, etc. Do you attempt to account for in-season roster management at all? How do you plan to determine (quantitatively or otherwise) whether or not your model actually worked once you have the actual end-of-season results?

 

[moleculo]

Quick background: i have a regular job developing financial models for investment banking clients. Over the past couple seasons, i've been working on a system that will model fantasy football leagues to better evaluate teams, and trades, with the basic idea is that everything comes down to wins (not points), and it is a zero sum game. The model uses a combination of actual results and standard pre-season projections and plays out the entire season against your actual schedule thousands of times to project returns. So far i think i've been fairly successful in creating a system that works, but want to extend some trial versions outside of my own leagues.Offer: I'll input your league into my system and model out your league and post the results (projected # of wins for each team in your league). You can fire off a few trade proposals, and I'll tell you how they affect the league, again in wins & losses.Catch: You have to fill out a sheet on your league - including everyone's roster, scheduling, and scoring. It should be simple, but will take a few minutes. I can only do 2 or 3 - first come first served. I won't give you the model - just the results.PM me if you are interested. Thanks.

in.

 

[emge]

I am in, pm me your email address so i can send you my details

 

[Ignoratio Elenchi]

I will also run your league's season through my proprietary model for free. Whatever info you send to NorCal8, send the same sheet to me via PM. I'll provide you the results, and then at the end of the year you can post your league results and tell us which model did better.

 

[NorCal8]

Transactions are essential and a fully functioning model should update weekly with all changes. I'm doing a pre-season/week 1 projection based on rosters today. If the guys in the test group want to keep updating, all they have to do is email me roster changes as often as they'd like it changed.

Models, generally speaking, are designed to project likely results over time in order to make the best immediate decisions. As inputs change, so do the models.

 

[Ignoratio Elenchi]

Transactions are essential and a fully functioning model should update weekly with all changes. I'm doing a pre-season/week 1 projection based on rosters today. If the guys in the test group want to keep updating, all they have to do is email me roster changes as often as they'd like it changed.Models, generally speaking, are designed to project likely results over time in order to make the best immediate decisions. As inputs change, so do the models.

Yes, I know how models work. That's why I was curious how you're going to assess how effective your model is, assuming that people aren't going to be sending you weekly updates of roster changes, etc. Not pushing buttons, just wondering what your goal is here. If your model projects me for 9 wins, and I end up with 8 wins but half my roster is turned over from the beginning of the season, is that a success for your model? And I'm serious about putting my models up against yours. If anyone wants in just send me whatever you send to NorCal. It'll be a fun little competition.

 

[moleculo]

Nerd fight!

 

[Donsmith753]

Iggy will you model bigger leagues?

 

[FreeBaGeL]

I'll jump in on this if there are any spots left.

 

[Ignoratio Elenchi]

Iggy will you model bigger leagues?

I just wanted to pit my system against OP's, so I'm not really looking to do any leagues that his model can't handle.

 

[moleculo]

I will also run your league's season through my proprietary model for free. Whatever info you send to NorCal8, send the same sheet to me via PM. I'll provide you the results, and then at the end of the year you can post your league results and tell us which model did better.

pm me your e-mail address and I'll send you the same stuff.

 

[NorCal8]

To your point however, how would we compare, since inputs change?

I built a model to give me an edge in several money leagues, and was hoping to get some field use feedback with a wider variety of league distributions and scoring systems to improve it. That's pretty much all i'm interested in at this point. I use over 10,000 game simulations and a couple hundred input variables. I think it is very reliable.