✨ New feedback, new leaderboard, 3 weeks to go! 🏃

We’re racing towards the end of this iteration of the competition, and I for one, am very excited!

With only 3 weeks to go, I hope you’re all buckling up for one final push on the markets :rocket:

:sparkles: Updated feedback

Some new elements are added to the weekly feedback from week 4 on wards (if not immediately updated, give it a few hours :slight_smile: ).

See the updated description of feedback here.

We are also displaying the average marketshare publicly on the leaderboards!

:new: New Week 10 leaderboard

For those of you who noticed, there is a 2-week gap between the final submission deadline and the 9th weekly leaderboard. We’ve filled this gap up with a 10th profit leaderboard. This will functionally be exactly the same as the other weekly leaderboards (after the bolstered size!).

:microscope: Research survey

As part of the research, we have a short survey (<5min) prepared, if you haven’t filled it our already then please do! :point_down:

On the fit_model function and retraining models :test_tube:

As you might know, there are two reasons that this is a code competition:

  1. Realism. No company knows details about their potential future portfolio in advance.
  2. Research. As part of the research being done by us at
    Imperial and UQAM, we use the code to gain broad insights into market
    dynamics at a large scale.

That means we will sometimes have to retrain your models for research purposes.

But a small minority of submissions don’t include their model fitting code and have found workarounds. This is against the spirit of the competition.

Note that if your model is not re-trainable it will be disqualified from winning any prizes in the final profit leaderboard.


As always we’re here to answer your questions and listen to any feedback you have :ear:

All the best

Ali and the Pricing Game Team

3 Likes

If there’s ever a research paper published using this competition as a guinea pig, can we make sure to share this paper to use once it’s published? :star_struck:

 

Ph.D. Thesis :: How a bunch of coding amateurs battled it out in a negative-sum game. :rofl:

 

EDIT: Oh and if you ever run mine… Sorry in advance, … you will be releasing 1lb of CO2, aka 10 hours of crunch time.

3 Likes

Basically yes :point_up_2:

And of course! once the research is ready and public it will be widely shared with everyone.

We’re also in the process of organising a less formal “Town Hall” event to have a discussion between the participants, organisers and some sponsors in the short period following the end of the competition!

Details will follow in the next week or two on that :mailbox_with_mail:

1 Like

So the winner will be decided by one’s profits on the final submission, right?

Yes exactly. The evaluation metric for this challenge is your average profit over the 100K policies with year = 5.

Public market share is a great improvement :clap:

1 Like

Thanks for adding market share!

1 Like