I’d definitely put Numerai on your resume. If you haven’t already got a job in the area I think data science competitions are the best way to stand out. I think personal projects (e.g. you read a paper, have an idea, try it, and document it) also do have their merits, but it doesn’t compare your ability to other people. Whereas in competitions like this you can express your ability as a number on your resume (your percentile rank on the leaderboard), which I think is a great way to summarise your expertise. The big problem with a lot of data science competitions is you’re not productionising your work, whereas in Numerai you’re staking money on your models and submitting them every week, so I think performing well in this kind of competition has a lot more weight for your ability. I think Numer.ai is great for your resume as given the tournament never ends you have a lot of time to try out novel ideas and try out advanced techniques (e.g. custom loss functions and training loops). If you can come up with a novel idea that works, or implement an advanced technique, then it’ll look great on your resume. I wouldn’t worry too much about GitHub as most recruiters don’t have a lot of time to look at that stuff in much detail. But if you can put your code alongside a report explaining what you did and why on your GitHub, then that’s definitely bonus points.
Although they don’t tell us exactly what the target is, if you watch the videos on their YouTube you can get a rough idea of what we’re predicting. Each week our models predict the ranks of global equities 4 weeks from now, and we’re scored on the correlation of our predicted ranks with the actual ranks at the end of those 4 weeks. But what those ranks actually correspond to I don’t think we know. So in brief I’d just say we’re ranking global equities.