Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Jane Street Market Prediction ($100k Kaggle competition)

kaggle.com

Post reply on HN
Print view View original

41–50 of 217 posts

Re: Jane Street Market Prediction ($100k Kaggle competition)

#41

Post by cultus »

1helloworld1 wrote: post #38

Isn't it pretty well known in the finance world that using stale public information to predict the market is a fool's errand? Unless you have some kind of specialized non-public data (e.g satellite images of number of cars parked outside parking malls, number of cargo ships moving in and out), trying to predict the market with historical data does worse than "Just give me some monkeys, darts and a dart board".

I'm sure they've got more/better data in production. There seems to be some arbitrage that can be shaved off the edges for players with innovative enough strategies and good and timely enough data.

Re: Jane Street Market Prediction ($100k Kaggle competition)

#43

Post by renewiltord »

aaronblohowiak wrote: post #37
renewiltord wrote: post #33

Earlier quoted context omitted.

With an engineer phone screen and three on-site interviews, that's 4 hours of engineer time. $150/hr compensation per engineer, so cost is roughly 2x, $300/hr. So $1.2k to run a candidate through the pipeline post-initial-qualification. To get one candidate and come out superior, acceptance rate should be 1%. (i.e. 99 failures). But if there are 50 leads from the program, and you convert a fifth, that's 10 candidates…

I think you are significantly underestimating the cost per hour of jane street employees

I have first party information, two years out of date. Do you have contradictory first party information? Yes/no will be sufficient for me to adjust my priors and I will be grateful.

Re: Jane Street Market Prediction ($100k Kaggle competition)

#44

Post by usmannk »

minimaxir wrote: post #40
1helloworld1 wrote: post #38

Isn't it pretty well known in the finance world that using stale public information to predict the market is a fool's errand? Unless you have some kind of specialized non-public data (e.g satellite images of number of cars parked outside parking malls, number of cargo ships moving in and out), trying to predict the market with historical data does worse than "Just give me some monkeys, darts and a dart board".

Granted, a typical Kaggle metagame-that-is-technically-against-the-rules is to use data from outside the dataset, which is one of the reasons winners have to be validated.

Generally this is allowed if you publish the data you're bringing in. They even create a sponsored thread for it in most competitions.

Re: Jane Street Market Prediction ($100k Kaggle competition)

#45

Post by logicslave »

1helloworld1 wrote: post #38

Isn't it pretty well known in the finance world that using stale public information to predict the market is a fool's errand? Unless you have some kind of specialized non-public data (e.g satellite images of number of cars parked outside parking malls, number of cargo ships moving in and out), trying to predict the market with historical data does worse than "Just give me some monkeys, darts and a dart board".

Theres actually still money to be made in small scale strategies. Sophisticated funds are running billions. They cant focus on strategies that only work for 100-500k. This is where big returns can be made. Even warren buffet will say, if he was only managing 1 million, he would get 100% a year returns.

Re: Jane Street Market Prediction ($100k Kaggle competition)

#46

Post by elil17 »

tcbawo wrote: post #27
nv-vn wrote: post #6

Earlier quoted context omitted.

>Jane Street has spent decades developing their own trading models and machine learning solutions to identify profitable opportunities and quickly decide whether to execute trades. These models help Jane Street trade thousands of financial products each day across 200 trading venues around the world. >Admittedly, this challenge far oversimplifies the depth of the quantitative problems Jane Streeters work on daily, an…

This contest seems like the equivalent of the "inventor's hotline" infomercial. If it identifies one promising new approach that they can iterate on, it has probably paid for itself. It also serves as a good PR and recruiting tool. The prize is probably designed to bring in clever non-professionals. It's a win-win for Jane Street

The inventors hotline infomercial is a scam where they get you to pay for expensive patent filing, consulting, and marketing packages. They never intend to actually use any of the inventions.

Re: Jane Street Market Prediction ($100k Kaggle competition)

#47

Post by rrjjww »

bigdict wrote: post #11

This would be much more interesting if the features weren't anonymized. At this point this is just a widest/deepest neural net competition on some unknown bunch of features.

I was just about to post the same thing. I'm a statistician in my day job and the raw math is only one part of building a model. Human judgement (while often flawed) can be key to improving actual model performance.

Re: Jane Street Market Prediction ($100k Kaggle competition)

#48

Post by smabie »

1helloworld1 wrote: post #38

Isn't it pretty well known in the finance world that using stale public information to predict the market is a fool's errand? Unless you have some kind of specialized non-public data (e.g satellite images of number of cars parked outside parking malls, number of cargo ships moving in and out), trying to predict the market with historical data does worse than "Just give me some monkeys, darts and a dart board".

no not really. you can use public information to give you an edge. And I say this as a person who trades and develops models at a vety successful market making firm.

Alternative data no one else can get easily certainly has tremendous value though.

Of course, predicting one or two seconds into the future (my primary concern) is easier than days or years, so there's that.

Re: Jane Street Market Prediction ($100k Kaggle competition)

#49

Post by minkzilla »

Goosee wrote: post #31

Sine this is a competition, does someone mind explaining the reason for people to publicly post notebooks in the "Notebooks" section? Seems counter-intuitive to provide competitors with free information, unless you are trying to throw them off.

Not a frequent Kaggle user but from what I’ve seen the ones posted in notebooks are baseline examples. The things anyone who is competitive enough to win has probably thought of and dismissed or could implement themselves in half an hour and iterate from there.
Post reply on HN