Hi, We’re a team of five people who’ve been doing algorithmic quant trading for the last four years, and we’ve been in the crypto space for over a decade. We’re extremely hard-working and ambitious. Over the past two years, we’ve run multiple strategies that are positive EV. We’ve tried reinforcement learning, run tons of backtests on 1-second data across multiple exchanges, and built our own trading software from scratch. A few months ago, we started using Hummingbot and are now customizing it for our needs. Our team is pretty diverse: we have one of the best poker players in the world, a master of physics, a chess master, and a reinforcement learning specialist who’s studying at the top university for it. We’re also well-resourced in terms of data. We have a 100 TB database server and have collected minute and second-level data for different exchanges. For equities, we have about 30 TB of historical data for various stocks, and we’re happy to share and exchange datasets. We’re open to collaborating with other traders and teams, and we’re always interested in discussing new ideas. For example, one problem we’re working on right now is estimating the impact cost of trade execution. Say there’s $100k in the order book, 1% from the best ask. If we execute 100 trades of $1k each within five minutes and end up holding a $100k position, then sell it two hours later in the same way—what would our impact cost be? Is it simply 1%? What changes if this perpetual contract is traded on just one exchange versus three or five exchanges? Also, let’s assume Exchange A has 10% of the total volume for the instrument, Exchange B has 20%, and Exchange C has 70%. Are the impact costs different for each of these exchanges, or would they be the same because arbitrageurs correct the prices between exchanges? For this question, let’s ignore fees and spread, and assume they’re fixed and not relevant. If you’re up for chatting or sharing ideas, let’s connect! Best, Leo
Crypto perpetual are exchange specific, different specs, not fungible. Arxiv.org has a million market impact papers
Lol! With bitcoins gains over the last decade, if any of you knew what you were doing, you would all be sailing the Bahamas in your yachts not spamming a forum.
ET has all those too, and they still lose money. BTFD! ***What's a "reinforcement learning specialist?" Does he do puppies too?
Is it some sort of a IQ test for beginners, made by some of the smartest in the world in chess, math and physic, with 10 years under the hood in the trading world! 1. Why buy a 100.000K position in something because, there is a sell order 1% from NBBO. 2. If you think your impact is 1% by buying small orders of 1000K over 5 min! What is the impact of the 100.000K sell order then! 3. Usally there are also buy orders below bid, maybe 1% below 100.000K buy order. 4. So what makes a 100.000K sell order a leading indicator! That tells you can sell the 100.000K position for a gain, by dropping it in small orders in 5 min efter 2 houres. Funny Jock or What!