- Note that this is NOT a test of coding skills. It is a test of whether you can find any meaningful insights (can be very small) using market data and statistics that gives an edge.
- If you already have an idea/algo on how to achieve this, you can directly use it and do a simple backtest with report.
- If you do not have your own idea but inspired to get one, then you can pick up just ONE algorithm below and test to find insights (it can be a specific value of a paramter than can increase the success rate, or a method for detecting a pair better than average, or any others). The insight can be very small, but it needs to be useful.
- After you done the coding, please provide simple comments on its assumptions or when/where will it work/or not.
- Token candidates: with top 50 Market Cap, remove the stable coin, and filtered by token with top 30 trading volume. (such data can be easily get using coingecko or other public sources).
- position allowed: either long token, or park as USDT (or other stable coin you like)
- Signal: choose any one you like.
- Benchmark: use BTC hold in the whole period
- Backtest period: past 1 year good enough. if you have a longer history, better.
- Evaluation criteria: Total return, sharpe ratio, maximum drawdown, drawdown recovery period
- Addon: anything you think that is useful.
- similar to strategy 1, the only changed part is you can either long or short tokens.
- Positions allowed: long, short, or park cash.
- price data: you can use token spot price data directly for long or short. (meaning ignore the spread of spot, futures for simplicity)
- similar as strategy 1 /2 , only if you prefer mean-reversion rather than momentum
- Time allowed: 7 days.
- Action: choose any strategy you like and code it out.
- Results to submit: code + a simple report