Read on Twitter
Time for another thread on algo trading basics as I've been getting many DM's on where to start:
1. Start with trading manually and finding what works for you, what trading style, indicators, R/R, etc.
2. If you know how to use Pine script on Tradingview start there to backtest
1. Start with trading manually and finding what works for you, what trading style, indicators, R/R, etc.
2. If you know how to use Pine script on Tradingview start there to backtest
2/ If you are not familiar, Python is usually the best/easy language to do algo/quant research in (with some R/Matlab for low level). This is a good course to begin with https://www.udemy.com/course/python-for-finance-and-trading-algorithms/
You'll need to understand dataframes in pandas to be able to analyze candle data.
You'll need to understand dataframes in pandas to be able to analyze candle data.
After that, you want to choose the right backtester. There is a quite a few out there, but for crypto and @QuantumLabs1 we built our own in-house. We even have ML/Evolutionary based optimizer to churn out variations of backtests/parameters so we aren't tweaking manually all day.
The biggest issue I see with new algo devs is they dont apply slippage/fees in their backtest and assume unlimited liquidity. You also want a strategy that has the least amount of negative months rather than being fixated on getting 100-200x returns. For reference..
We have a strategy that does 200x returns in a 3.5 year period with 1x leverage. However, if you had started in year 2, you may not get the full 200x returns. You'd have to start from Day 0 and hold until 3.5 years is up. This is not practical.
The ultimate goal is to have a decent sharpe above 1.75, sortino above 2.0, and have no more than 3 neg months in a 12 month period. That way if an investor/client uses your algo and they come in on a random month, you can assure they will have smooth returns.
Initially, we had algos that would theoretically return 200x+ returns but most of those returns were done in a few months time (think Nov 2017-Jan 2018) so if you came in on Feb 2018 your 200x is probably more like 2-5x returns. Timing is key, smooth returns is key, call it IRR.
@CanteringClark Tagging you as I thought you might find this interesting as I've seen you're interested in learning Python/algo/etc.
