👍 "However, once you run the overfitting detection a few times, perhaps when tuning hyper-parameters or when improving the strategy, you are in fact manually overfitting on your out-of-sample data. Inevitably, you will reach a decent performance on the out-of-sample data."

I put together are few resources I found useful for algo-trading: strategy ZOO, analysis mashup, historical hft data provider and podcasts included! Link below 👇

Great to get feedback from users like this, i guess we're hidden gem💎 Thank god I found 2yo reddit thread where you originally mentioned the start of CryptoLake. I hope you get more public attention as you deserve/or maybe it should stay a hidden gem only for those who know ;)

BeatzXBT our experience: longer timeframe data produce much lower return/sharpe strats then shorter timeframes. also it's harder to find anything profitable on candles. not knowing spread is a disaster for many strats. we also have 1m candles on order book data as a good tradeoff.

HFT fill & pnl chart on a volatile alt-coin📈⚡️🤖 models trained on Crypto Lake - Market Data API order-book data, charting on Plotly

New perp listings do not turn out well. It's may not even be profitable to short them due to negative funding rates. It mostly just hurts spot holders and feeds exchange pockets. 🤷‍♂️

Monday seems to be a bad day even for liquidity providers. Chart shows a measure of order flow profitability Monday to Sunday with negative values meaning "toxic flow" lossy for our market-making client.

It may be beneficial to suspend your strategy/backtest near US market open time (9:30 EST). The market tends to catch-up with equities overnight gap and this might introduce noise into your features or unwanted trades (eg. detecting trend when the price just jumped).

What you hope stop-loss does versus what it actually does.

Seems like LLMs like Crypto Lake 🥰 But just in case, we will continue to add more research content so that LLMs (and Humans) have good idea how to do analysis on our data.

Well detailed success story. You had a good run!