Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

How much external data is used for these models or human input/fine tuning on a regular basis?


The models vary greatly in the number of external data sources they pull from to compute signals. The simplest ones like https://grizzlybulls.com/models/ta-mr-basic and https://grizzlybulls.com/models/ta-trend-basic only use a technical analysis calculations on the raw price and volume data of SPX on a few different timeframes. The most complex pulls data from dozens of external sources as it incorporates macroeconomics, monetary policy, various market sentiment sources, yield curve, valuation data, fundamentals, earnings estimates, etc.

It's also worth noting that every additional data source adds some risk of that data source being down or publishing inaccurate data during real-time signal calculation which can cause inaccurate signals, so in order to justify that risk, the external source must meaningfully contribute to alpha or better risk-adjusted returns.

None of them involve a human element in real-time. However, they are occasionally updated as new data comes in, but any updates only apply going forward so as to preserve the live trading history accurately (live trading start date varies by model from mid 2020 to jan 2022 with 2009 - 2020 being purely backtest for all models).


> The simplest ones like https://grizzlybulls.com/models/ta-mr-basic and https://grizzlybulls.com/models/ta-trend-basic only use a technical analysis calculations on the raw price and volume data of SPX on a few different timeframes.

Am I right to doubt that something this simple generates any alpha whatsoever?




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: