Market capitalization based weighting, the basis of the Nasdaq-100 index and $QQQ ETF, probably constitutes a baseline for what can be considered "unbiased". An

Market capitalization based weighting, the basis of the Nasdaq-100 index and $QQQ ETF, probably constitutes a baseline for what can be considered "unbiased". Any AI agent that measures market "sentiment" can only be conditioned upon the quality of the data it is fed. Which will vary across companies. An example of one of the best algorithmic strategies I have seen is the following. During secular bull market eras. Simply buy and hold for a period of 24 months. Every IPO that comes down the pike. Regardless of sector. Backtesting this strategy yields annualized 50% rates of return. Which beats $FB performance the last four years :) No doubt, ML could further optimize selectivity, weightings, hold duration, etc. The central thesis is that growth in market cap is strongest during the growth phase of a company. Of course, today is the day another great algorithmic trading idea: fading volatility spikes. Unwinds in most violent and consequential fashion. Be cautious out there! Two Big Volatility Players May Be on the Loose as VIX Tops 15 https://www.bloomberg.com/news/articles/2018-02-02/two-big-volatility-players-may-be-on-the-loose-as-vix-tops-15 https://www.bloomberg.com/news/articles/2018-02-02/two-big-v...