How do you score equities based on multiple recommendations and target prices

data_science

#1

Hi folks,

I have been working on an investing platform for the last year. It’s in beta, going live soon but there are some additional features that are to be incorporated.

One of the things we are trying to solve is how to score a specific equity based on broker reports. Consider two kinds of stocks in the market - 1) popular blue chip stocks like Infosys or SBI, where there will be 10-15 reports available with BUY/SELL/HOLD recommendations on the stock and target prices; and 2) less well-known small companies with 2-3 broker reports available.

My questions:

  1. When summarizing broker feedback for a stock in a single number, can we state the average target price brokers have recommended, based on all the reports available? Or is there a more accurate way? SBI example for reference - top right number.

  2. With stocks that have just 2-3 reports in the past six months, what can be done? There doesn’t seem to be any way to talk about overall broker feedback on the stock, specially when the bigger equities have a much larger base of broker reports.

Thanks for the help!


#2

You might need to set it up as a regression problem and then decide how to score the stock. For the regression problem the y variable could be the change in stock price over a certain period after the report is issued. And the x variables could be name of analyst, type of recommendation, name of broker firm, target price, market cap etc. Hopefully you’ll then be able to identify which factors are important in predicting future stock prices.


#3

Thank you! Trying to finalize the scoring approach this week. Will update with the results.