I recently attended the 2018 New York Trading Show and one of the presentors in Advanced Analytics was Ronnie Shah, Head of US Quantitative Strategy for Deutsche Bank. He discussed a solution to avoiding the "Value Trap".

A little background...quotes from Investopedia:

What is a 'Value Trap'
A value trap is a stock that appears to be cheap because the stock has been trading at low valuation metrics such as multiples of earnings, cash flow or book value for an extended time period. Such a stock attracts investors who are looking for a bargain because they seem inexpensive relative to historical valuation multiples of the stock or relative to the prevailing overall market multiple. The trap springs when investors buy into the company at low prices and the stock continues to languish or drop further.
BREAKING DOWN 'Value Trap'
Successful in prior years with rising profits and a healthy share price, a company can fall into a situation where it is unable to generate revenue and profit growth due to shifts in competitive dynamics, lack of new products or services, rising production and operating costs, or ineffective management. For the investor who is used to seeing a certain valuation of the stock, a seemingly "cheap" price becomes interesting. However, it becomes a value trap to the investor if no material improvements are made in the company's competitive stance, its ability to innovate, its ability to contain costs, and management by the executives.
Is it a Value Trap?
An industrial company whose stock has been trading at 10x earnings for the past six months, compared to its trailing 5-year average of 15x.
A media company whose valuation has ranged from 6x-8x EV/EBITDA for the past 12 months, compared to its trailing 10-year average of 12x.
A European bank whose valuation has been below 0.75x price-to-book for the past two years, compared to a 8-year average of 1.20x

Because we often relie on fundamental data that is lagged at least by a quarter ( three months), it lacks information on what is happening now and thus we can easily fall into this value trap if thorough research on the company's current financial/operational status is not applied. Shah's solution to this problem is called "Nowcasting", a form of forecasting using LASSO, a linear model that estimates sparse coefficients. It is useful in some contexts due to its tendency to prefer solutions with fewer parameter values, effectively reducing the number of variables upon which the given solution is dependent. As the Lasso regression yields sparse models, it can thus be used to perform feature selection.

In his study, he presents LASSO with ~80 different raw fundamental variables with trailing historical data to 'nowcast' a particular value fundamental, say EV/EBITDA. The results of his study shows that replacing the latest EV/EBITDA with the nowcast EV/EBITDA, significantly increased Sharpe in addition to avoiding the "Value Trap".

There is a whitepaper on this entitled, "Avoid Value Traps using Quantitative Nowcasting Techniques", unfortunately it is available only to Deutsche Bank clients. If any of you are Deutsche Bank clients, it would be nice to share here.

Anyway, when I find time, I'll try to do my version of Nowcasting perhaps via Machine Learning.