Contributed by Brian Donovan, President of Stockcalc
The purpose of this article is to examine Analyst Consensus accuracy prior to a peak in the markets like we are currently experiencing. We will show how analysts consistently over estimate the number of stocks that are undervalued during times of significant market volatility giving a premature buying signal to investors.
Disclaimer: I am the President of Stockcalc.com a fundamental valuation platform. All analysis has been derived from the StockCalc site for this column. This is also my first article in “The Edge”
I was at an Advisor Trade show in late January and the talk on the floor was all about stock valuations. At the same show a year earlier no one was discussing value rather how far they could extend their client’s portfolios into FAANG stocks. This was just after the 15% Dec 2018 drop in the markets with most of that recouped by the time of the show.
Our booth at the show focused on our website – StockCalc. The stock most asked about was Tesla followed closely by Apple, Facebook and Amazon. Our models showed each of these to be overvalued. These were among the most common stocks Advisors were holding for their clients at the time.
I left the show looking to test for valuation bias in Analyst Consensus using our site as the counter test.
Hindsight is a teacher but every analyst will tell you it is not permissible in the models. Many quant traders rely on models to help understand valuation. We do the same. Each night we run 5 models on 8000 stocks (NYS, NAS and Canadian Markets). These models include DCF, 2 types of comparables (PE, PB, PCF,PS and EV) to EBITDA. We also run a set of comps for companies with negative earnings using assets, stock price comparisons and book value, the stock against itself historically and an adjusted book value. We then run each stock through a series of algorithms to derive an overall weighted value for each of the 8000 stocks across 146 industries.
To back test an undervalued bias I pulled our weighted valuations and analyst targets as of Feb 1st for the DOW 30. I also pulled prices as of Close Mar 20th, 2020 and compared against “Analyst Consensus” in Figure 1. There are two striking things that are apparent when doing this analysis: The first is that only 1 Analyst Consensus was overvalued whereas our modeling showed half the DOW 30 to be overvalued as of Feb 1st. The second is that with the recent drop in prices, driven by an exogenous trigger in Covid-19, our 15 overvalued stocks reached our valuations to the downside. Again this compares to only 1 Analyst Consensus of Overvalued stocks reaching target.
We have been seeing this same pattern going back a number of years. I also back tested all stocks in our database that had Analyst consensus for the last 5 years as of Jan 1st each year. I ran a test set of 3250 companies which showed Analyst Consensus at 80% undervalued, 20% overvalued versus our 49/51% split during that time period. The other of note is the Analyst consensus data was correct only 1 time out of 2 on a 12 month go forward basis whereas our valuations were reached 3 times in 4. (ie the stock reached the valuation to the up or downside within the next 12 months). This was in a market that saw the DOW go from 15,700 to 25,000.
As the markets became extended over the last 15 months, valuation increased in importance but seemed to stay off the radar as ‘growth is driving this market’.
Figure 1.
So where are we now with the recent drop and rebound in the market? This table shows Analyst community still has 29 companies as undervalued whereas now StockCalc has 27. Given the drops in the market we now have similar undervalued assumptions. We will revisit this again later in the year.
Leave a Reply