Deagol vs. Market... vs. Apple

Back in April I posted a couple of charts showing how my "fair value" and "target" forecasts compared to AAPL's price history over the last few years. Several readers have asked for an update, and I said I'd do one when AAPL made a credible move to catch up with its fundamentals, which is the basis for my FV and target forecasts. Seeing as AAPL has run about 80 points in 7 weeks, I was also curious to visualize how far towards a fair valuation has this recent attempt taken it.

I found I needed to make some changes to the chart, mainly on the vertical scale. The details from 2006-2007 were getting squished due to the scale now going up beyond $700, so I decided to use a log scale. Because Excel sucks at doing log scales, I've added a few custom labels and gridlines along the axis, which might seem odd if you're not paying attention (not every gridline is spaced equally in visual distance and some not even in the underlying values). Anyway, here's the chart (click to enlarge):

So there you have it, still way undervalued. Now, in case you haven't noticed from all of my quarterly earnings estimates, I should clarify how I come up with these FV and target values. It's quite simple, I have stuck with a 25-times forward-looking multiple on yearly EPS, as that has seemed a good mid-range valuation metric over the years of my analysis, except recently (as noted on the chart).

I also studied some relationships between DCF, PEG, and P/E valuation methods, and concluded that as long as Apple's estimated long-term future growth rate remained above 20%, using a 25 multiple on forward EPS seemed consistent with the more academic valuation methods. Needless to say, over the past few years Apple has in fact grown EPS significantly higher than 20%, anywhere between 40% and 70% (this is all just after the couple of years of iPod hyper-growth of 350% when EPS grew from a very low base). But the market has always remained skeptic of the sustainability of such high growth, and has kept the long-term, next 5 years growth estimate around 20%.

Thus, early on I decided I wouldn't argue with the market on which multiple was appropriate for AAPL, but instead I would try to beat the market by estimating a more accurate forward-looking EPS for the denominator. And this approach has worked relatively well for me, as shown on the graph above.

So, for the FV (the blue points along the blue curve), I take the next four quarters' EPS estimates, i.e. the annual EPS one year in the future, and multiply by 25. For the 12-month targets (the red points along the red curve) I do the same but with the following four quarters, that is, the annual EPS 2 years in the future.

In trying to assess the accuracy, reliability, and effectiveness of my attempts to beat the market, I can identify 3 sources of uncertainty. The chart above attempts to show the end result of the interplay of these three sources of uncertainty, with a clear visual of how my final goals of forecasting AAPL's price are being met (or not). First, the market expectations about AAPL could be wrong. And they have. Consistently. Way wrong. This is why AAPL can be held or tanked for months on end without regards to fundamentals, or how other times it can keep climbing, merely catching up, in the face of uninformed pundits rolling their eyes and flailing their arms in "things can't go up forever!" desperation. There's not much I can do about this, other than to take advantage of the opportunity and hope the market eventually corrects itself.

A second source of uncertainty, much harder to gauge, is that my views about the market's expectations could be wrong (i.e. my valuation methods, choice of multiple, targets, etc.). This is a field mired with land mines and controversy, from academy to practitioners, with wide ranging theories and practices from fundamental to behavioral finance to technical analysis and even esoteric stuff like astrology thrown in for good measure. I can't concern myself with any of it too much.

To me, market valuation is not much different than going to the grocery store and buying bread and milk: you look at the variety of offerings, check the prices, read a few labels, and pick the best deal in your subjective opinion. Some would say that price is always right, since it's the result of the interaction of all the market players coming together and agreeing on the best price. However, price is also always wrong, since it changes every day, every hour, minute and second. So at any given instant, yes that price was right, but the next second it's immediately proven wrong (not to mention in the next quarter, or a year or two, which is the timeframe I'm concerned with).

Anyway, my solution to this infinite complexity is to use a simple multiple, the one I see most consistently applied by the market over a long period of time, and intuitively assess whether any temporary deviation is justified. But that solution might no longer apply even in the long term, as suggested by the comment on the bottom right of the chart. Others have done a great job describing the current problem with AAPL's valuation and possible explanations seem quite elusive. See Asymco's Horace Dediu's excellent research on this over several articles. So I'll leave it at that.

Then there's a third source of uncertainty, the most critical one to me yet very easy to quantify, which makes the basis or inputs for the valuation methodology, and that's my EPS estimates. How certain can I be about those 1-year or 2-year estimates that I come up with? The chart above doesn't show this. It's not clear which of those targets might have been derived from flawed estimates, and that the market might have deviated from them not because it was wrong about AAPL, but because my expectations for AAPL were off. Quantifying this is a very simple exercise, yet I have never seen a single analyst do it for more than the most recent quarter. Because in order to assess the consistency and accuracy of our forecasts, a full history of them compared to actual results is required. Yet no one does this. I've attempted this with my estimates a couple of times, here and here (lousy attempts, I know).

So for today's exercise, I went ahead and aggregated all the EPS estimates I've documented making, more than 100 data points on different focus points looking into the future, and compared them to the actual reported results, charting the deviation bias (high/low) and accuracy (best/worst), and highlighting their statistical dispersion and skewness/clustering through colors, over the last 6 years and then again over the last 3 years. Here's what it looks like in four charts (click on each to enlarge).

There's 22 quarters worth of estimates for the same and next (1Q) periods, 20 estimates for two quarters into the future (2Q), 18 estimates for 3Q out, 17 for one year out, and 11 for 2 years out. The white lines separate each decile (10% of estimates within each column) so each colored block represents at least one estimate, and sometimes two, and sometimes in between (there's some interpolation involved). I tried to highlight the quartiles (25% buckets) through similar colors and gradients, not sure if that helps or not. Because this post is already extremely long, please feel free to use comments for further descriptions, interpretation, discussion, questions, etc.