Asymco details comparison between forecast and actual Apple figures
Earlier this morning I published a rough-and-ready run-down on Horace Dediu’s Apple forecast figures compared with the actual results. I reckoned they were pretty accurate. Now Horace has posted his detailed assessment and is a little less kind to his prognostications. Since the article gives a good deal of background and an insight into the analyst’s thankless task, I’m reproducing the text in full after the break.
Apple Earnings: Evaluating my performance
By Horace Dediu
As chronicled here, this quarter I went through three sets of estimates. This is a departure from my standard practice of sticking with one forecast made three months in advance. I felt last quarter was filled with materially important developments which deserved updating.
So how did I do?
I wanted to see if the additional information helped or hindered predictability so I put my three predictions side-by-side and measured the errors. I also added the “legal leak” from April 15th (published April 19th) to see whether it was an accurate predictor. The table below shows the error rates with the lowest error highlighted.
It’s clear that my error rate decreased as new information was used. It’s also clear that the “legal leak” was very accurate. What follows is a discussion of each line item.
The iPhone forecast was very accurate both early and late in the quarter. At 1.36% error it was even lower than what was leaked by Apple. I used the information about China and Verizon to boost sell-in and increased the growth rate to 110%. That proved to be the right call. I did not pay any attention to talk of component shortages as I did not consider them to affect Apple whose market share of devices is so low. I would score my “late” estimate as an “A”.
Even though I disclaimed it, my iPad forecast was very inaccurate. In contrast, the leak was very accurate. Although mitigated by the higher sell-through (5.1 million) shipments were constrained by the change in product. Details like the date of shipment, the date of announcement and the ramps made this a particularly difficult transition to predict. At 36% error, I score it as an “F”.
The iPod error was about 10%. That’s mediocre. The product dropped in sales by 17% (vs. 10% expected). Again, the mitigating factors, like higher mix of iPod touch and thus increased prices helped to keep the financial performance decent. I score this as a B-.
The Mac prediction was accurate. The error was less than 5%. I will discuss the Mac business in a separate posting but it suffices to say that the growth was excellent but predictable. I score my prediction as an “A-”.
Revenues were reasonably predictable however pricing seems to have been affected by exchange rate volatility. The iPhone turned out to have a huge average selling price ($660) which made up for the lower revenues from iPad. The error of 3.25% is acceptable. I score it as an “A”.
The Earnings number was very accurate. The error for the late estimate was less than one half of one percent. This bottom line number is a function of all the other numbers so it’s the most encompassing. You have to account for factors like interest income, changes in the number of outstanding shares and tax rates. None of these are predictable. However, I was lucky. With an error below 1%, I rate this as an “A+”.
My overall performance as a grade point average would be 3.43 or B.