It’s the second week of March. By this time of the year, especially if Apple has been unusually quiet, you start to see a level of restlessness among tech pundits and bloggers. There should be news. There should be stuff to write about.

In the absence of any actual news coming from the House That Steve Built, rumors tend to pop up, which helps keep the self-perpetuating echo chamber operating at full blast.

To be honest, though, this isn’t just a matter of click-hungry press running on fumes. Because Apple’s desktop hardware is now so incredibly old, there’s an entire army of Mac users who are trying to decide if they should hold out a few more months, or begin their long slide into the Windows side of the world. This is a serious problem. Apple’s most recent upgrade for its highest performing machine was more than 40 months ago.

Serious, expensive, long-lasting decisions need to be made and Apple’s customers are trying to predict whether to stay or go. To help provide some insight, I’ve decided to subject Apple’s announcement history to some rigorous analysis.

To conduct this analysis, I built a spreadsheet based on event data posted by AAPL Investors. This provided a data set on 26 Apple events going back a full decade, to the beginning of 2007.

spreadsheet-analysis-summarized.jpg

Here’s the starting dataset I built based on the AAPL Investors article. Yes, Apple used to hold iPod events.

These events became individual observations in the analysis. I defined variables corresponding to the invite date of the event, the event date itself, the season in which the event took place, and the product categories the event was promoting.

It should be noted that I’m not including WWDC events in this analysis, because we’re looking for whether or not new hardware will be announced. Summer WWDC events are usually used to preview iOS and MacOS changes so that developers can get started updating their apps.

Seasonality

How you view the seasons depends, generally, on which hemisphere you live in. Since I’m in the northern hemisphere, we’ll use northern hemisphere seasons. Even though seasons don’t really fall on month boundaries, for our purposes, here is how I’m defining each season:

  • Winter: December, January, February
  • Spring: March, April, May
  • Summer: June, July, August
  • Fall: September, October, November

As the following chart shows, Apple has held twice as many events in the fall as in the spring, with only a very small number in either the winter or the summer.

apple-events-by-season.jpg

Apple events by season

Events per year

On average, Apple holds 2.6 events per year. That number has been declining in recent years, with 2016 actually showing a slight bump, from two to three events.

apple-events-per-year.jpg

Apple events per year

Probability of a spring event

Given that there are far fewer fall events than spring events, a question can be asked: is there any way to predict, based on the number of events in a preceding year, whether there will be a spring event? In other words, is there a correlation between the number of events in a given year, and whether or not there is a spring announcement the following year?

To get a handle on this, I ran a linear regression analysis on a table containing the number of events in a given year and whether or not a spring event was held the following year. The following table summarizes the results:

regression-analysis-spring-event-probability.jpg

Linear regression: it’s not just for grad school anymore

Because the slope value on the coefficient for the number events is negative, this shows us that the likelihood of a spring event decreases as the number of events increase. Because that value is -0.16, the regression is really saying that the likelihood of a spring event goes down only about 16 percent as the number of events go up. But, that’s based only on the data from the last ten years.

The real question is whether there’s a valid, statistical conclusion in there. To find out, we use a tool called a p-value. In our case, the p-value is 0.387-yada-yada. While significance in p-values do not necessarily have to be based on the five percent rule, the standard is to reject the null hypothesis when p<0.05. For our purposes, the null hypothesis means that these are not the stat droids you’re looking for.

Since the p-value of 0.387 is substantially higher than 0.05, it is not possible to reject the null hypothesis. What all of that math means is that there’s no way, statistically, to relate the number of events of a previous year to whether or not we’re going to have a spring event this year.

Has this season passed us by?

To answer this, let’s first look at the date Apple sent out invites, rather than the date of the announcement itself. On average, Apple sends out announcements for events 8.6 days before the event itself. In our case, then, it’s unlikely we’ll have an event before March 21st or so, but there is still an open window for a March announcement event.

Here’s when Apple typically holds events. As the following chart shows, over the last decade, Apple has only held announcement events in six months: January, March, April, August, September, and October. In the spring, the only months where we can expect new products to be announced by Apple are March and April, with a substantially higher probability that there will be a March announcement than an April announcement.

total-apple-events-by-month-held.jpg

Total Apple events by month held

There are some rumors out there that Apple is preparing an event for early April. The premise is that since Apple is expected to open its new spaceship campus in early April, a product announcement event might be held, for the very first time, in the new campus’ specially-built amphitheater.

This is possible. It’s also possible that Apple might break with tradition and hold an event in May if the campus, or the new amphitheater, is not completed in time to comply with historical precedent.

In any case, the Apple spring announcement season, such as it is, is still very much in play.

What will be announced?

Let’s go with the assumption that there will be a spring event, since we’re very much still within the window of probability. Will there be new Macs?

Separating out the burning need for upgrades to key systems like the Mac mini and the Mac Pro, of the last eight events where Apple has announced new Macs, only one occurred in the spring. It is therefore unlikely (even if long overdue) that there will be new Mac announcements.

Be careful, though, with such pronouncements. Analysis of numbers on a spreadsheet is only that. Just numbers. Apple has a very strong, almost urgent need to refresh its desktop and performance line of computers to prevent mass defection, so even though it’s relatively unlikely based on historical patterns, there could still be a Mac announcement coming in the next month or two.

Far more likely are iPad announcements. Of the last eight iPad announcements, half of them were introduced in the spring timeframe (I’m including one January announcement in this pack). In other words, while there’s only a 12 percent chance we’ll see new Macs soon, there’s a good fifty-fifty chance we’ll see updated iPads.

In addition, since there were five Apple Watch announcements in the last seven Apple events, there’s a good chance the Apple Watch will have some mention, if only to announce another incredibly overpriced band for people who should know better.

So, should we have new Apple products by now?

Of the five March events that Apple has held in the last decade, all but last year’s event occurred before March 13. So, if we were going to have an event in March, it’s statistically likely it would have occurred by now.

That means we’re probably waiting on an April event. Unfortunately for those hoping for new products, Apple has only held two April events in the last decade. The last one was seven years ago.

Given that Apple has been quite anemic with Mac updates, as well as with iPad updates, and the fact that the new Apple campus is due to open shortly, pundits and rumor mongers are certainly talking about a spring announcement. The math says the window is still open, but increasingly unlikely.

That said, predicting anything from a dataset this small is dicey at best. There’s just not enough data to form a real pattern. Apple is going to do what Apple is going to do, regardless of any previous years’ pattern of activity. And, finally, as we’ve come to know from the U.S. election last year, time-tested prediction methodologies break down when the future somehow decides not to mirror the past.

So take this column for what it is: a fun exercise in the ongoing punditry of predicting what Apple is going to do based on experience, insight, rumors, lies, damned lies, statistics, and absolutely no real clue.

You can follow my day-to-day project updates on social media. Be sure to follow me on Twitter at @DavidGewirtz, on Facebook at Facebook.com/DavidGewirtz, on Instagram at Instagram.com/DavidGewirtz, and on YouTube at YouTube.com/DavidGewirtzTV.



Source link