How reliable is Ming-Chi Kuo?

KGI Securities' mysterious analyst scores 44% in Apple 3.0's first-ever analyst truthiness test.

AppleInsider refers to him as "noted" and "well-connected." 9to5Mac has taken to using the honorific "reliable." MacRumors calls him "the respected Ming-Chi Kuo."

Others take a dimmer view of what Kuo does. Asymco's Horace Dediu calls it industrial espionage.

We know almost nothing about KGI Securities' star Apple analyst except that he has made dozens of bold, often quite specific predictions about future products over the past six years. MacRumors lists 91 of them, starting at DigiTimes in April 2010.

Kuo's most recent prognostication, as reported by 9to5Mac, is that the iPhone 7 will lack "attractive selling points" and that Apple's flagship will end 2016 as the only top-five smartphone brand to register a year-over-year sales decline.

Do we believe it? Just how reliable is the reliable Mr. Kuo?

To try to shed some light on that, I've subjected Kuo's predictions to what I'm calling the Apple 3.0 truthiness test, rating the accuracy of his past predictions on a scale of 0 to 100.

I didn't test all 91 in the MacRumors list, but I tried to pick out a representative sample from the past two years, starting with...

Oct. 1, 2015

April 22, 2015

April 2, 2015

  • Prediction: Force Touch expected to bring about the "most significant change yet"; so significant that the first iPhone with it could be named the iPhone 7, not the iPhone 6s
  • What happened: The iPhone was called the 6s, as expected, and Force Touch has yet to prove its significance
  • Truthiness: 25%

February 6, 2015

Nov. 24, 2015

July 13, 2014

April 9, 2014

Average: 44%

The idea here is not to pick on one particular analyst, but to find a way to hold the Ming-Chi Kuo's of the Apple world to some kind of objective standard.

This is my first stab at something I haven't seen done before. If you see a better way to do it, I'm open to suggestions.

UPDATE: I've upgraded Kuo's score to 44% from 41% thanks to reader Keith Hope's correction in the comment stream. See also the 2015 Cult of Mac profile he spotted, with its rare photos of Kuo, some even more over-the-top encomiums ("the best Apple analyst on the planet"), and a pretty good evaluation of the accuracy of his earlier forecasts.


  1. David Emery said:
    How do you calculate/assess partial scores? For this to work, the methodology needs to be repeatable and should be doable by anyone.

    April 24, 2016
    • You’ve put your finger on the weakness of my system. I was following William Burroughs’ dictum; be just if you can, but if you can’t be just, be arbitrary

      April 24, 2016
  2. John Kirk said:
    I like it. The system isn’t perfect, but it’s still perfectly fine. 🙂

    April 24, 2016
    • Tom Sidla said:
      I agree with John. In fact, I bet PED’s method is more accurate than Ming-Chi Kuo’s.

      Next up, Trip Chowdry.

      April 25, 2016
  3. Aki Järvilehto said:
    Thanks for a very good summary! Would love to see more of these.

    April 25, 2016
  4. Kirk Burgess said:
    I think it’s worth separating his product feature predictions to that of his unit sales predictions. He gets some good scoops on product features, but his unit sales estimates are usually way off.

    April 25, 2016
  5. Ken Cheng said:
    Off-hand, I think the normal correct prediction rate for analysts is only 5%, 1 in 20, so 44% is fairly good.

    April 25, 2016

Leave a Reply