Analytics & Payroll loses to Talent & Drive

Congratulations to the Houston Astros for their victory in the 2017 MLB World Series game seven last night.  I live in Southern California, but I've not embraced the SD Padres as my "home team" yet, since I am a life-long Chicago Cubs fan.  The Cubs won last year...  after 108 years of trying... which apparently opened a trans dimensional portal that let a racist, xenophobic, ignorant, non-reading 10 year old in a 70 year old's carcass, with Vladimir Putin's direct assistance, become President of the United States.  The Cubs can wait another few dozen years before trying to win again - I don't want them to become the universally hated NY Yankees who have been to 40 of the 113 World Series (winning 27 of them).  The Houston Astros had never won a World Series, and it'd been 55 years since they were formed as a team...  so it was about time (even though I will ALWAYS think of the Astros as a National League team, as they've been the majority of my 1/2 century of life).

My actual Cub's hat, bought before they won the World Series in 2016

While the Houston Astros Did rely partly upon analytics, they had been drafting and recruiting some of the best players in major league baseball for years.  They wanted it.  They worked hard for it.  And their manager "could talk analytics all day long, but more often went with his gut and what the situation required" (I heard during Game 6's broadcast, Joe Buck saying it, I think).  The Brooklyn Dodgers... I mean LA Dodgers, paid millions of dollars for analytics, had The Largest payroll in all of MLB, were in the 2nd largest media market, but had worked out a despicable cable deal that MOST of the Dodgers home games Could Not Be Televised in LA when the stadium in Chavez Ravine that seats 56,000 was not sold out - the largest stadium in all of MLB, beating the damn Yankees by 3000 seats.  When the camera panned over the crowds during foul balls, you could see 1/3rd of the most expensive box seats showing EMPTY blue seat backs during Game Seven of the World Series!  EMPTY Seats!?!?  Wrigley Field only had empty seats during the World Series in 2016 from people using the bathroom, and it was < < much less than 5% of all box seats, even when some of those seats were going for $2000 each.

Thanks Jen for the gif!

So the Cubs had been the proverbial underdogs, and they won last year.  The Astros were the underdogs this year, and they won.  I'm a habitual under-dog supporter...  which I partially attribute to going to more than a dozen games at Wrigley as a kid (two of the three games I went to at Comiskey Park ended up with our vehicle being broken into... and the third one, my older brother and his friends picked up a Ford Escort and turned it 90 degrees onto the side walk so that our car would fit... they were very big boys).

Now,  you maybe asking yourself: "Why does JustJoeP oppose analytics so much?" Well, I credit that to my 20 years of working at one of the world's largest companies, where we made model after model to predict how the systems and components would perform in the field, how well they would be able to be repaired, how many times they could be repaired, when they'd need to be repaired, and when they would stop working for our customers.  It was Big Data, little data, good data, bad data. comprehensive data, partial data... and all that data was scrutinized by dozens of mathematics PhDs, data analysts, and engineers to come up with complex analytics that were supposed to add tremendous value, both for the company I worked for And for their customers.  Sadly, after more than a decade of trying to apply these analytics, revising and revising them, combing and scrutinizing the data and transfer functions repeatedly... the analytics Could Not Replace subject matter expertise; AKA experience.

Subject Matter Experts (SMEs) like myself were leveraged to make better models, sort & parse data, review repair reports, outage reports, customer issues, design reviews, root cause analysis... but no matter how hard we tried, and how many times we revised models, they still were far off the mark, much too conservative, or far too liberal, rarely ever accurate on predicting reality.  It was a Very Complex environment, with a very complex product and no two customers were the same or operated identically, that the analytic models Could Not Cope With the all of the Variation accurately - much like MLB!  Sure, if you start and stop very few times, and operate in a gingerly, gentle, non-stressful mode, the models worked marvelously!  But that was less than 5% of the all the customers, so the models and their predictions were virtually worthless in the big picture.   But saying that, admitting it publicly, was as deadly as saying that the Emperor had no clothes, when 99.997% of the sycophants had been telling the Emperor how amazing and awesome his clothes really were.

On a post-script, I do not and will not watch Wisdom of the Crowd as Crowds, by definition, are pretty emotional, stupid / ignorant / misinformed, reactionary, and prone to herd mentality.  Similarly, the same goes for the majority of analytics.  In limited applications, for narrow goals, Big Data Can Be useful, and perhaps partially accurate.  But in the Vast Majority of situations, applications, and concepts, data analytics usually don't have accurate predictions or reliable results.  Junk in = junk out.  Getting all the data "in" to be pristine, relevant, and representative of what is going to happen in the future is expensive, time consuming, and virtually impossible.  Anyone who tells you differently, is lying.