Pittsburgh Pirates Prediction Machine – What a Crazy Season

It’s been an up and down year for the Pittsburgh Pirates and also for the Prediction Machine.

We started this regular article to focus attention on specific Pirate performances for the upcoming week by making predictions about them.  Now that the end of the season is upon us, we pause to take a look back at the year.

But first, we have one more week’s worth of predictions to  review.  Last week we made four predictions and only one of them was correct.  Our performance pretty much mirrored the Pirates’ as they got swept by the Cardinals to finish their disappointing season.

Last Week’s Predictions:

We expected Andrew McCutchen to finish strong:  Heading into the final week, it sure looked like McCutchen was finding his bat.  He came into the week hitting .312 for the month of September and we expected this to continue.  Instead, Cutch finished the season much like he started it, going 4 for 22 for an average of .182.  This prediction was dead wrong.

We thought that Jameson Taillon versus Jake Arrieta pitcher’s would be a pitcher’s duel:  We expected both Taillon and Arrieta to hold the other team to one run or less.  Taillon lived up to his end of the bargain going six innings and allowing only a single run.  However, Arrieta fell apart in the 4^th and 5^th innings, allowing 3 runs in each, making our prediction wrong.  One of the silver linings of the season for the Pirates is they seem to have finally solved Arrieta.

We guessed the Cardinals would clinch a wild card berth against the Pirates.  Even though the Cardinals swept the Pirates, as expected, the San Francisco Giants had other plans.  The Cardinals played well down the stretch but ended up missing the playoffs by just one game.  This prediction was wrong.

We predicted the final record for the Pirates:  Actually, we said it would be between 79-83 and 81-81, but we didn’t count on a tie being in there.  If we take the tie out of the equation, then the winning percentage is very close to what we predicted, so this prediction was more or less accurate. 

Overall, we went 1 for 4 with our predictions for the week and were really off base on a couple of them.

Prediction Machine Season in Review:

Here at the end of the season, we thought it would be fun to look back at the various prediction machine articles and analyze our performance, much like we would do for a player.

Overall, we made 80 predictions over the course of the season.  Five of those were untestable as the player in question wasn’t able to perform for various reasons.  Of the remaining 75 predictions, 45 of them were correct for a .600 “batting average”.  In baseball, if you win 6 out of 10 games, you will be in the playoffs, but in the NL Central these days it would probably be in the wild card game.  We have higher aspirations than that.

Our accuracy record over time was fairly consistent, with our best month in July at 65% and our worst month in August at 54%.  In early September we had our best streak with eight straight correct predictions.

When we made a positive prediction (the player or team would do well), our record was decidedly worse that when we had negative expectations.  We tended to stay positive, with two-thirds of our total ending up in this category.  However, we were barely over .500 on those predictions.  On the other hand, we were right 64% of the time when we said something bad would happen.  The prediction machine is developing a platoon split!

The two hardest players to predict were Andrew McCutchen (1 for 5) and Jameson Taillon (0 for 3).  This shows the dangers of extrapolating past performance in baseball.  Cutch did far worse than previous years and Taillon outperformed all expectations.  The easiest player to forecast was Jordy Mercer.  We were a perfect 4 for 4 on Mercer predictions.

Our single best call was, without question, back in late June when we said Chad Kuhl would beat Clayton Kershaw.  That was fun!  Our most embarrassing call was probably when we said in late August that Tony Watson would quiet his doubters with no blown saves and yield less than one earned run.  That one didn’t develop like we expected.

Two predictions for next season

It has been a lot of fun writing the Pittsburgh Pirates Prediction Machine this season.  We hope you enjoyed reading it as much as we did writing it.  As we head into the off-season, we will leave you with two prognostications for next season.

Gerrit Cole will not be the opening day starter:  The Pirates have finally figured out that they need a lockdown pitcher to headline their rotation.  Gerrit Cole fits in as a very nice number two starter.  If the Pirates acquired someone like Julio Teheran of the Atlanta Braves, then they could boast a one-two punch to rival the Cub’s Jake Arrieta and Jon Lester or the National’s Max Scherzer and Stephen Strasburg.

Andrew McCutchen will stay a Pirate, but will start in left field:  This isn’t really controversial as this has been discussed and dissected already.  We just wanted to get it in writing that we expect it to actually happen.

Thanks again and see you next season!