FanPost

Contest results: predict Red Sox first half OPS's

Back on March 26 I launched a contest to predict what the Red Sox's OPS's would be through June of this year. We got six entries total (including my own). Now it is time to provide the results!

According to the schedule, the end of June would have coincided with their first 81 games - exactly half a season. However, they ended up playing only 80 games through June. My question, however, asked where they'd be at the end of June, so that's what we're going with.

To get the numbers in the tables below, I took the difference between the actual and predicted OPS, player-by-player, and multiplied by the player's number of plate appearances. Then I summed the numbers across all the players to get each contestant's score. Lowest score wins.

And now, without further ado, here are the final scores:

Results through June 30

Contestant Score
theloganscorpion 174.5
trespada 178.7
Alfredoz 186.5
On to Lansdowne Street 196.5
tomisphere 210.2
Carras_Prodigy 291.8

Hmm, that can't be right, I came in fifth of six. Er, I mean, congratulations theloganscorpion on your winning score!

Wait, that's odd. I calculated the scores with just one game to go so that I could get a head start on writing my narratives around these results, and when I did, Alfredoz was the winner:

Results through June 29

Contestant Score
Alfredoz 188.4
theloganscorpion 189.0
trespada 192.0
On to Lansdowne Street 202.0
tomisphere 224.0
Carras_Prodigy 294.7

Could just one game cause Alfredoz to fall all the way to third? Well, admittedly, it was a marathon 11-inning game which saw double the usual number of plate appearances for the Red Sox players. But still.

I wonder what things would look like if we'd gone an extra game, so we got to a full 81 games, exactly half a season?

Results through July 1

Contestant Score
trespada 166.8
theloganscorpion 168.6
Alfredoz 184.2
On to Lansdowne Street 190.8
tomisphere 198.3
Carras_Prodigy 286.1

Now trespada is on top! We'd have had yet a third different winner!

These scores sure are volatile. But not so volatile as to overcome the bigger differences in score - only volatile enough to overpower the smaller differences in scores among the top three finishers. There is only one thing to do. Declare this a three-way tie!

Congratulations theloganscorpion, trespada, and Alfredoz, you all three are winners!

Now for some eye-opening fun. Back when I announced this contest in March, I captured ESPN's predicted OPS's for all of these players, as well as those given by the projection systems available on Fangraphs: ZiPS, Steamer, and ATC. The results: they smoked us all. Every projection system easily defeated the best of our scores:

Results for projection systems

Thru 6/30 Thru 6/29 Thru 7/1
ESPN 147.4 153.4 141.0
Steamer 149.0 159.7 139.6
ZIPS 149.1 162.4 139.8
ATC 162.4 168.3 156.3
Previous season's OPS 259.0 262.6 250.7

I tossed in predicting based on players' 2016 OPS's (and 2015 in the case of Pablo Sandoval), which gave comparatively poor results. There's something to these projections systems, I guess.

Originally I had suspected that I might be too optimistic in my predictions. Did that suspicion prove to be correct? I changed the calculation to include negative signs, with high predictions producing a positive number, and low predictions producing a negative one. Now, a positive overall score means that on the whole you were too optimistic; negative overall score means you were too pessimistic. These scores provided a pretty clear answer to my question:

Optimism Scores

Contestant Score
theloganscorpion 88.8
trespada 84.7
Alfredoz 101.2
On to Lansdowne Street 81.1
tomisphere 155.8
Carras_Prodigy 49.1
ESPN 20.9
Steamer 34.7
ZIPS -9.9
ATC 1.6
Previous season's OPS 75.5

I was by far the most optimistic - not even close. It paid off in giving me the best scores for Mitch Moreland and Jackie Bradley Jr., who everybody underestimated (actually ESPN edged me out on Moreland); my optimism meant that I underestimated them the least. It also gave me a pretty good score on Vazquez, though trespada had me beat there. But my overoptimism did not pay off much apart from those three players.

Interesting to me was that we were all more optimistic than the projection systems. I guess that comes with being Red Sox fans. Or it may just be that we were more likely to base 2017 numbers on 2016 numbers - the optimism score predicted by the Red Sox' 2016 numbers was similar to that of most of us, and a lot higher than all of the projection systems.

Of course, last year optimism might have paid off. By the midpoint of the season, most of the Red Sox hitters were exceeding expectations. Would we optimistic fans have beaten the projection systems through the first half of 2016? Quite possibly.

Also, when we look at the results player by player, we can see that most of the team has underperformed expectations. Six of the 14 have underperformed the expectations of every person and projection system, all 10 of us; two more underperformed all but one of ours. That's 8 overall underperforming, against four overperforming expectations (Mitch Moreland especially, as well as Jackie Bradley Jr., Christian Vazquez, and Xander Bogaerts) and two performing about as expected (Sandy Leon and Deven Marrero, though Marrero's current hot streak may quickly change that situation). The optimist in me looks at the fact that the Red Sox are in first place despite the underperforming, and noticing the offense seeming to get better, and, well, is optimistic about what that means for the rest of the season.

I wanted to make an analysis along the lines of that last paragraph into a full post contribution to this weekend's Fanpost Friday ("What do you think of this team at midseason? What strengths have you observed? What weaknesses can be patched?"), but I probably won't be able to manage the time for it. Instead, I'll just post a fuller breakdown of the results by player and predictor, below. You can then make your own conclusions from the data! Players are listed by number of plate appearances, with those with the most on top.

Red Sox OPS predictions and actual through June 30

Player 2016 ESPN ZIPS Steamer ATC Alfredoz On to Lansdowne Street trespada thelogan scorpion tomisphere Carras_ Prodigy Actual PA
Mookie Betts .897 .870 .865 .882 .871 .856 .910 .915 .930 .912 .910 .840 354
Xander Bogaerts .802 .784 .775 .802 .770 .852 .810 .760 .830 .796 .780 .826 326
Andrew Benintendi .835 .822 .787 .778 .813 .843 .865 .815 .810 .825 .840 .782 320
Mitch Moreland .720 .787 .720 .752 .746 .755 .759 .760 .770 .780 .720 .818 303
Dustin Pedroia .825 .775 .758 .771 .785 .805 .796 .795 .800 .811 .830 .746 276
Hanley Ramirez .866 .824 .830 .839 .813 .879 .836 .815 .860 .915 .885 .768 268
Jackie Bradley Jr. .835 .773 .775 .777 .769 .804 .785 .830 .800 .848 .805 .860 255
Chris Young .850 .778 .749 .760 .768 .780 .808 .840 .800 .822 1.015 .734 162
Sandy Leon .845 .697 .678 .665 .696 .690 .657 .721 .660 .732 .650 .695 157
Christian Vazquez .585 .624 .642 .686 .623 .733 .670 .698 .650 .681 .605 .700 154
Deven Marrero .298 .499 .587 .599 .507 .479 .510 .592 .635 .660 .260 .552 119
Josh Rutledge .733 .609 .655 .674 .677 .695 .685 .687 .600 .703 .505 .558 118
Pablo Sandoval .658 .737 .719 .754 .698 .790 .780 .724 .730 .811 .740 .622 108
Marco Hernandez .730 .690 .672 .713 .671 .785 .712 .754 .720 .744 .730 .628 60
Brock Holt .705 .718 .714 .708 .704 .682 .671 .706 .670 .708 .710 .449 19
Steve Selsky .810 .683 .684 .703 .673 .746 .589 .768 .700 .802 .710 .333 9
Rusney Castillo .625 .000 .671 .699 .633 .000 .667 .815 .665 .790 .580 .700 0
Blake Swihart .720 .683 .675 .711 .705 .763 .767 .752 .720 .748 .695 .700 0

Red Sox OPS overestimates times PA through June 30

Player 2016 ESPN ZIPS Steamer ATC Alfredoz On to Lansdowne Street trespada thelogan scorpion tomisphere Carras_ Prodigy
Mookie Betts 20.0 10.5 8.7 14.7 10.8 5.5 24.6 26.4 31.7 25.4 24.6
Xander Bogaerts -7.9 -13.8 -16.7 -7.9 -18.4 8.4 -5.3 -21.6 1.2 -9.9 -15.1
Andrew Benintendi 16.9 12.7 1.5 -1.4 9.8 19.4 26.5 10.5 8.9 13.7 18.5
Mitch Moreland -29.6 -9.3 -29.6 -19.9 -21.7 -19.0 -17.8 -17.5 -14.5 -11.4 -29.6
Dustin Pedroia 21.9 8.1 3.4 7.0 10.9 16.4 13.9 13.6 15.0 18.0 23.3
Hanley Ramirez 26.2 15.0 16.6 19.0 12.0 29.7 18.2 12.5 24.6 39.3 31.3
Jackie Bradley Jr. -6.3 -22.1 -21.6 -21.1 -23.1 -14.2 -19.0 -7.6 -15.2 -3.0 -13.9
Chris Young 18.8 7.1 2.5 4.2 5.5 7.5 12.0 17.2 10.7 14.3 45.5
Sandy Leon 23.6 .3 -2.6 -4.7 .2 -.8 -5.9 4.1 -5.5 5.8 -7.0
Christian Vazquez -17.7 -11.7 -9.0 -2.2 -11.9 5.1 -4.6 -.3 -7.7 -2.9 -14.6
Deven Marrero -30.2 -6.3 4.2 5.6 -5.3 -8.7 -5.0 4.8 9.9 12.9 -34.7
Josh Rutledge 20.6 6.0 11.4 13.7 14.0 16.1 15.0 15.2 4.9 17.1 -6.3
Pablo Sandoval 3.9 12.4 10.5 14.3 8.2 18.1 17.1 11.0 11.7 20.4 12.7
Marco Hernandez 6.1 3.7 2.7 5.1 2.6 9.4 5.1 7.6 5.5 7.0 6.1
Brock Holt 4.9 5.1 5.0 4.9 4.8 4.4 4.2 4.9 4.2 4.9 5.0
Steve Selsky 4.3 3.1 3.2 3.3 3.1 3.7 2.3 3.9 3.3 4.2 3.4