Minnesota Twins predictions from FiveThirtyEight and ESPN are very different
In advance of Opening Day for teams that aren’t Seattle and Oakland, sports websites, including the one you’re reading, are releasing their 2019 MLB predictions. Two popular sites, FiveThirtyEight and ESPN, published theirs Wednesday morning. The sites reached their conclusions differently - FiveThirtyEight simulated the season 100,000 times and relies on analytics for statistical projections, while ESPN polled 31 in-house baseball experts - but did they end up with the same results? Let’s see.
(If you’d like to read them as well, here are the links to FiveThirtyEight’s and ESPN’s predictions, as well as those to the former’s AL “Favorites and Breakout Picks” and the latter’s power rankings. I’ve included each in my comparison.)
Division results
As expected, both FiveThirtyEight and ESPN project the Cleveland Indians to claim the AL Central. FiveThirtyEight predicts Cleveland will finish the season with a 95-67 record, tied for third-best in the major leagues, while ESPN’s experts are similarly bullish, with 28 of their prognosticators selecting the Indians. Their power rankings (ranked by poll, but with predicted records and World Series odds calculated by one person) put Cleveland seventh in the league, winning the division with a 92-70 mark.
While there’s not a lot of divergence regarding Minnesota’s placement, there is when it comes to record. Sorting the teams by their ELO rating, FiveThirtyEight puts the Twins 16th, second in the AL Central with an average but still respectable 84 wins and 78 losses. ESPN also guesses that the Twins will come in second - only three experts give the Twins the Central crown - and slots them similarly in their power rankings (17th), but projects them to win just 78 games, exactly the reverse of FiveThirtyEight’s average simulation.
Why the six-game difference - a difference that shifts Minnesota from a winning season to a losing one? We have access to more details of FiveThirtyEight’s predictions than ESPN’s, so it’s possible to extrapolate a guess.
Option 1: Intradivisional
One detail we can directly compare is intradivisional success. By comparing records within the AL Central, if the other teams benefit while the Twins lose more, we can assume that in ESPN’s prediction, the Twins do worse within the division and thus end with a losing record. But looking at those teams’ marks, we see something else:
In ESPN’s prediction, the White Sox and Tigers pick up an extra game, but the Indians lose three, meaning the six lost Twins games must come from outside the division.
Option 2: Top teams
Another likely candidate is poor performance against the top teams in the American League. After all, the AL Central is widely considered the weakest division in baseball, so if the Twins are to lose six more games, it seems reasonable that those losses will come against strong teams from stronger divisions.
Excluding Cleveland (5th by FiveThirtyEight’s ELO, 7th in ESPN’s power rankings), five AL teams are listed above the Twins by both sites:
I’m also including the Los Angeles Angels in the comparison, as while FiveThirtyEight ranks them 17th, one spot behind the Twins, ESPN (whose model is in focus right now) gives them the 16th spot, swapping them with Minnesota.
Comparing those six teams’ projected records, we see the following:
Here we see significant differences, but like before, we don’t see those that would explain the Twins’ drop in wins per ESPN’s predictive model. While the Angels gain seven wins and the Rays lose five, these teams in total drop five wins. If I’d included Cleveland, the drop would have been eight. Whichever factor results in the Twins losing more games should show a gain in wins for the other team. So we must cross this off and move on.
Option 3: Weaker competition
The only factor remaining in the American League is weaker competition, that the Twins will perform worse against the weaker teams in the AL and demonstrate that they can be as mediocre as anyone. Both models rank seven teams below the Twins. Three of those are AL Central competition, so we will exclude them for now. FiveThirtyEight ranks the Angels below Minnesota, but it’s ESPN we’re looking at, so I will leave them out as well.
The teams we are now examining are thus:
And their predicted records are:
Finally, ESPN totals more wins than FiveThirtyEight... but just a single W. And the Orioles gain four on their own. Something’s up here.
Only one variable in the Twins’ schedule remains:
Option 4: Interleague
The Twins play 20 games against National League opponents during the 2019 season:
Comparing their records, we have:
At last, ESPN bumps four wins off the Twins. It’s not six, but it’s the closest we’ve gotten. As this final option accounts for the remaining teams on their schedule, it’s as far as we will get.
Of course, not all extra wins or losses will come against Minnesota, but it’s fascinating how two teams can be considered in the middle of the pack, yet if one prediction were to come to pass, it would be considered far superior to the other based on a difference of just six games.
