How Atlanta United can win the Supporters’ Shield

Despite the team coming up short in the consensus biggest game of the season, both Five Thirty Eight and American Soccer Analysis calculate the current likelihood of Atlanta United winning the Supporters’ Shield at 73%.

While both models use sound, data-driven techniques to simulate the most likely final season table — and while these projections have been a helpful point of reference throughout the season — the trouble is that the few number of games remaining skews the “truthiness” of the situation. Because of the exceedingly slim margins currently separating Atlanta United and Red Bulls, the cited 73% figure may swing wildly with each match outcome from here on out. Accordingly, it might be better to discard the overall percentage and just talk through the discreet specifics of this thing.

With 3 games remaining and a max of 9 points left to be claimed by either side, Atlanta leads New York by a point. The next two Atlanta games are home to two bad teams in New England and Chicago while the next two NYRB games are on the road, first to a bad (but transitioning) San Jose and then to a legit playoff team in Philadelphia. On Decision Day 2018™, Atlanta travels to a tricky Toronto and Red Bulls host the hopeless Orlando City.

The earliest scenario in which Atlanta could clinch would be after 2 games, as a 4 point lead heading into the last week would be the ball game. They could do this by winning the next two matches so long as Red Bulls either lose one or draw both. Atlanta could also clinch before traveling to Toronto by winning one match if Red Bulls lose both (unlikely). Red Bulls could clinch as early as the penultimate match week with wins in the next two games and Atlanta failing to pick up a win (this would put the gap at either 4 points, or 3 points with the wins tiebreaker on lock - see appendix).

It’s tempting to expect the best two teams in MLS (and two of the best in MLS history) to win out. Don’t. American Soccer Analysis has both teams collecting all 9 points in 23 out of 1,000 simulations of their model (2.3%). Did I just tell you not to put too much weight into probabilities at this point in the season? Well, yeah. But this specific one is pretty extreme. It’s the sort of odds that C3P-0 reads off to Han Solo in an asteroid field. Side note: remember when USMNT was 93% to qualify for the 2018 World Cup with 1 match left to play against the worst team in the Hex?

I’m sorry. I am trying to delete it.

As gutting as the match was this past weekend, Atlanta should be in the driver seat here, up a point and hosting two bad teams in the Benz while Red Bulls scrap it out on the road (especially in Philadelphia).

Both contenders should be heavily favored in all home matches. But, of the three remaining away matches across both contenders, only the PHI-NY match features one of the two going up against a team averaging more points per game (1.7 ppg) at home than the contender is averaging per game on the road (1.5 ppg). While the Union likely won’t be favored in this one, this is as close as it gets. Anything can happen. I would watch this game if it was on TV.

You don’t have to agree with me, but in my opinion, this is not really a sport where you should talk about “controlling your own destiny.” Draws are a problem, because not only do they make the outcomes tertiary and not binary, but it’s also unclear what causes them.

Sometimes a team is playing for a draw. Other times, draws just sort of happen naturally in the wild — chaotic conflict between free will and space time. An endless ocean landing on an endless desert. Scoring goals is very hard yet they’re all we have to determine match outcomes. There’s no such thing as destiny and time is a flat circle. Having said that, the Five Stripes hosting two straight eliminated teams in the air conditioned Death Star with Julian Gressel on the right is as close as it gets.

If we’re trying to predict what the other guys might do, New York has averaged just under 6 points every 3 games in 2018. Their most common points return from a 3 week stretch this year was 6 points. Six seems like a good prediction, but I think 7 is very, very possible. It may well come down to goal difference — goal difference which Atlanta may desperately need to build some cushion around in the next two games. Looking at you, New England.

Appendix: Tie Ball Game

If the regular season ends and two teams are tied on points, the Shield will still be awarded to one of them.

For this to happen, Red Bulls have to get a point back somehow. Probably the most likely scenario is that Red Bulls grabs 2 wins and a draw and Atlanta 2 wins and a loss. Though the possible scenarios are myriad (NY 2 wins vs ATL 1 win, 2 draw for instance). In the unlikely event of a tie, here is a quick run-down of the MLS tie-breaker rules (I swear 1 and 2 aren’t flipped here) and where we stand:

1. Total number of wins - currently even, unlikely to decide it since level points will likely (tho not necessarily) include the same number of wins from here on out
2. Goal Differential - Atlanta United currently +2. Given ATL currently up a point, dropping a point on NYRB would also require a change in GD, so this tiebreaker might very much be in play. If in fact, we go to tiebreakers, this could be anyone’s game, or it could be level.
3. Goals For - Atlanta currently +8 on NYRB. If New York scores 8 more goals than Atlanta from here on out, they’re probably taking the Shield outright, so I’d suggest Atlanta has this one locked down in the case of tiebreaks, and we wouldn’t go any further down the list. Stop reading unless you’re a completionist and are looking for the last golden skulltula.
4. Fewest Disciplinary Points - likely not used. Romantic overtures aren’t included in this.
5. Away Goals Differential - likely not used (though Atlanta leads by 6)
6. Away Goal For - likely not used (though Atlanta leads by 5)
7. Home Goals Differential - likely not used (why would this decide the tie if 2 and 5 don’t?)
8. Home Goal For - likely not used (thought Atlanta leads by 3)
9. Coin Toss (tie of two clubs) or Drawing of Lots (tie of three or more clubs) - likely not used, but this is always tails.