Projections

Through the San Jose - Sporting KC game on Friday night, the model has "succeeded" 57.4 percent of the time during regular season play in 2015. That is, since we began uploading projections in week 25, the model's most probable outcome matched the game's actual outcome (win, lose, or draw) 47 of 82 times.

2015 Week 34 Matchups

Home Away Home Win% Away Win% Tie%
SKC COL 0.550 0.196 0.254
CHI NYRB 0.335 0.432 0.233
CLB DCU 0.620 0.180 0.200
MTL TOR 0.493 0.265 0.242
PHI ORL 0.509 0.244 0.247
NYC NE 0.388 0.377 0.234
SEA RSL 0.550 0.211 0.239
VAN HOU 0.499 0.226 0.275
FCD SJ 0.528 0.220 0.252
POR COL 0.499 0.218 0.283
SKC LA 0.465 0.279 0.255


2015 Week 33 Matchups

MLSsoccer.com posted a summary of what this week's games mean in terms of playoffs. Here are the probabilities of those potential outcomes.

Home Away Winpct Losspct Tiepct
TOR NYRB 0.465 0.308 0.227
FCD VAN 0.483 0.249 0.268
RSL POR 0.397 0.336 0.267
ORL NYC 0.555 0.224 0.220
SJ SKC 0.454 0.277 0.270
TOR CLB 0.569 0.223 0.207
NE MTL 0.499 0.253 0.248
RSL FCD 0.383 0.356 0.260
HOU SEA 0.504 0.230 0.266
LA POR 0.553 0.208 0.239
NYRB PHI 0.674 0.147 0.179
DCU CHI 0.457 0.301 0.242

2015 Week 32 Matchups

Not a lot of games this week. Montreal may have the highest away win probability we've seen this season, and they're nearly 2:1 favorites to get a least a point. Of course, earlier in the week they play the Red Bulls, and their chances at points are quite the opposite. Vancouver has coin-flip odds of getting all three points, which would probably even them up with FC Dallas for top-seed probability in our playoff seeding projections.

Home Away Home Win% Home Loss% Tie%
NYRB MTL 0.639 0.169 0.192
VAN FCD 0.498 0.237 0.265
COL MTL 0.371 0.341 0.288

 

2015 Week 31 Matchups

Home Away Home Win% Home Loss% Tie%
DCU NYC 0.494 0.279 0.227
TOR PHI 0.624 0.172 0.203
NYRB CLB 0.666 0.163 0.171
ORL MTL 0.417 0.318 0.265
CHI NE 0.425 0.328 0.247
POR SKC 0.443 0.285 0.272
SJ VAN 0.427 0.288 0.285
FCD HOU 0.523 0.226 0.252
COL RSL 0.464 0.267 0.269
SEA LA 0.424 0.312 0.264

 

2015 Week 30 Matchups

Many of the league's worst teams find themselves on the road this weekend, and thus home probabilities are soaring in week 30. You'd think it would be a gimme week for the model, but it's MLS, so not even one of the best teams is immune from a 3-0 drubbing at home. Oh, and apologies for last week's data debacle!

Home Away Home Win% Home Loss% Tie%
MTL CHI 0.578 0.201 0.221
HOU SKC 0.436 0.291 0.274
NYRB ORL 0.703 0.127 0.170
TOR CHI 0.628 0.173 0.199
MTL DCU 0.624 0.164 0.211
NE PHI 0.561 0.216 0.222
CLB POR 0.548 0.225 0.227
HOU COL 0.487 0.227 0.285
VAN NYC 0.634 0.159 0.207
LA FCD 0.536 0.225 0.239
SKC SEA 0.554 0.202 0.244
SJ RSL 0.532 0.223 0.244


2015 Week 28 Matchups

The best chances at road wins this week are where you'd expect them to be: in Philly and in Orlando. The Timbers could really use a home win; with San Jose hosting Seattle, Portland would be guaranteed to create some space for the last playoff spot with three points. At 68.6%, the Red Bulls have the highest probability to win of any team over the past three weeks. 

Home Away Home Win% Home Loss% Tie%
VAN COL 0.545 0.192 0.263
POR SKC 0.456 0.285 0.259
NYRB CHI 0.686 0.147 0.167
PHI CLB 0.462 0.311 0.227
HOU RSL 0.571 0.188 0.241
FCD NYC 0.657 0.156 0.187
COL DCU 0.488 0.247 0.265
SJ SEA 0.489 0.245 0.266
LA MTL 0.570 0.213 0.217
TOR NE 0.540 0.242 0.218
ORL SKC 0.409 0.325 0.266

2015 Week 27 Matchups

It's a short week, and there don't appear to be any surprises here, except maybe Columbus' large advantage over FC Dallas. This seems paradoxical, since FCD has the edge in both GD and xGD. But the model picks up a few hard-to-see things, like goals scored are easier to predict that goals allowed. Columbus is in the top three for both goals scored and expected goals scored, and is playing at home. 

Home Away HomeWinpct HomeLosspct Tiepct
NE ORL 0.593 0.185 0.222
MTL CHI 0.554 0.216 0.230
SEA TOR 0.446 0.307 0.247
SJ PHI 0.498 0.258 0.244
CLB FCD 0.563 0.215 0.223

2015 WEEK 26 MATCHUPS

This is the closest we've come to seeing an away team with a better chance to win than the home team. Between the Red Bulls and the Fire, it might as well be a toss up as to who wins--given it's not a tie. Sporting also has a relatively high road win probability in Denver this weekend. Odds are that at least one of those two teams wins in Week 26. 

Home Away HomeWinpct HomeLosspct Tiepct
CHI NYRB 0.397 0.360 0.243
COL HOU 0.419 0.288 0.293
SJ LA 0.446 0.296 0.258
TOR MTL 0.632 0.163 0.205
NYC CLB 0.522 0.267 0.211
PHI NE 0.526 0.243 0.231
ORL CHI 0.502 0.257 0.241
HOU VAN 0.485 0.243 0.272
COL SKC 0.409 0.321 0.270
FCD RSL 0.616 0.169 0.216
SEA POR 0.456 0.280 0.264
NYRB DCU 0.646 0.163 0.191

2015 Week 25 Matchups

Home Away HomeWinpct HomeLosspct Tiepct
CLB NYC 0.659 0.152 0.189
SKC SJ 0.607 0.161 0.231
POR HOU 0.487 0.238 0.276
VAN FCD 0.517 0.234 0.249
TOR ORL 0.674 0.134 0.192
DCU SJ 0.497 0.251 0.253
CLB SKC 0.525 0.237 0.238
MTL PHI 0.602 0.184 0.213
CHI COL 0.529 0.212 0.258
RSL SEA 0.462 0.274 0.264
LA NYC 0.654 0.154 0.192


2014 MLS Cup Finals Probabilities (Dec. 7)

Team Semis Finals MLSCup
LA 1.000 1.000 0.775
NE 1.000 1.000 0.225

MLS CUP PROBABILITIES (OCT. 29 - DEC. 7)

Team Semis Finals MLSCup
LA 1.000 0.745 0.578
NE 1.000 0.867 0.200
SEA 1.000 0.255 0.183
NYRB 1.000 0.133 0.039

Semis: Probability of reaching the MLS Cup semifinals, known as the Conference Finals. 

Finals: Probability of reaching the MLS Cup Final.

MLSCup: Probability of winning the MLS Cup.

ROUND 1, LEG 2 (NOV. 8 - NOV. 10)

Home Away HomePct
NE(4) CLB(2) 0.970
SEA(1) FCD(1) 0.685
LA(0) RSL(0) 0.645
DCU(0) NYRB(2) 0.115

Round 1, Leg 1 (Nov. 1 - nov. 2)

The chart below shows the probabilities of various spreads after the first leg of the MLS Playoffs. Probabilities were derived from teams' Expected Goals data from the 2014 season, and generated following a diagonally inflated bivariate Poisson regression model. 

Home Away H4 H3 H2 H1 H0 A1 A2 A3 A4
NYRB DCU 0.061 0.119 0.186 0.222 0.203 0.108 0.044 0.014 0.003
CLB NE 0.034 0.082 0.159 0.229 0.243 0.145 0.064 0.021 0.005
FCD SEA 0.031 0.079 0.157 0.232 0.249 0.148 0.064 0.020 0.005
RSL LA 0.011 0.037 0.098 0.192 0.270 0.207 0.114 0.047 0.015

Notes: In the playoffs, an away team may be even more conservative than it would typically be during the regular season, and thus it is possible that the actual probabilities of ties and one-goal deficits are higher than the model foresees. 

Play-in Games (Oct. 29 - Oct. 30)

Home Away HomeWin AwayWin
FCD VAN 0.698 0.302
NYRB SKC 0.712 0.288

 

MLS Cup Probabilities (Oct. 29 - Dec. 7)

Team Semis Finals MLSCup
LA 0.691 0.442 0.334
SEA 0.541 0.238 0.176
NE 0.529 0.283 0.091
RSL 0.309 0.139 0.085
CLB 0.471 0.239 0.073
NYRB 0.421 0.219 0.072
FCD 0.302 0.119 0.067
DCU 0.366 0.151 0.048
SKC 0.213 0.108 0.030
VAN 0.157 0.062 0.024

Notes: These probabilities are based on regular season data. While they were published on 11/11/14--during the playoffs--they are completely unbiased toward the playoff results that had already happened. 

Week 33 (Oct. 20 - Oct. 26)

Home Away hwin awin tie spread
RSL CHV 0.688 0.102 0.209 1.21
CHI HOU 0.444 0.270 0.286 0.37
SEA LA 0.481 0.248 0.270 0.42
MTL DCU 0.416 0.246 0.338 0.27
NE TOR 0.383 0.279 0.338 0.12
FCD POR 0.558 0.175 0.267 0.72
VAN COL 0.555 0.182 0.263 0.69
CLB PHI 0.305 0.389 0.306 -0.16
CHV SJ 0.404 0.254 0.341 0.27
SKC NYRB 0.544 0.219 0.238 0.68

Notes: Since these probabilities are based on Expected Goals 2.0, the results may not match what you'd expected if you look at the Playoff Push page.

Week 32 (Oct. 13 - Oct. 19)

Home Away hwin awin tie spread
HOU NE 0.566 0.163 0.272 0.76
POR RSL 0.273 0.460 0.267 -0.31
TOR MTL 0.544 0.179 0.277 0.72
COL FCD 0.429 0.282 0.289 0.27
DCU CHI 0.399 0.279 0.322 0.21
PHI SKC 0.510 0.222 0.268 0.54
SJ VAN 0.313 0.412 0.275 -0.13
NYRB CLB 0.629 0.117 0.255 0.98
LA SEA 0.573 0.199 0.228 0.76

Notes

*These projections do not include any controls for roster. Thus, teams that are resting stars may be overvalued.

*The model has yet to correctly pick an away win, but it has two more chances this week with RSL over Portland and Vancouver over San Jose--the exact opposite of what Timbers fans want to see.

Week 31 (Oct. 6 - Oct. 12)

Home Away hwin awin tie spread
TOR HOU 0.501 0.253 0.246 0.54
POR SJ 0.394 0.283 0.323 0.26
SKC CHI 0.609 0.153 0.238 0.91
SEA VAN 0.594 0.186 0.220 0.81
MTL NE 0.434 0.264 0.302 0.36
PHI CLB 0.616 0.130 0.254 0.94
NYRB TOR 0.531 0.175 0.293 0.60
RSL SJ 0.666 0.111 0.222 1.14
CHV COL 0.345 0.357 0.298 0.04
HOU DCU 0.551 0.156 0.293 0.71
FCD LA 0.457 0.243 0.299 0.31

Notes

*Though LA is the model's powerhouse, FC Dallas ranks not too far behind them at fourth in the league in xGD during even gamestates at home. This is why FCD gets the nod in this game.

*Portland has played poorly at home, and ranks dead last in xGD during even gamestates at Providence Park. The Timbers still have a plurality of the probability, but it's not going to be easy.

*Colorado is the only favored away team this week, though just barely over Chivas. We saw Chivas beat a strong RSL side this past week, so if you're thinking that teams out of the playoffs roll over and die, take my anecdote and smoke it!

Week 30 (Sept. 29 - Oct. 5)

Home Away hwin awin tie spread
PHI CHI 0.591 0.156 0.253 0.87
DCU SKC 0.307 0.398 0.295 -0.18
NYRB HOU 0.532 0.216 0.252 0.62
VAN FCD 0.459 0.250 0.291 0.33
NE CLB 0.433 0.214 0.354 0.32
LA TOR 0.612 0.149 0.239 0.92
SJ POR 0.436 0.221 0.342 0.38
COL SEA 0.397 0.307 0.296 0.13
CHI MTL 0.477 0.219 0.304 0.50
CHV RSL 0.270 0.458 0.272 -0.33

Notes

*Two away teams have the upper hand this week in Sporting KC and RSL. 

*The model is not convinced that Colorado is dead. The fact that the Rapids have the most underperforming underperformer by goals scored versus expected goals is just one indication that positive regression is likely. 

Week 29 (Sept. 22 - Sept. 28)

Home Away hwin awin tie spread
FCD SEA 0.506 0.241 0.254 0.46
SKC NE 0.643 0.133 0.224 0.97
TOR POR 0.542 0.153 0.305 0.74
DCU PHI 0.323 0.380 0.298 -0.04
SEA CHV 0.661 0.099 0.240 1.08
VAN RSL 0.467 0.255 0.278 0.36
CLB MTL 0.404 0.270 0.326 0.30
COL SJ 0.537 0.158 0.305 0.70
HOU CHI 0.506 0.179 0.315 0.58
LA NYRB 0.627 0.169 0.204 0.84

Notes

*The highest probability of a win goes to Seattle in its second game of the week, which shouldn't surprise anyone considering its opponent.

*Philadelphia is the only away team this week to enjoy a greater probability of winning than its host. Philadelphia is above-average, but this has more to do with the fact that DC United's shot statistics are pretty much the worst in the league. 

Week 28 (Sept. 15 - Sept. 21)

Philadelphia versus Seattle (Open Cup Final)

Since there is no chance of a draw, the model is not designed to tackle this game. However, thanks to some research by Nate Silver, we can approximate how much of the draw probability should go to each team. The model likes Seattle better in a vacuum, but the teams aren't playing in a vacuum (they wouldn't all fit!). Below are the adjusted probabilities for this knockout-style matchup.

PHI: 60%

SEA: 40%

The regular season probabilities can be found below for week 28:

Home Away hwin awin tie spread
RSL COL 0.608 0.157 0.236 0.82
POR VAN 0.303 0.436 0.261 -0.11
PHI HOU 0.545 0.221 0.234 0.62
MTL SJ 0.374 0.269 0.357 0.23
NYRB SEA 0.452 0.272 0.275 0.31
CLB NE 0.409 0.292 0.300 0.29
CHI DCU 0.435 0.224 0.340 0.40
LA FCD 0.572 0.203 0.225 0.67
TOR CHV 0.520 0.169 0.311 0.66

Notes

*The biggest surprise is probably Vancouver being favored on the road. Portland has been worst in the league this season in even gamestates at home, and that affects the model the most. 

*Despite the fact that xGoals 2.0 isn't too keen on RSL overall, its home xGD in even gamestates is up there with Seattle's for best in the league. This helps to explain why it has the largest probability of winning of any team this week.

Week 27

Week 27's games showed up as posts here and here.