Syzygy Analytics

Charlotte vs. New York Red Bulls

With Agyemang sold and Biel injured, I have NYRBs as a good bet not to lose here.  Charlotte FC is on a 6 game win streak but I think it’s sure to end soon and hopefully this is the game.

Seattle Sounders vs. Sporting Kansas City

Showing as a good home bet but I’m not too sure here – I think I’ve not got enough importance on the Rusnak absence.  Seattle also played mid-week so I’m leaving it.

Nashville SC vs. Orlando City

Orlando City are a favourite of my ratings right now and show as value here again.  I have accounted a bit for Orlando’s midweek game so I will be backing them again.

Real Salt Lake vs. Minnesota United

Salt Lake have new plays such as Rwan Cruz, Victor Olatunji and DeAndre Yedlin from FC Cincinatti.  They have also 3 suspensions including Diego Luna so it’s a tricky one to gauge.  I’ve already sided with Minnesota somewhat and there may still be value if RSL’s new lineup is not stronger than their average.

GOAL LINES

Just a quick bit on my total goals predictions

The orange and red on the right means I’m a bit low on my total goal counts across the board.  The MLS HAS dropped from 3.15 average goals per game last season to 2.93 which probably explains this however it surprises me how total goal expectations seemingly require more regression than individual team ratings.

My method for calculating total goals is possibly the culprit here I am not sure.  It goes something like this:

1. Calculate each teams total goals/xG per game removing the xG per game added by being better or worse than their opponents (a difference in ability creates extra (x)GPG.
2. Calculate the difference of this adjusted home team GPG and away team gpg to the league average gpg.  E.g. home team = 3gpg, away team = 2.5 gpg, league average = 2.7 gpg.  Home team = 0.3 gpg, away team = -0.2 gpg. 
3. Sum the home team and away team differences (league average +0.1 gpg)
4. Use a relationship between the difference in ability level of the 2 teams to gpg % added to create a final figure

I can work on a clearer explanation here although maybe my method isn’t deserving of one…

If I regressed my total expected goals numbers to what I would have expected last season they would match up much better with the betting markets.  I’m not sure why total goals seems to require more regression than team ratings.  More work required!