If the World Series has the Yankees playing the Braves,
the Yankees have a 59% chance of winning.

If the World Series has the Yankees playing the Padres,
the Yankees have a 74% chance of winning.

Yanks vs. Padres (Yanks win)
Yankees Win in 4,5,6,7 games 0.136473 0.202754 0.223496 0.174595
Yankees Win: 0.737317
Padres Win in 4,5,6,7 games  0.023017 0.062034 0.081386 0.096246
Padres Win: 0.262683

Yanks vs Atlanta (Yanks win)
Yankees Win in 4,5,6,7 games 0.080509 0.140451 0.188428 0.177615
Yankees Win: 0.587003
Braves Win in 4,5,6,7 games  0.046392 0.109147 0.125757 0.131702
Braves Wins: 0.412997



Below, we present a little analysis of this year's
predicted standings, comparing our predictions with those
of the experts.


Now that the regular baseball season has ended, I have been able to
do a comparison of our predictions of March 28, 1998 with those of
Sports Illustrated, Baseball Illustrated and the Star-Ledger. In all
cases and by all measures, our predictions outdid theirs (although in some
cases, all were pretty lousy).

Here are the details:

Who will make the playoffs: 
Of the 8 teams to make the playoffs,
Sports Illustrated picked 3, Baseball Illustrated picked 3,
the Star-Ledger picked 4, and we
got 4-1/2 (since I said the AL wild-card was a toss-up between Boston and
Baltimore and Boston is it - so we get half a point).

Who most accurately picked the order of how the teams would finish 
in each division?
On average, Sports Illustrated and Baseball Illustrated were off by 
1.13 places for each team, while the Star Ledger did slightly
better at 1.07 places off, while our predictions were off
by only 0.77 places.
(This is claculated as follows: Take the NL East
  Correct: Star-Ledger
1 Atlanta  Atlanta
2 Mets     Mets
3 Phils    Expos
4 Expos    Marlins
5 Marlins  Phils
Here Star-Ledger got Atlanta and Mets in correct spot. Phils were
off by 2 (3rd instead of 5th) and Expos off by 1 (4th instead of 3rd)
and Marlins off by 1 (5th instead of 4th) for a total error of 4
in this division. Totals for Sports Illustrated and Baseball Illustrated
were 34, for Star-Ledger 32 and for our Markov Chain approach 23.
Random placement of teams would lead to expected error of 45-2/3
or 1.52 per team.

The Star-Ledger also gave predicted number of wins for each team.
Our prediction was off in number of wins by an average of 6-2/3 per team, Star-Ledger
was off by 8-1/2 wins per team.

Finally, if you rank teams in expected order 1-30 and compute error
as about (e.g. if team had 5th best record but we predicted they would have the
10th best record, this would add 5 to the error), We were off on average
by 4.18 places per team and Star-Ledger was off by 5.73 places per team.
(I think Sports Illustrated might have made a prediction on this but I
no longer have my copy). Random placement of teams would lead to 
expected error of 299-2/3 or 9.99 places per team.

So in all cases, our Markov Chain approach beat the experts!!!!!!