MLBPET 2011 Baseball Forecast
By Bradford Doolittle. Filed in Uncategorized |Tags: Baseball
For a number of years now, I've prepared for the Major League Baseball season by running projections for players and rolling them up into team forecasts. The system is a counterpart to and actually predates my basketball system, NBAPET, and as such bears a similar name: MLBPET. Since I've sought to de-emphasis sabermetrics in my baseball work, I didn't really have a place to publish MLBPET's forecasts this season, so I thought I'd throw them up here, just to get them on record. I don't really feel like re-writing the overview of my methodology that I come up with every season, but here's what I wrote last year, for a site for which I used a Royals angle:
"My process (see, I have a process, too) for projecting the baseball season is really pretty simple, though it requires an encyclopedia of data and a spreadsheet that currently takes 31 minutes to open on my laptop. First, I generate projections for several thousand players, dipping deep into the minor-league organizations for each team. The projections are based on recent performance, with the most recent seasons given the most weight. I also apply aging adjustments to player forecasts, based on historical patterns.
"Once I have all those raw projections in the hopper, I then build each team's anticipated depth chart and divide up the playing time. This is an inexact science, but in general, players with a history of being injured or platooned are going to receive the more pessimistic projections for playing time. Once the depth charts are built, the system adds up the runs created and allowed for each team. The data is fitted to a predicted run-scoring environment which is based on league data from the last five years. At the end of this step, I have an initial projection of each team's record, which always turns out to be really close the the official projection.
"At this point, I input each team's predicted runs scored and allowed totals into a simulation engine, in which I've also entered the 2010 schedule. Then the fun begins: I run the simulation 100,000 times to see what happens. What I'm left with are average win totals for each team's run of simulations, as well as their percentage chances for playing on in October. In 100,000 simulations, it would be very unlikely for a team to project to a ZERO percent chance of making the playoffs. So it is for the Royals. So I say again, the Royals can make the postseason in 2010."
There you go. Let's start with the baseline won-loss projections with runs scored and runs allowed. No commentary--the projections are what they are.
NL East W L Pct RS RA Philadelphia 89 73 .549 767 684 Atlanta 88 74 .543 785 706 Florida 84 78 .519 762 731 New York 83 79 .512 759 739 Washington 71 91 .438 708 800 NL Central W L Pct RS RA St. Louis 86 76 .531 799 745 Cincinnati 83 79 .512 778 753 Chicago 78 84 .481 756 780 Milwaukee 78 84 .481 708 735 Pittsburgh 72 90 .444 737 824 Houston 69 93 .426 667 784 NL West W L Pct RS RA San Francisco 89 73 .549 759 681 Los Angeles 85 77 .525 717 672 Colorado 84 78 .519 835 797 San Diego 78 84 .481 682 703 Arizona 72 90 .444 735 822 AL East W L Pct RS RA Boston 94 68 .580 896 750 New York 90 72 .556 879 774 Tampa Bay 85 77 .525 811 766 Baltimore 80 82 .494 831 838 Toronto 68 94 .420 684 812 AL Central W L Pct RS RA Minnesota 83 79 .512 823 791 Detroit 83 79 .512 803 776 Chicago 81 81 .500 798 797 Cleveland 74 88 .457 772 839 Kansas City 67 95 .414 732 875 AL West W L Pct RS RA Texas 88 74 .543 847 763 Oakland 83 79 .512 742 716 Los Angeles 79 83 .488 719 736 Seattle 70 92 .432 656 757
Here are the results of using those baseline projections and simulating the season 100,000 times. List are each team's adjusted average win total, an their percentage chances of winning a division, a wild card berth and just making the playoffs.
NL East aW Div WC PL Philadelphia 89.3 43.64% 14.26% 57.90% Atlanta 88.1 34.55% 14.88% 49.43% Florida 83.2 12.83% 9.01% 21.84% New York 81.6 8.71% 6.76% 15.47% Washington 70.2 0.27% 0.23% 0.50% NL Central aW Div WC PL St. Louis 88.0 50.04% 6.83% 56.88% Cincinnati 85.3 31.14% 7.52% 38.66% Chicago 79.7 9.42% 3.12% 12.54% Milwaukee 78.7 7.37% 2.48% 9.85% Pittsburgh 73.4 1.64% 0.52% 2.16% Houston 69.1 0.39% 0.10% 0.49% NL West aW Div WC PL San Francisco 89.2 48.51% 10.52% 59.03% Los Angeles 85.9 26.53% 11.08% 37.61% Colorado 84.5 19.64% 9.46% 29.11% San Diego 78.3 4.60% 2.81% 7.41% Arizona 72.1 0.72% 0.39% 1.11% AL East aW Div WC PL Boston 94.0 57.78% 21.59% 79.37% New York 90.2 30.23% 29.05% 59.28% Tampa Bay 84.8 9.45% 15.35% 24.80% Baltimore 79.6 2.52% 5.27% 7.78% Toronto 66.1 0.02% 0.04% 0.05% AL Central aW Div WC PL Minnesota 85.7 39.12% 5.07% 44.19% Detroit 85.2 35.91% 5.02% 40.93% Chicago 82.4 20.70% 3.95% 24.65% Cleveland 75.4 3.86% 0.82% 4.68% Kansas City 68.1 0.40% 0.06% 0.46% AL West aW Div WC PL Texas 89.8 67.48% 4.43% 71.91% Oakland 83.5 22.82% 6.51% 29.33% Los Angeles 79.2 9.11% 2.74% 11.85% Seattle 69.7 0.59% 0.13% 0.72%
Finally, here is a forecast for each team by unit: hitting, overall pitching, fielding and bullpen. Listed are the club's predicted league rank for each unit.
NL East HIT PIT FLD PEN Philadelphia 5 3 5 16 Atlanta 3 4 12 6 Florida 6 5 15 13 New York 7 9 6 11 Washington 14 12 14 7 NL Central HIT PIT FLD PEN St. Louis 2 10 8 14 Cincinnati 4 11 2 10 Chicago 9 15 11 4 Milwaukee 13 7 10 15 Pittsburgh 10 14 4 8 Houston 16 13 13 12 NL West HIT PIT FLD PEN San Francisco 8 1 3 3 Los Angeles 12 2 1 2 Colorado 1 8 16 5 San Diego 15 6 7 1 Arizona 11 16 9 9 AL East HIT PIT FLD PEN Boston 1 1 9 4 New York 2 8 5 1 Tampa Bay 6 9 3 9 Baltimore 4 14 8 14 Toronto 13 11 12 8 AL Central HIT PIT FLD PEN Minnesota 5 5 13 5 Detroit 7 4 10 10 Chicago 8 10 7 3 Cleveland 9 12 11 12 Kansas City 11 13 14 2 AL West HIT PIT FLD PEN Texas 3 7 2 6 Oakland 10 2 1 7 Los Angeles 12 3 6 13 Seattle 14 6 4 11
