Initial Baseball Analysis 2009
The following was originally a post in our forum by Crdog who provides our baseball model.
We finally got our databases completed. The numbers below will be a tad better than whats tallied on the sportpunter site due to its grabbing of the best price available from multiple books instead of just the pinny line, it usually comes to about 1% better. Obviously its are worst season since we started in 03 by far, internally here its our first losing year. The big disappointment is the 5% and up overlays on teams -110 or better, thats been our bread and butter over the years and to come in at -4% turnover, well that hurt really bad. A lot of speculation as to why this happened, breaking it down by league, the NL was the problem, our projections for several teams this year turned out to be horrible. WAS, CHC and NYM really cost us huge.
Interleague did poorly again, although not as bad as usual, theres a couple of adjustments we make in these games involving the pitchers and DHs, however the data suggests that simply put the average AL player is better than the average NL player and a further adjustment will have to be made next season.
The biggest surprise is that the model won overall when including overlays that were faves. This has never happened before, its always lost when betting overlays that were -111 or worse. The fact that the model won overall when including the dismal performance of the dogs means the faves won at an unreal rate. The question is was this a one time type of thing or is there more to it…..
Also of note, we went back to our original pythag method of converting into win probs this season after last season use of a curve fitting conversion. And of course Murphy’s Law set in and the curve fitting method outperformed the pythag. After speaking to friends in the hedge fund arena theyve convinced me to go back w/the curve fitting conversion for next season.
Ive concluded that its type to consult some smarter people on these questions, so in January ill be hanging w/Jonno for a few days in Melbourne, i think he’ll be able to look at things from a fresh point of view and find some answers. In addition, the totals need some work, theyve always won, but never really won much in any particular year.
Here is our internal data, we use a static bankroll and a 1/15 kelly for our databases so dont really pay attention to the dollar figures, the turnover numbers are what matters.
Here are the numbers for the dogs using the pythag version we used this season (-110 or better at 5% and up overlays)
NL
Overlay W L Result $Bet Turnover
5% 108 137 (16,384) 155,108 -10.6%
10% 72 99 (17,931) 132,116 -13.6%
15% 42 65 (17,540) 99,781 -17.6%
20% 28 43 (11,730) 79,058 -14.8%
AL
Overlay W L Result $Bet Turnover
5% 120 127 2,264 210,990 1.1%
10% 83 92 551 184,403 0.3%
15% 57 72 (5,472) 157,075 -3.5%
20% 38 41 2,356 112,793 2.1%
Interleague
Overlay W L Result $Bet Turnover
5% 31 41 (2,297) 50,579 -4.5%
10% 24 30 (1,518) 45,979 -3.3%
15% 21 25 (384) 42,369 -0.9%
20% 12 12 2,303 28,011 8.2%
Total
Overlay W L Result $Bet Turnover
5% 259 305 (16,417) 416,677 -3.9%
10% 179 221 (18,899) 362,498 -5.2%
15% 120 162 (23,396) 299,225 -7.8%
20% 78 96 (7,071) 219,862 -3.2%
Here are the dogs using the curve fitting conversion:
NL
Overlay W L Result $Bet Turnover
5% 135 180 (11,170) 167,657 -6.7%
10% 66 105 (14,388) 120,765 -11.9%
15% 29 57 (11,874) 74,398 -16.0%
20% 13 34 (10,572) 45,542 -23.2%
AL
Overlay W L Result $Bet Turnover
5% 157 169 16,219 184,474 8.8%
10% 93 106 10,788 141,700 7.6%
15% 43 57 2,210 85,849 2.6%
20% 19 28 193 47,198 0.4%
Interleague
Overlay W L Result $Bet Turnover
5% 45 60 (4,791) 52,773 -9.1%
10% 21 37 (7,740) 38,528 -20.1%
15% 10 21 (7,484) 25,147 -29.8%
20% 2 10 (6,782) 10,258 -66.1%
Total
Overlay W L Result $Bet Turnover
5% 337 409 258 404,904 0.1%
10% 180 248 (11,341) 300,993 -3.8%
15% 82 135 (17,148) 185,395 -9.2%
20% 34 72 (17,161) 102,998 -16.7%
Model numbers for every overlay including faves using pythag:
NL
Overlay W L Result $Bet Turnover
5% 329 287 10,065 732,848 1.4%
10% 247 205 15,712 643,502 2.4%
15% 156 132 14,374 492,027 2.9%
20% 99 74 30,789 336,851 9.1%
AL
Overlay W L Result $Bet Turnover
5% 304 234 34,015 652,694 5.2%
10% 225 170 34,682 576,112 6.0%
15% 161 125 26,790 485,011 5.5%
20% 103 71 30,771 340,422 9.0%
Interleague
Overlay W L Result $Bet Turnover
5% 88 92 (28,707) 241,086 -11.9%
10% 64 73 (29,064) 217,557 -13.4%
15% 49 57 (24,190) 186,105 -13.0%
20% 34 33 (10,401) 141,869 -7.3%
Total
Overlay W L Result $Bet Turnover
5% 721 613 15,373 1,626,629 0.9%
10% 536 448 21,330 1,437,171 1.5%
15% 366 314 16,975 1,163,143 1.5%
20% 236 178 51,159 819,143 6.2%
Model numbers for every overlay including faves using curve fitting:
NL
Overlay W L Result $Bet Turnover
5% 202 219 (2,489) 253,748 -1.0%
10% 83 116 (11,144) 155,865 -7.1%
15% 34 59 (9,379) 86,156 -10.9%
20% 13 34 (10,572) 46,844 -22.6%
AL
Overlay W L Result $Bet Turnover
5% 207 200 23,210 242,716 9.6%
10% 109 118 12,959 171,501 7.6%
15% 50 60 6,013 99,775 6.0%
20% 22 28 3,884 51,932 7.5%
Interleague
Overlay W L Result $Bet Turnover
5% 66 83 (10,123) 89,206 -11.3%
10% 29 45 (10,595) 57,775 -18.3%
15% 10 23 (10,091) 29,997 -33.6%
20% 2 10 (6,782) 11,611 -58.4%
Total
Overlay W L Result $Bet Turnover
5% 475 502 10,598 585,670 1.8%
10% 221 279 (8,780) 385,141 -2.3%
15% 94 142 (13,457) 215,928 -6.2%
20% 37 72 (13,469) 110,387 -12.2%
Havent really poured thru the totals numbers yet, if i find anything interesting ill post later
