Ray2000
01-26-2013, 08:43 AM
A few years ago I sent a team of 50 monkeys around N America betting Win and Straight Exactas at various Harness tracks. They have returned and their Average %ROIs are shown below. There's no handicapping done here, just random selections for win and exacta.
Track Races Win%ROI Ex%ROI
MOH 2354 -22% -30%
TGDN 1367 -23% -27%
PRC 1800 -23% -32%
INDY 1680 -24% -38%
VD 2004 -24% -23%
FRD 1098 -24% -34%
CALX 2771 -25% -35%
MEA 7436 -25% -32%
WR 1262 -25% -32%
PCD 4718 -26% -31%
WDB 3558 -26% -37%
DD 5286 -27% -38%
BR 3623 -28% -33%
HAR 3812 -28% -36%
MAY 3425 -28% -34%
HP 1905 -28% -38%
HOP 1720 -28% -40%
BTVA 1271 -29% -34%
YR 8857 -29% -37%
RIDC 2466 -29% -28%
BMLP 2976 -29% -40%
NOR 1749 -30% -44%
PPK 2982 -30% -38%
STGA 5256 -30% -34%
Big M 2767 -30% -40%
SCD 1554 -32% -29%
PHL 5342 -32% -41%
MR 6908 -32% -42%
FLMD 1976 -34% -35%
NFLD 7109 -35% -45%
GEOD 1520 -36% -32%
FHLD 3547 -36% -42%
All 113710 races x50= 5,685,500 Bets
Win Bets -29% 3.23 StDev
Exacta Bets -36% 10.00 StDev
The table is sorted on Win ROI.
The monkeys had their best Exactas at Vernon and Tioga, strange.
Having a little fun here, but the purpose was to establish a base ROI and STDev for doing
some statistical hypothesis testing.
If you make the huge assumption that the right tail of the wager hits is normally distributed, then simple Z testing a sample run (using the values above for the population mean and Standard Deviation), can tell you how many "Lucky Monkeys" would beat your test run.
Example: My other thread on key box exactas has some odd win bet results right now, running +9.7% ROI on 231 bets. BTW I corrected the number there, on second look I knew that 20% couldn't be right.
Zscore=difference in means * sqrt(#bets) / population StDev
=(-.29 - +0.10)*sqrt(231) / 3.23 = 1.81
1.81 corresponds to a 1-normdist(1.8) or
.....4 Lucky Monkeys out of a hundred.....could do as well.
Track Races Win%ROI Ex%ROI
MOH 2354 -22% -30%
TGDN 1367 -23% -27%
PRC 1800 -23% -32%
INDY 1680 -24% -38%
VD 2004 -24% -23%
FRD 1098 -24% -34%
CALX 2771 -25% -35%
MEA 7436 -25% -32%
WR 1262 -25% -32%
PCD 4718 -26% -31%
WDB 3558 -26% -37%
DD 5286 -27% -38%
BR 3623 -28% -33%
HAR 3812 -28% -36%
MAY 3425 -28% -34%
HP 1905 -28% -38%
HOP 1720 -28% -40%
BTVA 1271 -29% -34%
YR 8857 -29% -37%
RIDC 2466 -29% -28%
BMLP 2976 -29% -40%
NOR 1749 -30% -44%
PPK 2982 -30% -38%
STGA 5256 -30% -34%
Big M 2767 -30% -40%
SCD 1554 -32% -29%
PHL 5342 -32% -41%
MR 6908 -32% -42%
FLMD 1976 -34% -35%
NFLD 7109 -35% -45%
GEOD 1520 -36% -32%
FHLD 3547 -36% -42%
All 113710 races x50= 5,685,500 Bets
Win Bets -29% 3.23 StDev
Exacta Bets -36% 10.00 StDev
The table is sorted on Win ROI.
The monkeys had their best Exactas at Vernon and Tioga, strange.
Having a little fun here, but the purpose was to establish a base ROI and STDev for doing
some statistical hypothesis testing.
If you make the huge assumption that the right tail of the wager hits is normally distributed, then simple Z testing a sample run (using the values above for the population mean and Standard Deviation), can tell you how many "Lucky Monkeys" would beat your test run.
Example: My other thread on key box exactas has some odd win bet results right now, running +9.7% ROI on 231 bets. BTW I corrected the number there, on second look I knew that 20% couldn't be right.
Zscore=difference in means * sqrt(#bets) / population StDev
=(-.29 - +0.10)*sqrt(231) / 3.23 = 1.81
1.81 corresponds to a 1-normdist(1.8) or
.....4 Lucky Monkeys out of a hundred.....could do as well.