How promising is this data?

[Huguenot]

For all of 2006..from the Robot
Considering its highest price is $37 and the decent win rate the IV, does this have a better than average chance of succeeding going forward?

Based on four primary factors (honed by a couple more including odds of 2-1/20-1 but will work using ML of 2-1/15-1) including WinF and K1-2. Dirt/poly sprints and routes. Fast only. All Tracks, all purses, excluding Maidens and 3 yr olds. In fact it works better on the sub-10K purse races.


00824 00232 28% 47% 62% 1.06 0.97 0.92 $7.5 0007 $37 2.30

[njcurveball]

Looks promising, how does it look when you group by track or class?

Could be more profit slicing it that way.

Nice job!

Jim

[DanG]

For all of 2006..from the Robot
Considering its highest price is $37 and the decent win rate the IV, does this have a better than average chance of succeeding going forward?

Based on four primary factors (honed by a couple more including odds of 2-1/20-1 but will work using ML of 2-1/15-1) including WinF and K1-2. Dirt/poly sprints and routes. Fast only. All Tracks, all purses, excluding Maidens and 3 yr olds. In fact it works better on the sub-10K purse races.


00824 00232 28% 47% 62% 1.06 0.97 0.92 $7.5 0007 $37 2.30
Andy,

Looks terrific!

Couple things…

• Would look even more impressive if you have access to rebates. (See Ian / PTC?)
• With that hit rate you can bet progressively if you’re playing live, or a modified progression if you’re betting early.

For me personally, the only other move I make at this point is try and eliminate the bottom 5-10% tracks. If I’m unsure and have an historical 6% edge, I try and find an additional 5-10% buffer by taking out suspect tracks. I’m sure there out there, but I’ve yet to find a play that performs the same at every venue.

BTW: [Anyone who isn’t a little unsure going forward is selling snake oil IMHO] ;)

Best of luck!

[Huguenot]

Thanks,
I'm still fine tuning a little bit and haven't tested it on turf.

Yes, next step is to find the suspect tracks but I'm afraid the sample sizes might be too small.

Interestingly May was a horrible month but it's bounced back so far in June with a 3-13 and 45% profit.

I am really intrigued by the randomness of statistics and why May was so poor.
To the end I am downloading every selection in May seeking to examine if there was a rational explanation for the bad month or just part of the yin and yang of spot plays and streaks.

[DanG]

Thanks,
I'm still fine tuning a little bit and haven't tested it on turf.

Yes, next step is to find the suspect tracks but I'm afraid the sample sizes might be too small.

Interestingly May was a horrible month but it's bounced back so far in June with a 3-13 and 45% profit.

I am really intrigued by the randomness of statistics and why May was so poor.
To the end I am downloading every selection in May seeking to examine if there was a rational explanation for the bad month or just part of the yin and yang of spot plays and streaks.
For what it’s worth Andy, every play I have has peaks and valleys. I break all my wagers in quarterly reports. I used to do it by month and spot plays have taught me it’s too narrow a snapshot to draw conclusions from. (At least in my case.)

May – July are often transitional months for many barns.

Briefs examples…

• Unusual class drops due to making stall space for 2yo’s.
• 3yo can jump up overnight and some can start beating older.
• Many barns are using turf as an experiment / conditioning tool and making reading pacelines very challenging.

It’s just a period when the sport is making room for new stock and animals are maturing or in decline after over racing in the winter IMO.

[njcurveball]

I am really intrigued by the randomness of statistics and why May was so poor.



Have you tried removing the 2 year olds?

[Mark]

Andy, FWIW I've had a frustrating time finding a similar low-odds 1000+ annual play. With 27 months of data the closest I came was a K1-K3 sample, 3/1+ M/L centered on CL, FR1 and HTR. The results below, from 40 tracks selected by availability and some comfort level, gave me pause.

I suspect the volatility inherent in productive ratings year-to-year works against a low-odds approach, but I hope otherwise.

March '05 through March '06 26% wins .96 ROI 2.17 IV

March '06 through March '07 30% wins 1.19 2.41

3/18/07 through 6/04/07 26% wins .85 ROI 1.96 IV

I've yet to attempt culling specific tracks which seems tricky. For the time being I'm going after higher odds and dry spells that appear to resolve themselves annually.

My apologies for the poor format.

[Huguenot]

I have a theory that you can be successful with these plays if you have the time to engage in a little qualitative analysis to go along with the raw data. WHY do some picks lose? Can you find a few common threads that will eliminate more losers than winners -- e.g. "F" or "E" horses who are too slow early, horses who ran a huge in its last, or a horse who ran a freakish number to qualify and will likely not repeat it. Or horses who got claimed away from a Dutrow type.
You probably wouldn't have to researc EVERY race in the sample to find out. If you look at 50 races I would bet you would start to find some patterns and then do a quick check to see if eliminating those types of plays would do more harm than good.

The key "if" here is the time.