problem with spot plays

[AwolAtHTR]

Sun 25 Feb 7

in the Spot Contest thread: http://www.homebased2.com/forums/showthread.php?t=6216



The problem with spot plays
--------------------------------------------------------------------------------
I have been monitoring a selected group of the posted spot plays and yesterday they did quite well, 18% wins and a ROI that exceed 1.5.

The problem was that the two biggest wins were at Will Rodgers Downs and with out those two winners the group of spot plays was in the dumper.

Don't know where your wager, but WRD is not available to me. With spot plays even when you win you can still loose.


======================
good point Mike

I see that this emphasizes the difference between a Contest Spot play and a useable play for day to day betting. This is the same as bets to win a tournament, ie, yes, take a shot SOMETIMES but daily betting the price play is a formula for trying to be lucky and losing money.

My spot play research uses the FAVORITES as the list of tracks that I know I can bet. So, I start with ALL tracks and tinker with conditions. When spot conditions are close to final, I will confirm with the Favorites.

I have a rough idea of the percentages and play count that I will accept as final and useable. However, I feel that I am at the embryo stage of this process and do not have any '..written in stone..' goals for a spot play to be final.

I like the Mutual Fund idea that DanG described as a general structure for a collection of spot plays which have different production rates.

I am interested in hearing methods used by others, so that I might revise my search method.

duane

PS: the QUOTE did not work but you see that it is from the other thread.

[edw]

As I stated in another thread today, I'm just starting a database and, additionally, I have no education in statistics or anything else related to the problem so I'm among the least qualified to answer your question. I offer what I've started doing only to stimulate more conversation in the area.

My suspicion is that my early approach is more or less heading in the right direction but not the correct solution. I've programmed this as much as a reminder to myself that -when I get enough data - I've got to more specifically address the problem.

In addition to the normal win%, roi, etc. calculations I'm doing 2 other parallel calculations.
The first eliminates the highest mutuel, lowers the play count by 1 and lowers the win count by 1. This lowers the average mutuel, lowers the win percentage and lowers the roi.

For some spotplays, eliminating only the highest mutuel is probably insufficient and I do a second calculation that eliminates the top 5% of the plays and (I think) it also reduces the play count by 5.0%.

The theory is that if all 3 methods (or subsequent derivations) show profitability, there is likely to be less of flukey nature related to the play. I also think there is merit in employing the median (instead of the average) mutuel but I haven't had time to think that through yet.

My suspicion is that those HTR members with a good understanding of statistics can offer some good insights.

[Donnie]

To AWOL's point:
Yes you must be able to play ALL tracks at ALL times or these spot plays as being posted in the other threads are meaningless. I would (and am) taking those being posted and "modifying" and researching them at available tracks. If your spot hits 12% of the time, will you be playing on the day and at the tracks where it does hit? A spot play is not normally a "grinding" play. With a 12-18% hit rate you can probably expect runouts near the 20-30 range....do you have the mental fortitude for these types of runouts? I did notice in my db that the hits tended to land in clusters.....one of my spots went to 32 straight losers before a couple of $6 winners hit. You better have some FAITH if you are gonna keep after it.

EDW-- usually a good idea to throw out that highest mutuel. That's the problem with spot plays...one or two EXTRA LARGE winners makes it look good on paper. $122, $144 and $170 winners don't come along too often. Remove them and how does the ROI look? Myself...I am weak...I need to have some winners coming my way every 6 or 7 races or I buckle.....so probably 25-35 % win rate has to be my threshold.

[tomcat]

I think Mike has a good point. I have played the same tracks for years and there is plenty
of action. I only download these tracks and only play them in tournaments. I realize horses are horses no matter where they run, but all tracks are different and the run different for different horses (horses for courses).

Maybe if we do this again we could agree on a list tracks, maybe 10 or so, then try our spot plays. Anyway it's a lot of fun.

[km]

I think two things solve these problems - and thanks for pointing them out and offering suggestions everyone.

1) We will insist on purse $10,000+, this prevents minor tracks and bush fairs from being involved as no one could make a bet on those anyway - 'Dees example with WRD hit home on that.

2) MVM has the best idea for how to award the winning spot play =

Multiply ROI x Plays

Example;
Spot Play (A) = 1.50 ROI with 500 plays
Spot Play (B) = 1.10 ROI with 5000 plays

1.5 x 500 = 750 (250 profit)
1.1 x 5000 = 5500 (500 profit)

Spot play B would win the contest because it produces more overall revenue even though it has a lower ROI. Using this method also does not require a minimum number of plays. The backfit small sample volatile plays would lose to larger samples that have more stable results.

[AwolAtHTR]

I
2) MVM has the best idea for how to award the winning spot play =


I think MVM idea is great.

consider a variation to pay three by using Win, Win and Place, and WinPlcShw

duane

[DanG]

Terrific ideas. :)

As KM outlined staying within the win pool only.

[tomcat]

Perhaps the spot plays should be evaluated on their going forward score....maybe 30 days. Backfitting doesn't always lead to positive plays going forward.

[MikeDee]

I like the idea of a play it forward test as well.

[DanG]

In my opinion this can be tricky.

A profitable bomb seeking spot play can make money over 5 years and easily lose in 30 of the months.

We have all read of the player who has a profitable play only to trash it after a bad two weeks. To me that’s the worst thing you can do. If anything after a sustained losing run its time to start a modified progression and up your wagers.

So many spot plays are like riding the Montu at Busch Gardens. Rarely do they provide a steady dividend check.

[MVM]

I know that it is what is available in the robot, but I think the >10k purse level is just a bit too high for the contest.

There are a large number of tracks available on YouBet (among others) that have the majority of their races with purses under 10k. Many of these tracks are actually viable contest tracks, even in Vegas. If you eliminate races under 10k, you eliminate many (and in some cases most) of the races from the following tracks among others (EmD, Beu, Tdn, Hou, Tam, LA, TuP, PM, RP, LaD, Pen, Hoo, FL to name just a few) all of which are readily available to bet via one or more internet wagering platforms. There are some big, perhaps unexpected names in that list.

I would suggest that the most representative way of scoring would be to exclude races from the tracks (like WRD, GrP, WMF etc.) that are not realistically viable for most people to play.

[AwolAtHTR]

---A profitable bomb seeking spot play can make money over 5 years and easily lose in 30 of the months. ----


yes DanG, I see that you help make the point: Is contest goal luck (bombs) or income (chalk)?

with a luck goal, any '..going forward..' testing would be probably change the winner.

with an income goal, the '..going forward..' test would keep the winner ONLY IF the orginal was really solid.

oh, but the income spot play may not be all chalk, so that, may be the wrong label for the income spot plays. also, I agree (expected naysayers) that any spot you bet will have an goal of '..income..' by making money in the long run. So, maybe the amount of risk should be used as a contest label, as DanG refered to the Mutual Fund structure for managing spot plays.

Of course, the '..going forward..' rule will have a debate about '..HOW MANY DAYS..'. Well, I will repeat my earlier opinion, that thirty (30) days should be best, with one week being too short for evaluation and three months as too long.

duane

[MVM]

BTW, I noticed that WRD, the Arizona and Montana fairs, Manor Downs, and pretty much the entire list of tracks that we would likely choose to exclude from the testing do NOT show up in the Robot track list.

Don't know if this means that they are not part of the testing when one selects the "check all" button or not.

And yes, the files for these tracks are in my directory.

[DanG]

yes DanG, I see that you help make the point: Is contest with have a luck (bombs) goal or or income (chalk)?

Actually Duane that wasn’t the point I was trying to make.

I don’t consider bombs “luck”. They happen every year racing is conducted, just at a much reduced frequency.

If I reevaluated my long shot spots every 30 days I would only be hurting myself and possible wind up in a rubber room!

BTW: Now that I think about it I agree with MVM on the 10k threshold. One thing I use in evaluating the “level” of a track is minimum claiming price offered.

53 tracks in the last two years have run for a tag of $2k or less…It’s a pretty representative list of minor league racing.
COUNT TRACK MIN-TAG
1 DUN 1
2 SAF 1
3 FLG 1
4 SON 1
5 ELK 1
6 GCF 1
7 DG 1
8 RIL 1
9 MOF 1
10 MAN 2
11 MC 2
12 MD 2
13 MIL 2
14 LBG 2
15 LA 2
16 MPM 2
17 KSP 2
18 JRM 2
19 HPO 2
20 GRP 2
21 LNN 2
22 FAR 2
23 ANF 2
24 ATO 2
25 BCF 2
26 BOI 2
27 CLS 2
28 CPW 2
29 GIL 2
30 EUR 2
31 GPR 2
32 FER 2
33 FNO 2
34 FON 2
35 FTP 2
36 KAM 2
37 GF 2
38 PEN 2
39 DAY 2
40 YAV 2
41 RUP 2
42 SUD 2
43 WW 2
44 WDS 2
45 SDY 2
46 WYO 2
47 SND 2
48 WTS 2
49 POD 2
50 PM 2
51 YD 2
52 SR 2
53 YKT 2

[njcurveball]

I like your list Dan, I think that may be one of the few things we all can agree on, those are definitely MINOR tracks.

Your point about Tracks that run the whole year is also germaine.

I am hoping to see the magic $2 hit by the late closers. I have learned much more from this contest than any $99 Nunamaker or Sport Stat book.


thanks for having it Ken!

Jim

[MVM]

I would agree with Dan's list, with the following exceptions (these are >.5 million a day handle tracks and/or are available on one or more account betting interfaces). They also appear in the HTR robot track listing (with 1 or 2 possible exceptions).

Boi-LesBoisPark
Cls-Columbus (Nebraska)
Fer-Ferndale (CaFair)
Fno-Fresno (CaFair)
Fon-FonnerPark
LA-LosAlamitos
LnN-LincolnNational
Pen-PennNational
PM-PortlandMeadows
SR-SantaRosa(CaFair)
Wds-TheWoodlands
Yav-YavapaiDowns

The rest on Dan's list are very small handle tracks with no readily available betting access except for locals.

[MVM]

BTW...I am somewhat embarressed to admit that i have actually been to 31 of the "minor" tracks on Dan's list...so if I seem a bit provincial or sensitive.......:)

[njcurveball]

No need to be embarassed MVM! One of our favorite trips to the track use to be the 2 hour+ drive to Dover Downs.

9 or 10 races with an $800 purse and the feature was for $1,500. They still paid in American and I would put the drivers efforts at the time on par with any major track.

I think Forrest Skipper set the World Record there running for $2,500 and barred from the wagering.

I know at least one (Penn National) won't be on the Minor list for much longer with their new slots.

Jim

[km]

Good observations

The Robot track list are the ones that were available for simulcast betting from outside their own state. The tracks that required on site attendance to make a bet, are not on the list.

There is an "Unlisted" item at the top at the top of that Robot track listing. This bundles all those hard to bet tracks. If a track not on the list becomes available for simulcast wagering, let me know and i'll add it as I did for ATL and YAV during a previous update. Is there any way to bet BOI (Boise), Will Rogers (WRD) or GRP (Grants Pass) outside of the locale rignt now?

If you check ALL the tracks on the list, the Robot will test anything with results, even the "Unlisted" ones. Or they can be tested in one bundle by checking the "Unlisted" by itself.

Next contest maybe we'll use NO "Unlisted" tracks for the minor track exclusion and not use Purse value > 10k as the filter, since that would eliminate quite a few races from medium sized tracks.

[Mike T]

Yes, there is one offshore that offers these books. 5Dimes offers pretty much every track out there. They also offer a 10% bonus on all winning Win and Exotic bets at major tracks, which is better than Pinnacle's 8% on amount wagered if you are better than the track take. Below is a link to the tracks they offer.

http://www.5dimes.com/RacebookFrame.htm

[MVM]

Ken...

Boise is on YouBet....I haven't seen the other 2 on any of the 3 (now 2) account wagering services that I use.

[MVM]

Excluding the unlisted tracks would be the perfect solution IMO.

I'll revisit the robot list and post any tracks available on YouBet or XPressBet that are not included.

[MVM]

Well what do ya know....I log in to YouBet and sure enough, there is a listing of upcoming events...included in those is 3/3 Will Rogers - Opening Day (WRD). Looks like another one to add to the list.

I could be self-serving, and lie, saying that GrP (Grants Pass) will be on YouBet this year.

I've got a great, very simple spot play that works there. It gets a play in most races, but not all....the single criteria is listed below.

nLegs=4

:-)

[DanG]

I've got a great, very simple spot play that works there. It gets a play in most races, but not all....the single criteria is listed below.

nLegs=4
:-)

LOL :D :D :D

[MVM]

There are only 3 tracks available on Youbet, XPressBet or TVG that are not on the robot list:

Tim - Timonium
Boi - LesBoisPark
WRD - The aforementioned Will Rogers Downs