Playable races

[George]

A quick comment about the query recently discussed in Selections section using best cramer# in 90 days with rank 2 or 3.

Appears to be very streaky long term and not a good "mechanical" play. The only
automatic plays I make are high fr1 longshots where if I look hard at them will be
afraid to bet them.

Think the c90 query is an excellent "race finder" to take a harder look at the race.
Many days I don't have time to slog thru 150 races looking for a few "good"
plays. Have found these races to be "cullable" and able to find one to six
nice races to play in less than 20 minutes.

Average plays spit out a day is 12. Using all the htr tools can find several strong
plays quickly. Many of these races are 2-3 horse match races which greatly
simplify the exacta requirement. Most can be hit with a $4 key and average
around $38 payoff.

Reccommend anybody looking for a quick way to home in on some HTR
profits to take a look at using this as a race identifier method.

[fred4now]

Appreciate your comments George, will take a look.

[MtKen]

Geo:

I was going to start a new thread but this looks like a good spot to post this.

This is from 2 months of downloads (Nov & Dec '01).

Trying to find playable races where the favorite or another heavily bet (under 2/1) horse tanks .

I looked at 2 types of horses, all those <2/1 actual odds & those with ML Rank #1 & ML is 2/1 or less.

C90 rank was my only test factor.

1st, those with actual odds under 2/1.....

1....2369/1017.....43%.....lost $-8.8%
2....1177/403.......34.2%.........-$21.2%
3....622/217.........34.9%.........-$17.6%
>=4...805/212......26%............-$37%
0.....295/89..........30%.............-$31.6%

Good luck if you're betting against C90 1 favorites.....you have to beat the track take + another 10-12 %.

The vulnerable favorites are those ranked 4 or higher which actually give you an advantage vs. the take. They show up in 15% of the races. If you play against the 0's too that's 1 race out of 5 where you're betting into a positive situation. (Actually I forgot about races where the favorite didn't get counted here so these situations show up less frequently--I'll figure it out later but its probably every 7 or 8 races.)

The 2nd method is to use the m/l, rank 1, m/l odds <= 2/1.

The disadvantage to this method is that it comes up with 1/2 the # of plays.

1's & 2's are basically the same, although the 2's did a little better by losing only $16.5%.

3.........329/94......28.6%......-$32%
>=4....388/79.......20%.........-$43%
0.........134/40.......29.8%......-$21.3%

I don't know why the 3's lost 6% more races & 15% more $ here, probably need a bigger sample. 4+'s lost big time again as expected.

0's did a little better here but there was a $17 winner in this small sample.

I didn't break 0's into FTS or layoff horses but that might be interesting.

Also I have to wonder if those turf/dirt figs distort everything somewhat.

As usual, it would be great to see what would happen with a larger sample.

[George]

If am following you right agree it is tough to go against c90 top rank favs.

What the query I am using does is find high quality races for older more experienced horses. Most are STR and HCP or high dollar claiming races with only one to three serious contenders entered. Interesting how many are scratched at last minute by trainer who obviously did not expect one of the competitors to be in the race. Good horse to put on "watch" list for next race.

Have also noticed that in these races the c90 rank 1 has a distinct tendancy to "bounce" leaving it open for the c90 rank 2 or 3 to score. Think that is a little different situation than you describe but not sure.

Another strange thing I have observed in the claiming races found by this query is that often one of the top 3 cramer number horses is rising abruptly in purse. An example would be a horse saw recently named Linkoman. Had ran several 9k purses at HAW for another trainer and suddenly is shipped (loaned?) to a new trainer at FG. Promptly enters him in 3 straight 28k purses at FG and wins 2 of them. Then back to the 9k's at HAW.

Sorry to rattle on but find these to be interesting races and think most are safe to try and beat the top cramer number if number 2 or 3 has something unusual about them.

[MtKen]

George:

Most of these C90s favorites do lose but I think they are tough to beat & you need a pretty good reason to bet against them. The situations where the weak favorites pop up may be a case of "somebody has to win" so find somebody to bet. My point is that if these weak horses under 2/1 are taking up 1/3 to more of the win pools & lose 3/4's of the time you might be able to bumble your way into a decent ROI.

On the other hand, there are 1 of these C90 #1 favorites in 1 of every 2 or 3 races---might have to try to weed out the losers & get the win % up into the K1 range.

[tomcat]

How about un-playable or high risk races?
Scratch the top 2 or 3 ML, the use C90 or just run the software...crazy races deserve crazy handicapping.
Hey MtKen, scratch the #1+2 rML, then look at C90 or GL8. Let us know.

[hurrikane]

couple of observations. Can't test any until the new export comes out with the actual c90 #s instead of the ranking.

First....a c90 rank 1 with only a point or 2 difference from the c90 r2 is rather meaningless. 10 points seem to be the cutoff and you can usually reach that in the first 3 rankings.
I like taking these horses and finding the money.

Second...is the c90 r1 near of better than the Cramer par for the race. If it is way below as you will find with MDN 2 and 3 yo..the numbers seem somewhat meaningless.

As I said.. no way to test it right now.

[MtKen]

Tomcat:

This was just a quick (2month db) observation on those lo-odds horses that aren't in the top 3 C90 & they are very beatable--apparently. I haven't looked at all at any more contentious or crazy races, figuring that my best bets right now are against those vulnerable C90 4+ or 0 favorites.

Those C90 #1's that go off at 2/1 lose even less $ in my 2 month db---$7.6 % compared to -8.8%. Again, I think it would be tough to make $ betting against them. At 5/2 they do a little worse; losing $13.6%......still less than the track take, which gives you an extra 'tax' if you bet against them.

'Kane:

The overlays I like are C90 top 5 who have 1 of the best races in their last couple or F1 1st or 2nd rank who have a competive fig anywhere (within 200 days) in their pps. Then, like Tomcat suggests, throw out the vulnerable favorites & bet these overlays or whatever other spot plays fit. I'm just looking to cherry pick the races with the vulnerable favorites.

[hurrikane]

Sounds good to me Ken. Like your logic.

These days I play the ole' F1 spot plays and after that it's just p4s for me. So I'm stuck having to play 4 races off a card. I usually look for cards with false fav, strong k1**** to get a single...or, like friday, had 2 spotplays, off a query that hits 40%, in the p4 at SA. ended up with 3 singles on the ticket. I had 5 live tickets going into the last race and got nosed by a pig.

Cheapest p4 ticket I've ever played and felt like I had it in my pocket. Only thing the 1 horse had was a d/s cramer rank of 1....rank of 1 and the only horse that ran near the cramer par for the race. How the hell I missed him I don't know.

Still crying in by beer over that..paid 2300.00. :(

[George]

Ouch!! So near...and yet so far....those hurt.
Get'em next time Hurrikane.

[MtKen]

'Kane:

Ouch is right...hopefully you'll hit a good one sooner than later. I'm amazed at some of those Pik 4 payoffs. E-horse caps exotic payoffs; need another account for exotics.

Did you play the Pik 3 with your singles? I'm thinking I should be doing more doubles & Pik 3s using these for singles.

I'm also looking for ways to get the win % up to K1 level on the favorites. 1 variable I like is when the C90 #1 has the best fig in the last race vs. everybody else's last 2 races/figs but its not the C90 1's best fig or the fig that made it C90 # 1. This (hopefully) protects against a 'bounce' & gives you a horse who doesn't need his very best effort to win. I just started looking at these & I think they'll hit above 50%.

Good luck on those Pik 4s!

[George]

Ken...Interesting approach on finding c90 r1 with an edge. Do you have an automated way of finding them with Excel or just eyeballing them ?

Until we get the ability to d/l several races per horse, no way to automate in access.
Have become really reluctant to run these things by hand one at a time.

[MtKen]

George:

Mainly eyeballing them, waiting to see if anything else is going to be put into the 2001 export before I go & export more past files. Anyway, never found anything that held up at lo-odds in the db. Maybe there are too many 'defects' that go unnoticed by the data limitations for lo-priced horses. I think i will export a couple more months of data & do some twiddling around, though.

In my mind , the perfect favorite has the same solid trn/jock combo it has run well with recently, isn't changing in distance or surface, has a short layoff, matches up well vs. today's pace, has a fig edge from its last race & better figs besides, isn't a sucker horse, & might as well include a nice shiny coat too and a big fat 8/5 on the tote!! Might see that every other month or so:).

Trainer switches I hate, talking favorites here, along with jockey changes for no apparent reason. if a horse is stretching out I like to look at the HTR "Classic" pp screen to check out the pace figs-- I like a horse that doesn't show deacceleration ( 98-92) at 6f if its going 7 or 8f next. Anyway lots to look at, always avoid those favorites that throw a clunker but it seems many of them come back in their next try. No way to db test C90 1 horses that still get bet that have run 2 poor races but maybe these can be beat.

Appreciate your ideas from playing K1 favorites as to what worked or didn't. I'm using E-horse & getting a rebate of up to 5%---if you could break even place betting these horses & find enough of them you could make decent money off the rebate. Churn,churn,churn.....

[George]

Ken.....real problem is too many players that can identify the "ideal" horse in a race. Did an analysis not long ago where I extracted by exact distance all the losers based on top ranked horse for each of six pace factors. Then extracted all the winners. Ran both sets through a statistical analysis prog.

The loser set for each factor had median odds of 9-1 at post. The winner set had median odds of 3-1 at post. Combining sets the factors had hit rate of 29 to 30 percent and each lost 17 to 18 cents on every dollar wagered.

In other words, our competition is very very good at identifying "live" versus "false" favorites. Tough to beat!! The k1**** played carefully can eke out a small positive return in both win and place but don't think it is worth the stress.

Based on above think the tote-board is as good a false fav identifier as any. Unfortunately all the late simul money makes even that hard to id until it is too late to wager.

If I ever find a foolproof "false fav" technique will share it.

[hurrikane]

good banter my cold weather friends. :) (I say that with a 10 degree windchill outside! *$^#! that 's probably a heat wave for you two!)

I have, as you, never found a good blind fav picker or false fav picker. What I have found is, as you have eluded too, a query that pumps out races to look at closer and make a decision.

On the other hand...check the SA, AQU, and LRL models..thier are some very strong factors in their that will lead to some very stong, ie 40% hits rate and 35% return, races to look at every day. Only 2 or 3 but good mutuals. To catch 2 at the same track in the P4 get me..well..let say..excited. :D

Hmm..perhaps I am stating the obvious.

Ken, no, didn't play the pick 3..should have..foolish on my part. Will have to start streching the envelope again.

Have changed venues and am no longer on the 15hr a day Bin **&^% Laden chase. Hopefully I can get my life back again. :D

[MtKen]

Geo & Kane:

Spent awhile dinking around with C90 1 horses <= 5/2 odds....

Came up with this:

Dis: 5-7f

Sur: <> turf

Field: <=8

MLO: <= 5/2

MLO rank= 1

Actual Odds: < 3/1

AP (2001) : <=4

-bet$1350

Won $1322

Place $1306

Didn't write down the exact W/P %'s---around 50% winners (675 horses) & 65-70% place.


So I think "Great", about breaking even on a decent sample size here. So I get lost in the db for a couple of hours tweaking this sample until I have decent numbers & still have a decent sample left.

Then I go into the only 'control' sample I have exported 2001 into & check the 1st query......after almost breaking even on 675 horses in the large sample it tanks in the 2nd--losing 21% in a 105 horse sample. Moral: Don't start tweaking a small query before you check it out on all your samples.

Amazingly the bigger queries i came up still nearly broke even in the little sample--if I discount 1 large flop with Qt3 20+ races (F1 prices go up in those, wonder if favorites have a tough time generally in those?) I've been avoiding back-exporting stuff in 2001 past last Nov but I might do another month or two.