Eric Bowers narrowing the field
- Thread starter Olwenba
- Start date
retriever
Stallion
Batman For Ever - non-runner.
Forecast prices were a bit topsy-turvy in that race. I don't like to put to much faith in W/Hill's forecasts, that HRB uses. Way off the mark a lot of the time. Sometimes I think they do it on purpose to screw you over.
Even if the forecast price was half that it still wouldn't have made the cut with an OR of 74. You have to look at it that Dance Thief gave you a great run for your money at 11/1. Nearly a different story....
.
Last edited:
Just going through some of the numbers I haven't gone to too much detail so far, not including today's races I thought I'd share for anyone interested. I removed any horses on the shortlist which became non runners after posting. I've got a spreadsheet of all the results since the start of the thread and a separate spreadsheet for the shortlist selections. The method looks to eliminate 50% of runners in each race, my shortlist contains 49% so consistent with that. Still operating around 70% I want to get this to nearer 80% in time (hopefully) to do that I may have to use forecasts odds from more than one source as suggested before in the thread. I've taken a quick look at the spread of winners and the price of winners has been lowest even money right up to 20/1 at SP. However, of the 47 wins, only 5 were double digit odds at SP. So a focus more in the front end of the market it seems?
RACES 66
WINS 47 71.21%
RUNNERS 774
SHORTLIST 382 49.35%
RACES 66
WINS 47 71.21%
RUNNERS 774
SHORTLIST 382 49.35%
retriever
Stallion
Well, yes and no.
Yes, if you want to increase the current 71% upwards and have a better strike rate. You wouldn't necessarily be more profitable fishing in that selection pool. The method is already geared towards the front end of the market anyway and I don't see how you can change that.
Random selection dictates that you will have a strike rate of about 10%. You have 47 winners captured from 382 selections which is 47/382= 12.3%
You are on the right side of the fence here.
Can you tell me the total sum of the winners SP, and if you have it, at BFSP?
.
SP 301
BSP 365.43
5 AW races to have a look at....
| time | track | horse | runners | Bower Rating |
17:00:00 | Wolverhampton | Written Broadcast (IRE) | 11 | 5.5 |
17:00:00 | Wolverhampton | Kenstone (FR) | 11 | 5.75 |
17:00:00 | Wolverhampton | Last Date | 11 | 9 |
17:00:00 | Wolverhampton | Angel Amadea | 11 | 9.5 |
17:00:00 | Wolverhampton | Outreach | 11 | 11.5 |
17:00:00 | Wolverhampton | Midnight Flame (IRE) | 11 | 13.5 |
| time | track | horse | runners | Bower Rating |
18:30:00 | Wolverhampton | Peripeteia | 11 | 5.5 |
18:30:00 | Wolverhampton | Dame Sarra | 11 | 5.75 |
18:30:00 | Wolverhampton | Wilkie | 11 | 9 |
18:30:00 | Wolverhampton | Pysanka (FR) | 11 | 11 |
18:30:00 | Wolverhampton | Scylla | 11 | 12 |
18:30:00 | Wolverhampton | Virtual Hug (IRE) | 11 | 14 |
| time | track | horse | runners | Bower Rating |
19:30:00 | Wolverhampton | Brazen Idol | 11 | 3 |
19:30:00 | Wolverhampton | Bright (IRE) | 11 | 6 |
19:30:00 | Wolverhampton | Tyger Bay | 11 | 7.5 |
19:30:00 | Wolverhampton | Ultramarine (IRE) | 11 | 13 |
19:30:00 | Wolverhampton | Level Up (IRE) | 11 | 16 |
| time | track | horse | runners | Bower Rating |
20:00:00 | Wolverhampton | God Of Thunder (IRE) | 11 | 1.75 |
20:00:00 | Wolverhampton | Waterloo Sunset | 11 | 11.5 |
20:00:00 | Wolverhampton | Le Rouge Chinois (DEN) | 11 | 15 |
20:00:00 | Wolverhampton | My Harrison George (IRE) | 11 | 15.5 |
20:00:00 | Wolverhampton | Miss Dynamic | 11 | 22 |
| time | track | horse | runners | Bower Rating |
20:30:00 | Wolverhampton | Goldsmith (IRE) | 10 | 3 |
20:30:00 | Wolverhampton | Blue Yonder (FR) | 10 | 4.5 |
20:30:00 | Wolverhampton | Ellexis (IRE) | 10 | 10 |
20:30:00 | Wolverhampton | Prince Ali | 10 | 10 |
ARAZI91
Mare
retriever - start using a site called Betmover where the Betfair feed is free.I'll put up a snapshot of the 5.00 at Wolverhampton , gives you updated Betfair odds in increments from 12 hours to 1 minute to the off , you can pick 9 options - also gives the % of the price move, just need to sign up with an email / password. They also use WAP(Weighted Average Price) which is slightly more accurate than standard Betfair pricing as it only accounts for bets that have been matched to a niminum liability of £100. They also have the subscription based datafeed to TPD's sectional times for in-running but its quite costly but the Betfair pricing and feed is free. Good little site.
I'll download a snapshot of the first just before the off at Wolverhampton and post up the link to the site.
I have a script in python that updates the Betfair Odds every 30 seconds, also gives you volume traded and have included WAP as well, without out i could not update my probabilities in a Bayesian sense.
Remember if anybody is using Bookmakers Odds to normalise them to 1.00 so they become "fair odds" and extract the overround - there are a few methods of doing this but unless your backing 75-1 shots + , the difference between each method is minimal. Most use the "Shin" method but testing the four methods the differences were minimal - i would just use standard normalisation for simplicity.
You can see that within Industry SP they Starting Prices returned (which are really Closing Prices) there is a 0.17184 over-round to beat or in % terms 17.184% - that's a big hurdle to overcome - simple Normalization extracts the over-round and gives a more "truer odds", although there is no such thing as "true odds" within horse racing, we can only estimate until after the race where the whole probability mass is assigned to the winner.
LukeyBoy
Yearling
I use Margin relative to odds just for pure simplicity.
I'll put up a snapshot of the 5.00 at Wolverhampton , gives you updated Betfair odds in increments from 12 hours to 1 minute to the off , you can pick 9 options - also gives the % of the price move, just need to sign up with an email / password. They also use WAP(Weighted Average Price) which is slightly more accurate than standard Betfair pricing as it only accounts for bets that have been matched to a niminum liability of £100. They also have the subscription based datafeed to TPD's sectional times for in-running but its quite costly but the Betfair pricing and feed is free. Good little site.
I'll download a snapshot of the first just before the off at Wolverhampton and post up the link to the site.
I have a script in python that updates the Betfair Odds every 30 seconds, also gives you volume traded and have included WAP as well, without out i could not update my probabilities in a Bayesian sense.
Remember if anybody is using Bookmakers Odds to normalise them to 1.00 so they become "fair odds" and extract the overround - there are a few methods of doing this but unless your backing 75-1 shots + , the difference between each method is minimal. Most use the "Shin" method but testing the four methods the differences were minimal - i would just use standard normalisation for simplicity.
View attachment 140666
You can see that within Industry SP they Starting Prices returned (which are really Closing Prices) there is a 0.17184 over-round to beat or in % terms 17.184% - that's a big hurdle to overcome - simple Normalization extracts the over-round and gives a more "truer odds", although there is no such thing as "true odds" within horse racing, we can only estimate until after the race where the whole probability mass is assigned to the winner.
Did have an idea for this thread and use the FC odds as bookmakers odds and then adjust to 100% book using the Margin relative to odds method.
Not an advert but horse-race-data is £50 for the year and has downloadable excel results have prices I think from 15 mins to post time which is handy.
Last edited:
ARAZI91
Mare
retriever - this is the first at Wolverhampton - you can connect directly to Betfair and receive all the increments from 24 hours to 1min to post i'm sure - they also show charts with volume traded using 5 increments - the % move is cumulative based on the largest price available in the last 24 hours , not increment to increment
These are the charts which can be made larger by clicking on a horse
Heres the 6.00 just a few minutes ago
Using WAP (weighted average price) to a fixed liability is around 2% more accurate than last matched or BFSP - it may not seem a lot but cumulatively over 1000's of races that little edge is compounded
retriever
Stallion
Sorry
ARAZI91 I got 'lost' within the site. Thanks for this, it will be a great help!
I've been a subscriber to Bet Angel in the past ( which uses TPD and the like ) but I'm not totally keen on the trading aspect, preferring to access and bet my own horse racing selections....backing and laying. This will give me added useful info!
.
ARAZI91 I got 'lost' within the site. Thanks for this, it will be a great help!I've been a subscriber to Bet Angel in the past ( which uses TPD and the like ) but I'm not totally keen on the trading aspect, preferring to access and bet my own horse racing selections....backing and laying. This will give me added useful info!
.
formtheory
Stallion
ARAZI91
Mare
No not really , there is a lot more of what goes in to a value bet than just price. When i said that WAP pricing is around 2% more accurate than Betfair i was taking a "Big Data" approach over thousands of races although Tyger Bay had plenty going for it tonight on my models , subjectively thought Brazen Idol would struggle off that mark tonight for a horse that started last season off a mark of 46 and was a very weak race he won at FFos Las Last time out, Ultramarine has never really fired at this C/D and i could not see much else in the race TBH , in the last 30 mins Tyger Bay went from 10.5 to 6.4 at the off whilst there was no other notable moves. Tyger Bay certainly looked well handicapped tonight based on some of his turf runs at the start of the season where he was rated as high as 84 but his record shows he's more effective on the AW and had come into a form cycle. One thing that helps to spot "value bets" is your own probability distribution line , oddsline , tissue - call it what you want - then you have something to compare the live market with, but that's a whole different ball game.Hi
I’ve created an account how do you suggest using WAP prices ? Sorry if being daft.
My selection tonight Tyger Bay has a WAP price currently of 4.95. The Betfair current price is 5.2. So would you say that is currently a value bet ?
Thanks
Tonight's OR's v Prev. Max OR - Bright making his Hcap Debut.
BTW
formtheory i have done the analysis on the <=10 Hcap's or Less , done it in Python last night and exported it into Excel , just need to do a bit of formatting and i'll post it up tomorrow.
Last edited:
mick
Sire
I liked the above comment because while the market has proven long term accuracy it is based on opinions and forming your own can prove to be time well spent. The way i do so is to first find a horse of interest Re its winning chance and i then create a MAP Minimum Acceptable Price. This is arrived at by pondering the negatives to its chance ( and there are always some ) plus of course perceived strengths in any of the opposition.
As
ARAZI91 rightly says a different ball game which took me years of trying before i reached the point whereby i felt confident enough to apply the thinking but on balance it now works well enough. While the main use is to back winners at better than chance odds its application also enables me to swerve some perceived poor value losers which is close to same difference Re our long term profit lines. ?There were 9 races that I had initially started to look at today, anyway, had a bit of time to look the the F/C odds and compare. I checked 5 lots of odds HRB/Oddschecker/Sportlinglife all had the same F/C prices. RP and Timeform each had their own prices. If I've done the calculations right then Timeform were some way ahead when it comes to having a 100% F/C book, highest book was 106% in the 3.05 and 8.00 races. Whether that's just coincidence or not but I think I will sample their forecasts going forward on the method. Left with 5 races to look further at;
| time | track | horse | runners | Bower Rating |
14:05:00 | Lingfield | Light Speed (IRE) | 9 | 4 |
14:05:00 | Lingfield | Enough Already | 9 | 10 |
14:05:00 | Lingfield | Tiger Beetle | 9 | 11 |
14:05:00 | Lingfield | Romilda (FR) | 9 | 15 |
14:05:00 | Lingfield | Avon Light | 9 | 15 |
| time | track | horse | runners | Bower Rating |
15:05:00 | Lingfield | Sea Me Dance | 10 | 11 |
15:05:00 | Lingfield | Sharp Distinction | 10 | 11 |
15:05:00 | Lingfield | Artisan Dancer (FR) | 10 | 14 |
15:05:00 | Lingfield | Fair Dinkum (IRE) | 10 | 21 |
15:05:00 | Lingfield | Smith (IRE) | 10 | 22 |
| time | track | horse | runners | Bower Rating |
19:30:00 | Wolverhampton | Whatwouldyouknow (IRE) | 12 | 6 |
19:30:00 | Wolverhampton | Waleyfa | 12 | 9 |
19:30:00 | Wolverhampton | Ibiza Rocks | 12 | 12 |
19:30:00 | Wolverhampton | Harbour Vision | 12 | 12 |
19:30:00 | Wolverhampton | Landlordtothestars (IRE) | 12 | 14 |
19:30:00 | Wolverhampton | Kingwell | 12 | 17 |
| time | track | horse | runners | Bower Rating |
20:00:00 | Wolverhampton | Min Till | 12 | 4 |
20:00:00 | Wolverhampton | Lauras Breeze (IRE) | 12 | 7 |
20:00:00 | Wolverhampton | Rose Light | 12 | 11 |
20:00:00 | Wolverhampton | Snooze Lane (IRE) | 12 | 11 |
20:00:00 | Wolverhampton | Get Set Jet | 12 | 17 |
| time | track | horse | runners | Bower Rating |
20:30:00 | Wolverhampton | Louisiana Bay (IRE) | 12 | 6 |
20:30:00 | Wolverhampton | Kemerton (IRE) | 12 | 8 |
20:30:00 | Wolverhampton | Trusty Scout (IRE) | 12 | 9 |
20:30:00 | Wolverhampton | Moon Over The Sea | 12 | 10 |
20:30:00 | Wolverhampton | Black Smoke (IRE) | 12 | 10 |
20:30:00 | Wolverhampton | Optician | 12 | 15 |
ARAZI91
Mare
formtheory - your query on the amount of Handicaps won by horses who have had <=10 Prev. Hcap runs got me thinking that this might be good information if done by trainer, then it would confirm if the Prescotts, Gosdens, Haggas & Varians of the training population actually do work this way so i thought it would be better to post it here as if you know your trainers MO then it may add something else to the Bowers method - I'll post each trainers P/p Ratio and Average OR during the sequences (i also done <=5 Prev Hcap runs and <= 3 Prev Hcap Runs. I've added some notes at the bottom of each sequence that are applicable to the analysis. The timeframe runs from 01/01/2013 to 02/12/2023. Then there's the question of sample size , a trainer who only has had 150 runs within that timeframe works out at under 15 runs per year so i used convex optimisation to give me a sample size estimate that was optimal - so was left with 64 trainers.Trainers record with horses with <=10 Hcap Runs
Trainers record with horses with <=5 Hcap Runs
Trainers record with horses with <=3 Hcap Runs
As can be seen it's virtually the same bunch of names at the top of each table in each sequence which again supports the theory that "precocity" has a big input into these results - the Official Rating has not much of an influence other than when they start their handicapping careers they are "under-capped" - Charlie Appleby , Haggas , Varian , Stoute have had owners who are prepared to spend money on the Frankels and Dubawis at the sales and they know that their buying horses who generally will improve rather than a Ruth Carr who if she gets a winner might have to wait a long time for that horse to win again due to it's limited ability and some Handicapping Redux - that's not to say Ruth Carr is a bad trainer , she does an excellent job with limited raw material. Prescott tops the tables over the 3 sequences but i would not put him in the same category as Gosden, Appleby or Haggas - it' simply his MO of buying middle distance /staying types and getting them "under-capped" as 2yos by running them over distances that are not optimal to their pedigree - the fact he tops the table over the 3 sequences shows how proficient he has been with this method.
Rob
added - rejigged table 1 - <=10 runs or less as made an arse of it primarily , sorted now.
Last edited:
formtheory
Stallion
ARAZI91
Mare
The P/p Ratios and average OR's in the <=10 runs sequence (in average OR order , highest first) were
I'll not bother posting the <= 5 Prev.Hcaps or <=L3 as the results are very similar but you can see the differences , a 30lb span between Charlie Appleby at the top and Tony Carroll at the bottom.
Rob
I'll not bother posting the <= 5 Prev.Hcaps or <=L3 as the results are very similar but you can see the differences , a 30lb span between Charlie Appleby at the top and Tony Carroll at the bottom.
Rob