![]() |
Communal Ratings
All this talk of no interesting strategic threads or no threads in relation to backing winners has prompted me to put forward this idea and see who is legitimate about helping each other to find profitable strategy(s) or who is just throwing hot air about.
I propose that we put together a communal ratings spreadsheet much like the one Shaun has provided with the following ideas: 1) We can use this thread to go through each form angle bit by bit, have a discussion on the best way to score and/or include that form element into the ratings and then move onto the next form element. 2) I can code a modified version of the spreadsheet I use to take into account the approach we arrive at as a group. Note my spreadsheet brings in form from SkyForm, Betfair Form (R&S) and Tatts Race page. Can basically come up with equations, ideas, theories as a group and I can try to put this into a logical equation in the spreadsheet to build a rating. 3) I'll only email the final version and/or versions along the way to those who actively contribute to the discussion. That way no pesky guests can just get along for the ride without contributing. Thats the basic idea at the moment. Like I said in a thread in the general area I almost guarantee a very small portion of those who are actually seeking decent discussion get onboard with this idea. Or if you think this has no merit or have a better idea either post it in here or start another thread. Would be interesting to see what we come up with, who knows it could be crap. But it could also help people think different about their own ratings or about how to approach certain form elements. |
Good idea Josh. My 2 cents worth is this:
1/Just run a bot and collect data of the field at an agreed time (official start perhaps?) 2/ Make use of the ratings that you scrape to find a flaw in the algorithm, which all ratings incorporate. 3/ After collecting price, price rank, volume, volume rank, rating, rating rank for several thousand races, you start to look at what combinations beat the inferred probability and one's that do. A scatter plot usually reveals this quite quickly. 4/ You can also rank or rate track conditions, class of race, distance, number of runners blah, blah. At the end of the day you've created a monster that requires some number crunching. I think fellow participants are best using a bot to collect the data themselves and paste into excel. After that, be they backer, layer or anything in between, they're on their own? OR you subscribe to a proprietory program to do all this for you, which this forum is well serviced for. Nothing new really, just dedication to be patient and see it through. If you're prepared to put in the time you will reap the benefits. |
I've already got my own spreadsheet that I collect data from each night when I get home from work. Save some details into a spreadsheet for the appropriate races and build up my collection of data moving forward.
I'm more thinking we can set it up so people can tweak the allocation of points to each form area themselves if they like similar to how Shaun's works. However its more how the form factors are assessed that we can contribute to and discuss as a group. Basic example: Say the first point we all wish to discuss is Days Last Start - Do we want to look at this in isolation of when the last start was, or do we want to look at it as a collective for fitness based on when the last 2, 3 or 5 starts were? Do we want to allocate more points towards a horse that started 14 days ago because typically they win more often (strike rate) or do we want to allocate more points to a different range because 14 days is overbet? There's a lot of discussion to be had around each form factor that possibly as a consensus we can bring together everyone's good points, put it into practice and it formulates part of a rating and move onto the next form factor. Once we've formulated as many form factors etc as we deem necessary people can then use the spreadsheet to back, lay, record data or do whatever else they like with it. They could find niche's using the ratings on group races only. They could find a niche using just 1 form factor in metro races only. They could find a different combination of scoring the form factors that makes it more profitable based on forward testing it etc. So many possibilities with the final 'ratings' product, but many good minds here that could help contribute to discussion on each form factor. |
Quote:
I threw one up re an ADEL/METRO method I bet on a few days ago. Provided data, time frames, filters, the selection, the race criteria, filters to build strike/rates and, Did not get "ONE" single response. May have been different it was a ADELAIDE METRO LAY PLAN (L.O.L). The only people who benefit from this are the Guests & non contributors. I also asked if an area could be set aside for exactly what you propose. Restricted access to members with a certain no's of posts, Or those who wish to join but must contribute . Again no answer. Good Luck EVAJ an admirable approach but like I found out Think we are flogging a dead horse, you can't make people join, Or contribute it is their right & the owners of the forum to set terms. What you can have is the idea I threw up, To those who do want to help. Much of the important stuff I do just goes to members here I converse, With via email. But as I met them here I try to give something back, To the forum. People who want to post to everyone can continue, those who want to give out ideas but not for the masses can ask the thread be restricted. OR IN THIS DAY & AGE IS THIS CALLED DISCRIMINATION (L.O.L) Cheers. |
Further to Josh. Anyone can say weight the ranking more for this and less for that. That's all it is, an opinion! Without the data to back it up, how can you know? That's where RACE CENSUS and RTW have it all down pat. If don't want to pay for the historical data, then it leaves you no choice but to collect the data yourself. I favour this method as most people will probably use a bot to place their bets. A betting bot needs to trigger at a specified time, so any data you collect must be at the same specified time.
I was looking today at building a VBA dashboard to just punch in a set of ranks like say 1, 4, 3, 6, 1 (each being a weight for a particular rank, price, factor) and it spits out profitability or otherwise for the duration of your database. I think something like that would be a useful tool to assist forumites. I know I've saved many hours using Shaun's scraper. So maybe developing tools to aid forumites find profitable outcomes, rather than spelling out what keys to the kingdom win? |
Member Restricted forum section would be I think a good idea.
|
[QUOTE=Rinconpaul]Further to Josh. Anyone can say weight the ranking more for this and less for that. That's all it is, an opinion! Without the data to back it up, how can you know? That's where RACE CENSUS and RTW have it all down pat. If don't want to pay for the historical data, then it leaves you no choice but to collect the data yourself. I favour this method QUOTE]
You are now where I was 20 years ago. What you are attempting is admirable but will take you forever, To get it right in analysing this way. I can tell you now if I had to do it again no way. Unfortunately R/P is 100% correct. R.2.W & Race Census have it there for you. Whether it is right for you I cannot know. I learnt early, you have to spend money to make money Subscribing to rating services, subs to racing magazines,(PPM) Buying racing books, Punters Choice systems, racing seminars racing two horses (syndicate). While getting "GELDED" along the way I was learning. One thing for the guests & non-subscribers to read. OPEN BM-60 RACES (NO-AGE RELATED). ALL HORSES MIN OF 3 STARTS (BEFORE SCR). SKY-1 (6G) 02/02/2014-05/06/2015. OUT-3,400 (4W) (11.7%). RET- 1440.(B/O/3 -S/P). -1960.(L.O.T 57.6%). AV-DIV-$3.6. BIGGEST DIV $5.0. Last winner 02/09/2014 R6-5 TAMWORTH.. Since then 18 consecutive losses. I have hundreds of these & worse, Regarding wizard panels opening favourites, At certain tracks (MORNINGTON MY GOD), 1st opening fav in some race types. As a bookie open your bag & let the punters throw, Their money in it will not be coming back out. But they are lay plans & who gives a F--- about them. The point here is how can you analyse things like the above, Into a ratings plan & make it profitable by analysing factors. Do not want to put you off but that's the challenge you face. So what can we learn with this post. Review a certain Race-Type. Look at the exact age & sex of the horse(s), That win them What selection(s) in your ratings compliments it. THEN. starts. price distance. track condition(not to mention bias on the day to be recorded) state. country-provincial on & on it goes. That's my 2 bobs worth. Good luck with the thread & the ratings challenge. Cheers. |
Quote:
Cheers. |
Yes members only would be good.
Here is my 2c worth. Punters all regard win and place % highly. I use the opposite in conjunction with the two I mentioned above . Hope it's not too cryptic? |
Quote:
Hi Gary, If you can come up with these type of LAY plans, then you know where not to bet and you bet around those horses using your own strategies. So for those that don't give a flying potato about laying, you're actually finding value around those layable horses. My small contribution to the discussion for backing methods and ratings is this.... In 1974 an American called Fred Davis produced a book titled "Percentages and Probabilities". In the book he displayed tables, which included a new statistical technique for analysing horse racing information: the Impact Value. Soon after other books started to display Impact Value figures such as the excellent "Winning At The Races" by Bill Quirin. Over the next few decades nearly all US handicapping experts mentioned impact values somewhere in their books. punter reading the form - the same info read by a million punters each day The past ten years has seen impact values really take off in the US. Their importance is now worth more than speed figures in 'Handicapping' software packages. Here in the UK impact values are mostly ignored. It seems we are too obsessed with form figures, top trainers and jockeys, 'hyped' horses, newspaper naps and tipsters. Only Nick Mordin promotes the use of impact values. You can read his excellent work in the Sporting Life / Racing Post Weekender, and in his own books. What Exactly Is An Impact Value? An impact value is a statistical technique, which produces an index derived from the percentage of winners having a characteristic divided by the percentage of starters having the characteristic. Lets look at some examples. Last Time Out Winners Examining 69234 runners in 6678 races on the all weather we find that 5548 were last time out winners. Of those last time winners 1107 went on to win this race. IV = % of winners that were LTO winners / % of runners that were LTO winners IV = (1107 / 6678) / (5548 / 69234) Impact Value = 2.07 This means that, on the all weather, a last time out winner is twice as likely to win than a horse who did not win lto. First Run On All Weather 22208 horses ran in 1896 3yo+ Handicaps. 2513 of those horses were having their first run on the all weather. 144 of these 'first timers' won the race. IV = (144 / 1896) / (2513 / 22208) Impact Value = 0.67 In 3yo+ Handicaps a horse having his first run on the all weather is 0.67 times less likely to win than a horse who has run on the surface before. 2nd LTO in 2yo Maiden Stakes 33476 horses ran in 3127 2yo maiden stakes on the turf. 2264 of them were 2nd on their last race. Of this group 636 won the race. IV = (636 / 3127) / (2264 / 33476) Impact Value = 3.01 In 2yo Maiden stakes a horse that finished 2nd last time out is 3 times more likely to win than a horse that did not. An impact value of 1.00 means that horses have won no more or no less than their fair share of the races. Below 1.00 means they are performing below expectation, above 1.00 means they are performing above expectation. How Can Impact Values Find Winners? There are two ways to use impact values. You can either look for positive features to determine if a horse is worth backing, or negative features when trying to eliminate horses. But you should not look at the impact value figure in isolation. You should examine both the impact value and the ROI% figure. A strong positive value for both of them is what you need to find as this indicates a group of horses who are winning more races than they should, and are going of at prices higer than they should. Looking again at the 2yo maidens who were 2nd last time, they may have an impact value of 3.00 but a level stake bet on all of them would have made a 10% loss. In 4yo+ long distance (12f or more) Handicaps on the all weather, Colts have an impact value of 1.84 and have also returned a profit of 32%. These are the kind of positive stats you should be looking out for. This is courtesy of flatstats, but explains it well. I consider Impact Values important, along with prices, they are not the road to riches alone! |
All times are GMT +10. The time now is 06:11 AM. |
Powered by: vBulletin Version 3.0.3
Copyright ©2000 - 2024, Jelsoft Enterprises Ltd.