Have I found value?
100 gives an indication but still too soon for any conclusions
here is an obviously losing strategy that still shows a profit in 2/50 simulations after 100 trades:
another simulation of the same system has all variants in a clear loss by the 500 trade point: I believe that depending on the actual expected value of the system, it will take a different amount of time to reach the point of near-certainty about profitability. As you'll typically need to work those out using actual results just continue doing what you're doing and check back at a consistent interval e.g. 100, 200, 300, 400 etc.
you could compare each set of 100 to each other and if you're receiving similar stats between the sets that's a sign of accuracy, if they vary keep upping the size of the sets until they start to show the same results
here is an obviously losing strategy that still shows a profit in 2/50 simulations after 100 trades:
another simulation of the same system has all variants in a clear loss by the 500 trade point: I believe that depending on the actual expected value of the system, it will take a different amount of time to reach the point of near-certainty about profitability. As you'll typically need to work those out using actual results just continue doing what you're doing and check back at a consistent interval e.g. 100, 200, 300, 400 etc.
you could compare each set of 100 to each other and if you're receiving similar stats between the sets that's a sign of accuracy, if they vary keep upping the size of the sets until they start to show the same results
You do not have the required permissions to view the files attached to this post.
-
- Posts: 70
- Joined: Mon Aug 31, 2020 3:01 pm
What site is this on?eightbo wrote: ↑Mon Jan 25, 2021 2:49 pm100 gives an indication but still too soon for any conclusions
here is an obviously losing strategy that still shows a profit in 2/50 simulations after 100 trades:
◘.png
another simulation of the same system has all variants in a clear loss by the 500 trade point:
◘◘.png
I believe that depending on the actual expected value of the system, it will take a different amount of time to reach the point of near-certainty about profitability. As you'll typically need to work those out using actual results just continue doing what you're doing and check back at a consistent interval e.g. 100, 200, 300, 400 etc.
you could compare each set of 100 to each other and if you're receiving similar stats between the sets that's a sign of accuracy, if they vary keep upping the size of the sets until they start to show the same results
I've found systems and methods to be difficult to prove/disprove.
I'll give my own personal example. I do a lot of DOBBING. If you're not sure what that is it's quite well explained on the forum and internet. Essentially it's a kind of offset betting.
In early 2017 I thought I had a good set of criteria to make the DOBBING selections. I tested it on 600 races over April - June. I was succesfull around 60% so I thought I'd go live with it and record the results:
I go from July to June each year.
2017-2018 1697 races 58% success rate, £34k traded £5.7k profit.
2018 - 2019 2220 races 56% success rate £81.5k traded £8.3k profit.
2019-2020 1486 races 56% success rate £21k traded £2.8k profit (I lost my income due to covid).
Currently to date I've been struggling to find suitable selections. Previously I could expect to find around 4 good selections a day. Lately it's a miracle if I can find 1.
Currently just halfway through the period 579 races 56% success rate just £6.5k traded and only £579 profit.
The strike rate has been consistent but it's been very difficult to find good selections.
My point with illustrating this is that it's quite possible to find good strategies but watch out for issues like, no of selections drying up or underperforming.
Good luck on your journey.
I`ve been looking if it`s possible to extract dobbing candidates from historical data and my selections look like this. The Yellow ones.
What I have noticed is that despite the form being good, if it has highish odds on the day they attract little movement in odds to get a position to DOB. Have you noticed this?
What I have noticed is that despite the form being good, if it has highish odds on the day they attract little movement in odds to get a position to DOB. Have you noticed this?
You do not have the required permissions to view the files attached to this post.
Your very welcome. Why not trial your strategy with small stakes? Say £2 - £5? You can then gain valuable experience and insight. This will aid you in recognising value and the right markets to trade.
Your'e very welcome and thank you. I hope things do pick up again as I was hoping to get back to how it was before.
Why not trial your strategy with small stakes? Say £2-5? You'll gain valuable insight and experience and learn what works and doesn't.
TBH I've found that my higher priced selections perform just as well. If you want to look at histroric data then Timeform is quite good as it shows the horse's Betfair in-running highs and lows against each previous run. Also Winning Warlock is a free site that gives statistical in-running data for each runner on the card.Atho55 wrote: ↑Mon Jan 25, 2021 5:46 pmI`ve been looking if it`s possible to extract dobbing candidates from historical data and my selections look like this. The Yellow ones.
Kemp 1745.jpg
What I have noticed is that despite the form being good, if it has highish odds on the day they attract little movement in odds to get a position to DOB. Have you noticed this?
I`m using the Promo data to determine the average BSP to IPMIN gap for each runners races since 2018 then picking the runners who have the highest % of races > Average.
It started off as that but started to notice that a few went on to win. Yesterday 93 picks yielded 14 winners but looking deeper the pick runners who were also rank 1 or 2 accounted for 11 winners from a selection of 29 instead of 93.
Interesting what you said about your selections doing ok at higher odds.
At this moment in time, Jazz Legend & Sir Rodney look to match the criteria.
It started off as that but started to notice that a few went on to win. Yesterday 93 picks yielded 14 winners but looking deeper the pick runners who were also rank 1 or 2 accounted for 11 winners from a selection of 29 instead of 93.
Interesting what you said about your selections doing ok at higher odds.
At this moment in time, Jazz Legend & Sir Rodney look to match the criteria.
You do not have the required permissions to view the files attached to this post.
The sample size you need depends on your strike rate.
eg 100 coin flips give you more confidence of an unbiased coin than 100 rolls of a die.
The "gold" standard is often sad to be 100 winners (which usually means hundreds of bets) but you should start getting an idea of whether it's worth continuing with after maybe 50-75 winners.
eg 100 coin flips give you more confidence of an unbiased coin than 100 rolls of a die.
The "gold" standard is often sad to be 100 winners (which usually means hundreds of bets) but you should start getting an idea of whether it's worth continuing with after maybe 50-75 winners.
Hi you take an interesting approach.Atho55 wrote: ↑Tue Jan 26, 2021 1:41 pmI`m using the Promo data to determine the average BSP to IPMIN gap for each runners races since 2018 then picking the runners who have the highest % of races > Average.
It started off as that but started to notice that a few went on to win. Yesterday 93 picks yielded 14 winners but looking deeper the pick runners who were also rank 1 or 2 accounted for 11 winners from a selection of 29 instead of 93.
Interesting what you said about your selections doing ok at higher odds.
At this moment in time, Jazz Legend & Sir Rodney look to match the criteria.
1355 Sthl.jpg
I realise I didn't properly answer your question re the longshots/bigger prices. As I stated I do find that the particular ones I select perform as well as their shorter priced counterparts. However having said that on occasions I do find that some long priced selections perform with credit during a race but don't shorten significantly. Sometimes this is down to the in-running bettor's 'prejudice' against longer priced runners. I always factor in that bigger priced horses are expected to lose and have to 'prove' themselves more in a race.
I've got a couple of recent examples. yesterday I could only find 3 possible DOBS. The first one up was a horse called 'Luna Wish' in the 2.10 Chelmsford. It was generally around odds 21/23 on Betfair earlier in the day but when my bet was struck it was at odds of around 18.0. I couldn't watch the race as the video link was down. During the race Luna Wish's odds seemed to be going the wrong way, drifting so I though yet another poorer performing selection this season. To my surprise it went on to win. My other 2 DOB bets lost.
I looked at the 1.25 Southwell race you highlighted. My concern was that there was likely to be a lot of pace in race with many prominent racers. I was drawn to 'Duke of Firenze' as a horse that could pick the pieces. I should really have DOBBED it (Betfair odds 30.0 available pre off). I decided instead to back it for a place at odds of 11.0. It came through for 2nd.
One thing that occasionally happens with my longer priced selections is if they are suddenly backed late in the minutes leading up to a race. This can spoil their chances of DOBBING. For example a horse is 12/1, Betfair odds 22.0. It gets backed to 8/1 betfair odds now 12.0. Instead of only having to drop to odds of 11.0 in-running it must now go down to 6.6 and may only drop to odds 8.0. Esseentially the value was taken out of it pre-race.
Which site/s do you use to look up the horse's in-running data?
The data comes from here https://promo.betfair.com/betfairsp/prices This is sorted then dumped into a database.
Then it`s just a case of extracting todays runners, determine the tick gap between BSP and IPMIN and measure how each of todays runners has done against an average of the days runners past performances.
The winner of the last race Fayathaan, from 9 races has an average tick gap of 81 and of those 9 races has gone 55.56% of times > Avg.
This is not the average of the entire database but just the runners today. Each runner today will usually have a variety of BSP Ranks they started at. You create an average of all the Rank 1`s, 2`s etc and it`s that value you are comparing against.
Then it`s just a case of extracting todays runners, determine the tick gap between BSP and IPMIN and measure how each of todays runners has done against an average of the days runners past performances.
The winner of the last race Fayathaan, from 9 races has an average tick gap of 81 and of those 9 races has gone 55.56% of times > Avg.
This is not the average of the entire database but just the runners today. Each runner today will usually have a variety of BSP Ranks they started at. You create an average of all the Rank 1`s, 2`s etc and it`s that value you are comparing against.
You do not have the required permissions to view the files attached to this post.