Trading or betting automation, and with what data?
-
- Posts: 10
- Joined: Tue Feb 23, 2021 10:40 am
Thanks Atho, you see, this is exactly what I have been doing so far, playing with implied probs and try to find correlations with other factors. But nothing lasts (and it is not supposed to, ideally). Did you manage to find "rules" that never change?
I tend to analyse the horse data rather than grey data so the example I posted is new as I just wanted to show you what could be done. Saying that, the underlying concept is not new so be interested to see how it does myself.
If I was to look for rules that never change then they would likely be odds based rather than Rank or Trap.
If I was to look for rules that never change then they would likely be odds based rather than Rank or Trap.
Why? value doesn't disappear based on todays 'ROI'
You can't make money in the former without the latter.liltbrockie wrote: ↑Wed Feb 24, 2021 9:56 amDefinitely exploitable if trading... not so much for winning bets.
You think wrong. What on earth are you basing this off?
- The Silk Run
- Posts: 918
- Joined: Mon May 14, 2018 12:53 am
- Location: United Kingdom
I would assume trading greyhound markets is difficult. But, their are successful traders in the domain, with long term success. It only takes a track manager to put the bowser out before every race and that's all that form work, data, drained away !!! And a Tesco colleague could price up this market with his / her price gun better than the bookmakers, exchanges.
- ShaunWhite
- Posts: 9731
- Joined: Sat Sep 03, 2016 3:42 am
This! And beautifully put too.LinusP wrote: ↑Wed Feb 24, 2021 3:21 pmYou can't make money in the former without the latter.liltbrockie wrote: ↑Wed Feb 24, 2021 9:56 amDefinitely exploitable if trading... not so much for winning bets.
I laboured for way too long under the misconception that "trading" was different to value betting. Once the penny dropped it all fell into place.
??ShaunWhite wrote: ↑Wed Feb 24, 2021 6:02 pmThis! And beautifully put too.LinusP wrote: ↑Wed Feb 24, 2021 3:21 pmYou can't make money in the former without the latter.liltbrockie wrote: ↑Wed Feb 24, 2021 9:56 amDefinitely exploitable if trading... not so much for winning bets.
I laboured for way too long under the misconception that "trading" was different to value betting. Once the penny dropped it all fell into place.
-
- Posts: 10
- Joined: Tue Feb 23, 2021 10:40 am
Of course, both are based on understanding which prices are off with respect to the real probabilities (or will be off in the future). I am having problems, as most, in understanding where the clues on the real probs are.
ShaunWhite wrote: ↑Wed Feb 24, 2021 6:02 pmThis! And beautifully put too.LinusP wrote: ↑Wed Feb 24, 2021 3:21 pmYou can't make money in the former without the latter.liltbrockie wrote: ↑Wed Feb 24, 2021 9:56 amDefinitely exploitable if trading... not so much for winning bets.
I laboured for way too long under the misconception that "trading" was different to value betting. Once the penny dropped it all fell into place.
If you have some data try comparing the implied probability v what actually happens. Take 2.0 odds. Does it win 50% of the time.
Easier to =FLOOR(BSP,0.01) to lump them together rather than deal with millions of individual BSP`s.
If you are struggling to get your head round it PM me and I will explain further
Easier to =FLOOR(BSP,0.01) to lump them together rather than deal with millions of individual BSP`s.
If you are struggling to get your head round it PM me and I will explain further
-
- Posts: 10
- Joined: Tue Feb 23, 2021 10:40 am
Hi Atho,
I have done this many many times.
The figure below is derived from the BSP of about 90k UK and AUS races, without factoring the commission.
This is a plot for all the data, but of course I have made it on: a) shorter timespans; b) segmenting by week-day; c) segmenting by track; d) segmenting by fav; e) accounting for volume.... and many other combinations.
Long story short, yes, each day the blue line has bumps above the red line, but all of that regresses to the mean over time, so any strategy on this is doomed to fail for the gambler's fallacy. I have tried also backtesting systems that look where the bumps where yesterday (or last week on the same day, or over n previous days) and use those intervals for the day after, but it does not work in the long term. The BSP data are extremely efficient in the long term. That somehow is a relief, but tells me that there is need of external data to make a viable strategy.
I have done this many many times.
The figure below is derived from the BSP of about 90k UK and AUS races, without factoring the commission.
This is a plot for all the data, but of course I have made it on: a) shorter timespans; b) segmenting by week-day; c) segmenting by track; d) segmenting by fav; e) accounting for volume.... and many other combinations.
Long story short, yes, each day the blue line has bumps above the red line, but all of that regresses to the mean over time, so any strategy on this is doomed to fail for the gambler's fallacy. I have tried also backtesting systems that look where the bumps where yesterday (or last week on the same day, or over n previous days) and use those intervals for the day after, but it does not work in the long term. The BSP data are extremely efficient in the long term. That somehow is a relief, but tells me that there is need of external data to make a viable strategy.
You do not have the required permissions to view the files attached to this post.
-
- Posts: 10
- Joined: Tue Feb 23, 2021 10:40 am
Since I am here I will also share the only thing that I have found that is consistent.
The plot below is the result of a strategy based on a logically-justifiable inconsistency in the SPs distribution in the Aussie dog market. The data are about 20 months worth of races.
The strategy is a betting strategy mixing back and lay, and the results consider the higher commission paid on aussie dogs. The criteria used to determine the dogs to bet are very simple, and the only filter on BSP is loose enough to suggest that this could work by using the near-price close enough to race start.
I am hesitant to implement it at the moment (lots of swings, although they are in the nature of such systems, but I will implement it sooner or later, since it is remarkably consistent), but I think it is interesting because the same does not happen for the UK market, even changing the parameters. I wonder if it is because AUS has more runners, or less volume, or more/less bots....
The plot below is the result of a strategy based on a logically-justifiable inconsistency in the SPs distribution in the Aussie dog market. The data are about 20 months worth of races.
The strategy is a betting strategy mixing back and lay, and the results consider the higher commission paid on aussie dogs. The criteria used to determine the dogs to bet are very simple, and the only filter on BSP is loose enough to suggest that this could work by using the near-price close enough to race start.
I am hesitant to implement it at the moment (lots of swings, although they are in the nature of such systems, but I will implement it sooner or later, since it is remarkably consistent), but I think it is interesting because the same does not happen for the UK market, even changing the parameters. I wonder if it is because AUS has more runners, or less volume, or more/less bots....
You do not have the required permissions to view the files attached to this post.
I wondered if it was because in Aus BF is a secondary player to TAB but in the UK it is a major player and can align the various platforms such as Exchange and Sportsbook. Whatever the reason, I agree that the markets are different.
- Crazyskier
- Posts: 1166
- Joined: Sat Feb 06, 2016 6:36 pm
Cardano wrote: ↑Wed Feb 24, 2021 11:31 amI have one greybot place market lay only strategy that is profitable long term, though the drawdowns can be long and somewhat painful, even with 100 races traded per day. But I've never had a losing month overall and am slowly scaling it. The main issue is place market liquidity and how late the money comes.I'm also experimenting with the aussie dogs overnight, which has some insanely short-priced favourites overturned, but the liquidity again can be an issue, so leaving bets in the market until suspend can mean some value can be gained occasionally.
CS