Hi,
I would like to identify when a Selection is genuinely backed and or Layed not just traded back and lay.
A Selection could have a high matched volume, but hasn't moved much in ticks because there was just as much support wanting to Lay as to Back so it may only fluctuate only +/- a few ticks up and down?
The selection I would like to see is one that has a substantial Matched Volume or matched % compared to the total matched but has moved its price up or down, because the money has forced it too.
I do not know how to do the Math for and hoping someone may be able to help or suggest another way?
Many Thanks
Michael
Many Thanks
Selection movement Math
It's not simple to do because you need to compare the (% of total matched) with something else(s). For example. you'd need to have a benchmark for what % of total matched is 'normal' for a selection at each price point. So you could build a historical database and compare (although that's fraught with issues.)1856 wrote: ↑Sun May 03, 2020 8:32 amHi,
I would like to identify when a Selection is genuinely backed and or Layed not just traded back and lay.
A Selection could have a high matched volume, but hasn't moved much in ticks because there was just as much support wanting to Lay as to Back so it may only fluctuate only +/- a few ticks up and down?
The selection I would like to see is one that has a substantial Matched Volume or matched % compared to the total matched but has moved its price up or down, because the money has forced it too.
I do not know how to do the Math for and hoping someone may be able to help or suggest another way?
Many Thanks
Michael
Many Thanks
Or you could just use price as a single signal, which probably gets you most of the way there.
Traders also genuinely back & lay selections too. If you're trying to identify it, then its likely to have already happend by time a calculation has identified it. If you're spending time looking for what's happened before you or a piece of automation will act then you're always behind what's happend, if you're wanting to find them because you plan to bet on them, then it would probably be a point where it's no longer value if it's already been backed or layed.1856 wrote: ↑Sun May 03, 2020 8:32 amHi,
I would like to identify when a Selection is genuinely backed and or Layed not just traded back and lay.
A Selection could have a high matched volume, but hasn't moved much in ticks because there was just as much support wanting to Lay as to Back so it may only fluctuate only +/- a few ticks up and down?
The selection I would like to see is one that has a substantial Matched Volume or matched % compared to the total matched but has moved its price up or down, because the money has forced it too.
I do not know how to do the Math for and hoping someone may be able to help or suggest another way?
Many Thanks
Michael
Many Thanks
- ShaunWhite
- Posts: 9731
- Joined: Sat Sep 03, 2016 3:42 am
That should be a matter of comparing implied chance with the % of the vol on that runner. 50% chance, 50% of the vol. But all measurements like this are hit and miss. They might hold true for a bookie maintaining a balanced book but with exchanges being mutilple participants there's no inherent mechanism to force a relationship between traded vol and price.
Last edited by ShaunWhite on Mon May 04, 2020 2:40 am, edited 1 time in total.
- ShaunWhite
- Posts: 9731
- Joined: Sat Sep 03, 2016 3:42 am
... The other issue you have is defining the word 'move'. What's the timescale, is it ticks or %age etc?
I think the way to do what you want is by comparing the volume of back bets with the volume of lay bets on a selection and looking for pressure on one side or the other. You can obtain a rough approximation of that via Excel data capture, looking at the vol change and ltp to determine which side volume is being taken on. However the data updates aren't issued for each market change and there's a degree of conflation (batching) that nets orders placed and taken within the update cycle. The ltp could be a small back bet when 5 other transactions in the same 20ms window were all lays. .... And you also have the complication of cancelled orders; prices can move because money is withdrawn as much as it moves by being taken.
Before getting into specific maths, you need to specify what parameters and measurements you actually want to include in your calc and how you can obtain them. And then define what range of values you want for your indicator/singal, perhaps from -1 to 1. Form there it's trial-and-error to some extent, juggling your variables until your signal starts to correlate with the near future.
I think the way to do what you want is by comparing the volume of back bets with the volume of lay bets on a selection and looking for pressure on one side or the other. You can obtain a rough approximation of that via Excel data capture, looking at the vol change and ltp to determine which side volume is being taken on. However the data updates aren't issued for each market change and there's a degree of conflation (batching) that nets orders placed and taken within the update cycle. The ltp could be a small back bet when 5 other transactions in the same 20ms window were all lays. .... And you also have the complication of cancelled orders; prices can move because money is withdrawn as much as it moves by being taken.
Before getting into specific maths, you need to specify what parameters and measurements you actually want to include in your calc and how you can obtain them. And then define what range of values you want for your indicator/singal, perhaps from -1 to 1. Form there it's trial-and-error to some extent, juggling your variables until your signal starts to correlate with the near future.
That's not inherently true.ShaunWhite wrote: ↑Sun May 03, 2020 1:26 pmThat should be a matter of comparing implied chance with the % of the vol on that runner. 50% chance, 50% of the vol.
Consider a race with two 3.0 chances, compared with a race with a 3.0 fave, and the second fave at 8.0.
- ShaunWhite
- Posts: 9731
- Joined: Sat Sep 03, 2016 3:42 am
There's a couple of ex-bookies on the forum who would know for sure and I'm happy to be corrected, but it's my understanding that if you were a bookie then in the situations you describe your book would ideally hold.Anbell wrote: ↑Mon May 04, 2020 12:27 amThat's not inherently true.ShaunWhite wrote: ↑Sun May 03, 2020 1:26 pmThat should be a matter of comparing implied chance with the % of the vol on that runner. 50% chance, 50% of the vol.
Consider a race with two 3.0 chances, compared with a race with a 3.0 fave, and the second fave at 8.0.
Race 1: Fav 1 = 33.3% vol, Fav 2 = 33.3% vol, remaining 33.4% across the rest of the field.
Race 2: Fav 2 = 33.3% vol, Fav 2 = 12.5% vol, remaining 54.2% across the rest of the field.
But on an exchange that's not necessarily the case because nobody is driving the book towards any sort of balance and it's technically possible for a 10/1 outsider to have seen the most money change hands without creating a sitiation where anyone was unduely exposed on that particular horse.
I don't like guessing so taking all UK horse racing for Feb20 (2653 selections), and looking at the percentage share of market volume vs the implied chance of each horse, I get this below. The market share is approximately in line with implied chance but it's guide rather than a given.
Bearing these results in mind, I'll change my reply to a 50% shot taking between 40% and 75% of the book
You do not have the required permissions to view the files attached to this post.
I've learnt so much from your many posts that I just *had* to jump in and give a little back for once
I did the analysis a few years back for AUS, and the faves consistently over-weight to quite a large extent, meaning that the others all underweight, of course. From memory, from my data-set (I think I only used from 2.0 to 100 cutoff) the breakeven was about 4.
ALso - what's that green line? Should it go from corner to corner? If so, you can see that most data points are above the line if the implied chance is > (at least) 40%
I did the analysis a few years back for AUS, and the faves consistently over-weight to quite a large extent, meaning that the others all underweight, of course. From memory, from my data-set (I think I only used from 2.0 to 100 cutoff) the breakeven was about 4.
ALso - what's that green line? Should it go from corner to corner? If so, you can see that most data points are above the line if the implied chance is > (at least) 40%
- ShaunWhite
- Posts: 9731
- Joined: Sat Sep 03, 2016 3:42 am
The green is just a linear trend and theoretically should run from 0,0 to 1,1. You're spot on though, favs are slightly over backed and lomgshots the opposite. The tipping point looks like it's iro 0.25....ie those priced at approx 4.0. I'm glad you've found some of my posts interstesting, do enough and a few might be worthwhile just from the law of averages. They're all just opinion and speculation so I wouldn't take everything as fact if other people have a different take.
- ShaunWhite
- Posts: 9731
- Joined: Sat Sep 03, 2016 3:42 am
More money 'on' it (in the traditional sense) or more traded vol?
I'd have thought it would be a bit of both, disproportionatly high traded vol because so many traders focus on the fav, plus the punter bias towards what's perceived to be the 'favourite'.