Hi all, when it comes down to the Betfair charts, we cannot download this or use any time before we come to the markets I.e. if I come to the pre off 5 mins before the off all the charting data can be visualised by me but I can’t download it or use it in my automation.
Issue is for me to paint a picture I’d need to be storing the value of prices when the markets first open and I do y think that’s even possible to do as im not allowed infinite stored values.
Im thinking That I have two options ? Either I take the stored values and export them into excel as soon as the markets open for them to be built there and then equations applied the the data OR there is a bet angel graph which has stored all of the prices and volumes from the previous opening hours before 5 mins since I joined the market ?
Anyone here have any suggestions on this please
Download price data
-
ShaunWhite
- Posts: 10424
- Joined: Sat Sep 03, 2016 3:42 am
Export them as they happen. For lower frequency samples you might be able to use a history list.
https://www.peterwebb.com/bet-angel-new ... et%20Angel.
https://www.peterwebb.com/bet-angel-new ... et%20Angel.
Thank you Shaun for the link that could be a useful read.
After having some time. I know you do quite advance stuff, but what I'm after doing is basically getting all of the graph data thats displayed below the trading ladder.
Its still early to think this as I'm still doing lots of research and reading. But it looks like that bet angel ca record the charting. but you have to have the market open ready for it to be recorded i.e. 24hours before the race. that way all of the price and volumes get stored in the back ground. I'm hoping that the advance charting is capable of storing this for me, else it looks like an API export to python job or another way, but these become alot more complex and challeneging.. .
After having some time. I know you do quite advance stuff, but what I'm after doing is basically getting all of the graph data thats displayed below the trading ladder.
Its still early to think this as I'm still doing lots of research and reading. But it looks like that bet angel ca record the charting. but you have to have the market open ready for it to be recorded i.e. 24hours before the race. that way all of the price and volumes get stored in the back ground. I'm hoping that the advance charting is capable of storing this for me, else it looks like an API export to python job or another way, but these become alot more complex and challeneging.. .
-
ShaunWhite
- Posts: 10424
- Joined: Sat Sep 03, 2016 3:42 am
You can't access info that happened before you start BA. The charts at the bottom of the ladder aren't data as such they're images (png files I think) that are downloaded every so often.
Iknow thats why I said I need to open then 24hours before the race. so it logs all of that data. However do you think its best to export the data into excel real time? and apply the different calculations to stuff or can it all be done inside bet angel?
-
ShaunWhite
- Posts: 10424
- Joined: Sat Sep 03, 2016 3:42 am
24hrs is a bit excessive, what's happening 4hrs ago or even an hour ago that would change your option of 'now' ? And you probably won't want to store a value every 20ms so you might get away with a history list sampled at a lower frequency (guessing, I've never used them).
But as always, what you want to do with the data determines what you store, how often and where. If it's for auto then I guess you're interested in ranges and volume, so having all the info to draw a chart and 'look' at the shape of it is overkill.
But a bit of a rule of thumb from data modelling, you need about 10-20 times as much history as you want to predict fwd. So if you want to 'guess' what a price will be in 20s time you'd only need about 5mins of history. And whatever time history you have should be cut into around 3-600 samples. So a 5min history would be sampled at about 0.5 to 1s intervals. A 1 hour history would be samples taken about every 10s... Or thereabouts. So whether it's 24hrs or 24s you have roughly 3-600 samples.
One other advantage of that is your deltas (rates of change) can be sample frequency based not time based. Ie the change in the the last ten values stored, not necessarily the change over the last 10s. That way the scale of the calcs matches the scale of the history you're analysing.