Originally posted on 10/02/2017:

Quote Originally Posted by KVB View Post
That's all sounding very familiar now Doves.

For current relevant information, I've been running a lot of Excel, do a lot of repeating, and do far too much manual input. On top of that, the main sheet I use to input data is a hand written guide, one I use for every sport. That's part of the sharp forecast and many current state of the market metrics.

The stacking forecast is more involved and requires a lot more data manipulation.

Good stuff Doves, and a bit provocative if I do say so myself.

Excuse me, I didn't smell my toothbrush before I brushed my teeth this morning, I think I need some breath freshener...

Derivatives of stats can be easily done with SQL queries or programming. My old method was to have a database of box scores and derive advanced stats/splits/season stats from that. Haven't worked with any artificial intelligence programming frameworks, but am looking to get my feet wet, gonna check out decision trees. I write code all day, any interface that you are looking to create is probably easier/cheaper than you think. For "live interface" do you mean gathering real time stats? For live betting? Anything can be automated pretty easily. This other poster you speak of sounds like a legend. You never really elaborated much on this "interface". Anything you can imagine can probably be done though.

I look at a lot at rankings of stats, not raw stats, to compare to league averages. Raw stats will change but there will always only be 30 rankings. Not sure how to use the information yet, gonna start playing around with linear regressions with my own stats I create, maybe filtering out weak opponents or other filters. I can still get creative with team stats without having to analyze individual lineups/player collections. Planning on doing regressions based on advanced stats, traditional stats, recent performance, performance against certain types of opponents. The team stats are right next door.

I tend to underestimate points scored and other basics stats, will keep that in mind to not underrate them. I used to scrape web pages, now the name of the game is to figure out which interfaces the websites call behind the scenes. Most websites will do a call to an interface to return JSON data (the format I posted earlier) which is then displayed on the webpage. You can bypass scraping the webpage and directly call the interface yourself if you know the parameters.

One of the interface that MLB.com and MLB At Bat mobile app calls:
https://statsapi.mlb.com/api/v1/game...0170930_013023

You can use this data to run regressions/make predictions during the first few innings, even first inning, maybe check out spin rate, movement, or metrics of the pitcher to determine when a starter doesn't have "his stuff" and make a live bet.

Here is one of the interfaces that NBA.com calls, hustle team stats. You can pass parameters to get data just for the first minute of the game, over a certain time period of the game, etc. Before you would have to scrape the play by play data and parse it which would be a huge task. Now you call the API, give me the stats for the first 5 minutes of the game.

http://stats.nba.com/stats/hustlesta...eID=0021601154

Here is an example what NFL calls

http://www.nfl.com/liveupdate/game-c...12200_gtd.json

You can paste the URL/data into this site to make it more human readable if you want to check it out (prettify)
https://jsonformatter.curiousconcept.com/

Scraping things like odds, news reports, would be harder potentially, but gathering sports statistics is easier than ever.