Why do people get an edge from historical data?
-
- Posts: 95
- Joined: Mon Jan 06, 2020 12:48 am
There are plenty of websites, some free and some subscription only (like goal profits) that provide all sorts of historical data that apparently enable people to bet / trade only based on this historical data, e.g. "Team A are playing team B and Team A have scored 80% of the time in the first 30 mins while Team B have conceded 60% of the time in the first 30 mins, so "I'll bet on over 0.5 goals in first half" type of thing. But surely all these stats are already accounted for in the available odds? So why do people get an edge from such "publicly" available data? I understand that this could be possible if one watches the live game as well, but plenty of websites (like the one I mentioned above) claim that watching the game is unnecessary. My guess that this approach only could work in some obscure lower division leagues where the pre-match odds are not as heavily modelled by professional quant teams?
What do you mean by 'why' do they get an edge? Are you assuming there is an edge there? The rest of the post sounds like you're answering your own question in that there isn't an edge to be found in those sorts of stats. If that's what you're saying then I definitely agree with that. Blindly betting based on those sorts of basic, available-to-everyone stats, I just can't see anyway to profit on that, not when there's so many variables that make up a match and not to mention what's actually happening inplay.
-
- Posts: 95
- Joined: Mon Jan 06, 2020 12:48 am
Well, as I wrote in my initial post, websites like goal profits claim that their members are consistently making a profit using such strategies without watching a game at all; so I am wondering whether it's basically all marketing bullshit, or are some people genuinely finding edges using such stats.
Data mining in this manner is statistical deja vu. Just because a horse (or football team) did something or there appears to be a correlation doesn't mean it predicts anything. Correlation is not causation. So it's best to use data to model or understand something, but not to create a strategy based on it repeating or not repeating.
We have tons of data on a number of different sports but put them into Bet Angel from a modelling perspective because of this.
We have tons of data on a number of different sports but put them into Bet Angel from a modelling perspective because of this.
-
- Posts: 518
- Joined: Thu Oct 03, 2019 10:51 pm
But what I don't understand is that football is so unpredictable that how do you make a profit it from it either short or long-term regardless of how much data you have. Unless you able to regularly get high odds it seems almost impossible. Liverpool's defence now seems very weak for example. The historical data for Liverpool over the last three seasons has now been thrown out the window.Euler wrote: ↑Tue Oct 13, 2020 9:11 pmData mining in this manner is statistical deja vu. Just because a horse (or football team) did something or there appears to be a correlation doesn't mean it predicts anything. Correlation is not causation. So it's best to use data to model or understand something, but not to create a strategy based on it repeating or not repeating.
We have tons of data on a number of different sports but put them into Bet Angel from a modelling perspective because of this.
PSG were winning 1-0 yesterday at half time and the opponent had 10 men. I backed PSG to win with more than 3 goals but I traded out after about 70 minutes for a loss. PSG won 4-0.
- wearthefoxhat
- Posts: 3207
- Joined: Sun Feb 18, 2018 9:55 am
eg: Premier League; No 0-0 draw this season (around 40+ games) therefore, if trend continues, there won't be one this season....Euler wrote: ↑Tue Oct 13, 2020 9:11 pmData mining in this manner is statistical deja vu. Just because a horse (or football team) did something or there appears to be a correlation doesn't mean it predicts anything. Correlation is not causation. So it's best to use data to model or understand something, but not to create a strategy based on it repeating or not repeating.
We have tons of data on a number of different sports but put them into Bet Angel from a modelling perspective because of this.
-
- Posts: 518
- Joined: Thu Oct 03, 2019 10:51 pm
So why have all that historical data if its no use to anybody?
It is possible to use historical data to predict by analysing the data and making an assessment on how likely the desired outcome will happen. I have numerous models that do such a thing on the Aus, UK and Ire racing on a daily basis.
Or couple that historical data with what you are seeing happening in-play
From a tennis perspective, history data shows you that player A recovers from a break down, 1st set 70% of the time against an opponent with Player B's serve\break retention stats. The in play data tells you Player A is serving well, has scored x points\earned BP's against Player B's serve in previous games, and shows no sign of injury\disinterest. That gives you a level of confidence to make the entry. It's arming yourself with as much information as possible to give you a reason to trade.
Further to that, I'm pretty sure Team Profit has an inplay stat module and the advice is to use that in conjunction with the historical data. Not been a member for ages though.
From a tennis perspective, history data shows you that player A recovers from a break down, 1st set 70% of the time against an opponent with Player B's serve\break retention stats. The in play data tells you Player A is serving well, has scored x points\earned BP's against Player B's serve in previous games, and shows no sign of injury\disinterest. That gives you a level of confidence to make the entry. It's arming yourself with as much information as possible to give you a reason to trade.
Further to that, I'm pretty sure Team Profit has an inplay stat module and the advice is to use that in conjunction with the historical data. Not been a member for ages though.
- shakey1964
- Posts: 118
- Joined: Mon Nov 19, 2018 2:56 pm
- Location: Usually in my truck
As Atho55 says you can make an assessment of the data and thus make an informed decision. Any pay to look website will claim consistant profits but perhaps the raw data is not enough and further digging is neccessary.
The "soccerstats" timing percentages are complete gobbledygook to me so I made my own which I nearly understand
The "soccerstats" timing percentages are complete gobbledygook to me so I made my own which I nearly understand
Once you have done the spade work you can take the optimum positions from your dataset and use this information to test recent data against. This uses a single Promo sheet
Data from this is fed into the decision engine
which in turn throws out a set of instructions
With this game the only limit you have is your own imagination
Data from this is fed into the decision engine
which in turn throws out a set of instructions
With this game the only limit you have is your own imagination
You do not have the required permissions to view the files attached to this post.
A model can give you an idea of what the odds of something should be, because of crowd mentality the actual offered odds may have value due to over reactions and crowd following. If you can spot this value, then you have an advantage...JustLukeYou wrote: ↑Sun Oct 18, 2020 9:53 amSo why have all that historical data if its no use to anybody?