xG - Any use from a footy betting perspective?

[harry_rag]

Idle musing, mainly (for me) from a player perspective but let's pitch it at a global level, match xG, team xG etc.

I know most of the old pros treat it with disdain (as they do pretty much anything invented this century) e.g. Lee Dixon with a recent "don't get me started on xG" comment. Might as well have said "don't get me started on any of the plethora of subjects I don't understand"!

Any views welcome on it's value or otherwise from a punting perspective. Here's my dilemma from a goalscorer angle:

• Player A has scored 10 goals and has an xG of 10 - so far so good, I can take his scoring stats at face value when considering betting on him to score
• Player B has scored 10 goals and has an xG of 5 - do I conclude he's been lucky and treat that figure of 10 with a degree of scepticism or do I assume he's a better than average finisher from any given opportunity?
• Player C has scored 10 goals but has an xG of 15 - has he been unlucky or do I assume he's a poor finisher?

In terms of the latter scenario, Dominic Calvert-Lewin is pretty much the worst player for underperforming his xG and my take (based on constantly losing on him when thrown up as a system bet) is that he's a dodgepot and best not backed unless with someone else's money. Players like Haaland and Martinez tend to outperform their xG and the idea of being better than average finishers doesn't seem too far fetched.

My feeling is that xG could be of some use (even if relatively limited) in my betting but I'm struggling with how to interpret it.

[MCLARKE]

From what you say it would appear to be a good sign if the XG is better than the actual.

I guess you would only know after you had kept some stats.

I've always avoided the goals market because the overall overound is so high but you seem to have the football knowledge to overcome that. Unfortunately I don't have that knowledge, I am driven by the stats.

[harry_rag]

55 minutes ago, MCLARKE said:

I've always avoided the goals market because the overall overound is so high but you seem to have the football knowledge to overcome that. Unfortunately I don't have that knowledge, I am driven by the stats.

A bit of both to be honest. Where I'm at now is the culmination of over 20 years of betting on football in a certain way (largely based on using the spread prices to try and find value in the fixed odds markets). I've ended up specialising in betting on goalscorers. I don't regard myself as an expert on football as a sport but I'd claim a degree of expertise in football as a betting medium. Much of that is stats driven though.

I now have stats for 8430 players who were all priced at 15+ (based on midpoint) by the spreads. Blindly backing every player at best bookies price would have had an ROI of -3.29%. That's not too bad a starting point to try and sort the good bets from the bad ones, especially when you factor in how often you'd get better on the exchange. The shorter the price the better the returns. Blindly backing every player >35 has a positive ROI of over 7% but that's based on a small sample of 209 players as I used to disregard those players.

The ROI during the "two firm" era was -4% from 6082 bets whereas, post-SPIN, it's -1.4% from 2348 bets. That suggests a bit of short term good fortune that would explain why returns have been so decent over the last few months. Better batten down the hatches for a reversion to the mean! 

 

[harry_rag]

Read a bit to suggest that it might be a bit too random to rely on hugely for my betting. Might take a look for players who could be flying under the radar for goals from free kicks or outside the area, based on stats for shots from free kicks and average shot distance.

I’ve won twice on one player (from 4 bets) who hadn’t scored when I noticed how high in the rankings he was for free kicks. He had 0 goals but an xG of 1.5 and had scored a few goals previously. He’s gone in twice now (both posted on here) at 14/1 and once at 50/1 from outside the area. I think it might be possible to find a small “stable” of similar players worth following at big prices till the end of the season.

[harry_rag]

Just been doing a bit more analysis of the system data. 8430 players recorded now. ROI for blindly backing all at best price is -3.29%. It was -4% in the "2 spread firm" era but -1.4% since then. I can only assume that's random variance as the players in the overall data sample don't know there's only 1 spread firm now!

There were 1956 players where computer said "BET" and they returned 117.73 points profit with an ROI of 6.02%. It was 5% during the old era but 8.4% since. Again I assume at least some of that better return is down to variance but it would be nice if some of it was down to the tweaks to the system based on the analysis I did.

The 6474 "NO BET" players  returned a loss of 394.68 points with an ROI of -6.1%. That was -6.65% in the old era and -4.62% since.

There's been a huge change in the mix of which of the 2 "BET" indicators has made most of the profit (one has gone from -11.36% to +18.55%!) so I'll be interested to see how that develops as the new data sample increases.

Got to be happy overall though and just keep on doing what I'm doing. Real money ROI YTD is 11.5% (including study based bets on players who either aren't in the data or didn't qualify as an automatic bet). Last year it was only just over 3% for all bets so I'd be surprised if it stays in double digits but we'll see how it goes.