Factoring in home advantage

Have a read of this paper from the football data website, it will give you a few pointers:

http://www.football-data.co.uk/ratings.pdf

how do you decide the sides are equal in the first place letsalldance?

The home side ground advantage is between 3 and 4 goals
per 10 matches, and your calculation of 44/29/27 is about
right. Have a look at the spread superiority and compare
the diff with the book prices offered, and after a while
you may get the idea. in the example you offer, the spread
superiority (mid price) would be around 0.3 and 0.4. GL.

some teams at home have a greater advantage than others ,some years i have seen a team with a better away record


in summary

don't waste your time reseraching unless it's one team

Thanks for the comments, I'll check out the PDF and try and get my head around spread superiority.

The deciding about whether team's are equal would be purely based on my own opinion(granted team's are very rarely completely equal in reality, it was more just for the example) I believe I have a good knowledge of the league and all the teams. So in a sense once I've factored in my knowledge/opinion and team news all I've got the factor in is home advantage.

So it seems the 44/29/27 is a good starting point for that. I've also been adjusting slightly if a team as a real bad away or home record comparable to their quality,Fulham and Stoke come to mind has being extreme examples of being especially strong at home but historically poor away from home. Hence I would nudge a few percentage points in the appropriate direction. Thought West Brom were a cracking price this weekend for example given Stoke's poor away form and the fact that West Brom have been in great form recently. Granted they lost but I thought they where value given I consider West Brom the better team and the added home factor over what is typical.

was it two years ago? when in division two, home sides had a freakishly appalling record week in week out

coachmaster, that is what happens with real figures, they jump about,
but that doesn't the stats are wrong.
Try it with roulette, spin the ball 36 times, ignore the zero, I would
be amazed if the evens chances came out level.

Personally I would probably have the % chances as  44/29/27 (HW/D/AW). Is this totally wrong or am I on the right lines?

Yes you are on the right lines for sure

I wouldn't pay any attention to 'strong home teams' myself .In the premier lg , long term, pretty much everyone gets 60% of their points at home and 40% away  .You did get some teams over 1 season that have a different balance but these things never last

eg in 2010/11 Man United got about 70% of their points at Home and only 30% away , so the tv pundits had them fixed as a strong home team that struggle a bit away .Then 2011/12 , with much the same players ,they got almost 50% of their points away, and over the 2 seasons the split was about 60-40 as you'd expect

Kennels,  stats are just that ..stats. Everything evens out from point zero  (eventually)  even a roulette wheel

Looking for patterns using home advantage is very unreliable . Stop it now !

Correct, everything evens out eventually, but in the short run,
like a season or two, they jump about. Concentrate on one team
who buck the stats in a season, or even half a season and it
is more than likely to return to the norm. GL.

I'd have home advantage at somewhere around 2.1, and would probably have the draw at a slightly higher price, somewhere between 3.5 and 3.55 but that's just me.

Dec 4, 2012 -- 10:18PM, Coachbuster wrote:

Kennels,  stats are just that ..stats. Everything evens out from point zero  (eventually)  even a roulette wheel Looking for patterns using home advantage is very unreliable . Stop it now !

Agree with Coachbuster

Are you proposing to use the average results distributions to price up individual games letsalldance?

hejik, what exactly are you agreeing with
coachmaster over ?

Dec 4, 2012 -- 11:52PM, kenilworth wrote:

hejik, what exactly are you agreeing with coachmaster over ?

Interesting question Kenilworth.
I don't actually know

Suppose I agree with this vague statement "Looking for patterns using home advantage is very unreliable"

What I should have said is, average results distributions should not be used to evaluate and price individual match ups.
I am also not clear as to what the OP proposes to do so perhaps shouldn't be commenting at all until he clarifies

hejik, the OP is not proposing to do anything, but asking 'how do
I factor in home advantage' into the betting, not an easy
question to answer on a forum.

Regards your highlighted sentence above, I think it's possibly
reliable to a degree. This weekend we have the two Manchester
sides in opposition and the compilers have the home side about
3 fractional ticks superior to the visitors, yet lie 3 points
behind and a bigger price for the Championship, the price has
home advantage built in. If the match was at OT I am sure the
betting would be reversed, United being the 3 ticks fav, the
advantage of being at home. A simple example but significant.

Yes Kenilworth but nothing to do with Average Premiership Results Distributions

hejik, not sure what you mean by that, could you
clarify/

I mean the compilers haven't given thought to the average results distribution for the entire league in their sums for this particular game

Why do you think that? Do you have a different idea of
what the betting should be?

I'll grant you may be some reversion worked in to keep your prices square
But if you're saying that if pricing for a game between Manchester United and QPR at Old Trafford then you should base most of your decisions on average results distributions for the premiership then you are very very wrong

Wasn't directed at you Kenilworth in particular
Was a general observation

I would do it on...
League positions
Matches won..
Matches drawn..
Goals scored..
Goals conceded..

ln that match of course I already
know what the betting was, so point
in guessing but if the match was at
QPR I would have had United either
8/15 and 1/2.

i wouldn't place too much on league position either kenilworth
or matches won/drawn etc..

west brom were 3rd and aren't that good
and, does wins against fulham, wigan and stoke equate to same form as wins over man united, chelsea and man city?

that would put me off that approach

hejik, there has to be some subjectivity with sides
like WBA, as there is with Liverpool, Arsenal, ManU,
Swansea etc these, but what they are doing at the
moment is a consideration.

How good are Wba? How bad are Liverpool?
What about Tottenham, or even ManU?

despite the general belief that Liverpool are short every week, that is not the case at all in my opinion
or the millionaire odds setters it would seem
that's my point about focusing on recent results and league tables etc.

So you believe they are being priced correctly,
but perhaps just on a poor run ?

I believe they're simply struggling to see out matches kenilworth
they are certainly performing at a level much higher than most seem to think
obviously goals are a big thing in football though
but it cannot continue as it has been

I don't see what relevance average results distributions could be to me though in my analysis

Interesting conversation we've got here in my opinion. I'm not sure what you mean by 'average results distributions' but what I'm looking for is just a general over-riding idea of much impact home advantage as in the Premier League specifically. I don't know if one can general be applied to the league or whether you'd have to calculate separately for each team.

My own approach is more or less purely based on the underlying quality of the team as well as a bit of current form. I don't tend to look at the league table that much because it can be a bit misleading. West Brom being a good example this season I guess, yes they are in good form you've got to factor in that but in all reality I would still only have them priced as a good mid-table side rather than a top four side obviously. On the contrary I think Liverpool are contentiously priced up as if they are challenging for the top four, which in my opinion clearly isn't the case.

The Manchester derby this weekend is a cracking example. For me there's very little to split the two teams and therefore I would argue that the 44/29/27 mentioned before would be the "right prices" for this matches which spookily match the current odds available exactly (2.28/3.4/3.6). Depending on your interpretation of home advantage you could make an argument for finding value in these odds desperate drawing the same conclusion that the game would be a coin toss at a neutral venue.

letsalldance, a good post and I would say you have a
good idea of what is required, GL

letaalldance,

a little more to Liverpools price than meets the eye. They have dominated in almost everymatchthey've played. If they could put away chances then I assure you they would be challenging for top 4.

How have you arrived at 44/29/27?

hejik, are you a Liverpool supporter?

NO I AM NOT!

ok, you don't have to SHOUT !lol

Letsalldance. - How would you rate Eve v Tot then. I would say that they are currently at the same level and the price there is 2.1 for Everton.

I'm not saying which price is right and wrong, but it's interesting that you have the Man C price at 2.28 and the Everton price at 2.1. for 2 games both where you may think the teams are equal.

I would rate Spurs far superior to Everton

Would actually make them favourites Away

i wouldnt go as far as to say far superior imo...however id fancy everton to beat spurs at home and likewise spurs to beat everton with home advantage

When Man C played spurs on 10 Nov they were 1.61, when they played Everton last week they were 1.58. Make of that what you will.

The only thing i would say in addition to this is that Spurs form has probably improved since 10 Nov.

Hejik - Just out of curiousity do you think that Liv are fair price at 1.88 away to Udinese without Gerrard and Agger. Personally i feel it's a bit short.

U.A. I would say Spurs have had better results than Everton
but haven't necessarily been playing better. There is a diff.

Dec 6, 2012 -- 5:56PM, U.A. wrote:

Hejik - Just out of curiousity do you think that Liv are fair price at 1.88 away to Udinese without Gerrard and Agger. Personally i feel it's a bit short.

UA, I have no idea in this match, haven't even looked at it.
Suspect that price is more  about Udinese
Don't think they can qualify anyway? Don't know
Wouldn't be backing though at that price at first glance but have not looked in any detail

A lot to like about Tottenham giving 1.25 tonight with Pinnacle though
2.0

Ken - Maybe, i suppose all I am saying is currently the two teams are fairly equal if the game was to be played at a neutral venue (that's just my opinion of course). Similar with Man C v Man U (although I personally think Man U might be ever so slightly shorter odds at the moment at a neutral value).

It's quite interesting the difference between the 2 prices for the home team. Of course i may be completely wrong with my opinions. What do you think?

Was exactly my point in first place UA
Isn't near as much a factor as some seem to think
What happens if you have side that performs better away from home?
Are you going to give them "Home Advantage" still?

looking at average results is a waste of your time
the fact the 2 weekend games that have been mentioned just happen to coincide with the averages is probably why they've been picked
what about all the other games that don't fit at all?

What happens if you have side that performs better away from home?

In the long term, there aren't any .

By the way if you had layed Liverpool for a set liability for the last 5 seasons for every single game in the Premiership (not including this one) according to my calculations you would have a Return on Investment of 13.5%. And that's after taking off commission of 5%.

Fairly sure you would be in profit at this point in the season as well.

U.A. I'm not disputing your figures but how do you know?
Do you have Liverpool prices and results for those 5 seasons?

Dec 6, 2012 -- 10:20PM, tobermory wrote:

What happens if you have side that performs better away from home?In the long term, there aren't any .

Yes Tobermory I am aware of this

But how long term are you going?
Season 02/03 or 04/05 has no relevance at all today
And many sides will display a negative home rating even over a couple of seasons

Of course many sides will display a negative side over
a couple of seasons, 'a couple of seasons' in, say, the
Premier League home matches is just 38 matches which is
actually a very small sample to draw a conclusion from.
spin a ball on a roulette table 38 times and see the results
you get, follow that by 380 spins and see the difference.

Hi Ken.

In a word Yes. I have all EPL match prices (from Betfair) going back 5 seasons before the start of this one. They are not starting prices, but over 99% of prices recorded are on the day of the game, and a general principle is to record them as close to kick off time as possible. 100% are recorded either the day before or on the day of the game.

You might get a slightly different result if you look at starting prices but in my experience I find that Liverpool tend to shorten before kick-off so I would expect the percentage to be fairly similar or maybe even ever so slightly higher.

Dec 7, 2012 -- 9:18AM, kenilworth wrote:

Of course many sides will display a negative side overa couple of seasons, 'a couple of seasons' in, say, the Premier League home matches is just 38 matches which is actually a very small sample to draw a conclusion from.spin a ball on a roulette table 38 times and see the resultsyou get, follow that by 380 spins and see the difference.

Yes Kenilworth
Thanks for that
Components of a Roulette wheel are identical for all 380 spins
And it doesn't take 20 years to get there
Your logic suggests we should be looking at every every Premier League season played in order to rate sides today
I'm sure that's not what you meant

No Hejik, just the last 3 seasons i'd say

Teams may have a season where they seem to be home or away 'specialists' but there would be no reason to expect that to continue ,as it would just be coincidental

For the Everton v Tottenham game I'd give Everton the very slight edge and price it up has 55/29/26, meaning Everton should be priced at 2.22 in my view making the current price at 2.06 available a value lay. Granted you could make arguments for Tottenham having the better squad of players but they are missing Bale and I honestly think outside the two Manchester clubs Everton have been the best side in the league this season. Incidentally I often find Everton to be a little short (well apart from last week against City where I did lay City) especially at games at Goodison, they are often seem to be around the 1/2 mark at home against your average sort of mid-table teams which I'll be happy to lay all day.

Given how I price things up there seems plenty of value around this weekend in the Premier League. Once again I think Liverpool are too short. I take the point about Liverpool missed chances this season but that's a direct result of a lack of goalscorers in their team bar Suarez, who will be a big miss in this game. I also think both Arsenal against West Brom and Swansea against Norwich make for good value lays this weekend.

Someone was asking where the 44/29/27 came from, well it came from partly looking at historical HW/D/AW percentage as well as my general knowledge of how games tend to be priced up when two equal teams play each other. But there's no saying this is definite or anything I just find it useful to have a starting point to work from and then adjust the percentage according to your opinions on each team.

What I mean is that small batches of results throw up
unusual sequences. For instance Blackburn Rovers have
drawn 7 from 10 away matches, while Wolves haven't
drawn yet playing the same amount of matches. Should
we be backing the former to draw away, and the latter
not to draw, for the rest of the season?
ManU haven't drawn this season but ManC have drawn 6
times from the same amount of matches, should we bet
accordingly? These sort of distortions are allover th
place. Small batches are meaningless, that's what I'm
saying.

Calculating draw odds confuse me as well to be honest. Interestingly enough in the Manchester derby the draw odds have been drifting out somewhat, perhaps taking into account the fact Man United haven't drawn this season?

I would say there is definitely a higher draw tendency for certain teams. Especially teams that set-up defensively and don't concede many goals. If for example the two Manchester clubs had looked much more secure in defense and conceded a lot less goals this season surely the draw wouldn't be has high as the current 3.65/3.7.

Statistically 44/29/27 is not far away for the Prem.  Since season 2000/2001, the actual percentages are 47/26/27.  They have varied very little season by season, the highest home win percentage being 51% (twice), the lowest being 43%   

This, however, does include every game, and you may want to remove the strongest teams from any statistical modelling as it skews the figures.  eg. Man United draw very few home games each season as they win a large number of them.

I agree with kenilworth that you don't want to get too hung up about certain teams being home'away/draw specialists.  As an example, West Brom lost 10 home games last season, more than any club outside the bottom 3, yet they are perceived as home specialists this season - 6 wins from 8.  Everton drew only 3 home games last season, but have drawn 4 out of 7 this term.  And so on....

Yes Kenilworth I agree
And it goes to my original point
You could possibly use it for some degree of reversion to work those kinks out
But there's no relevance to the sample if you have a worthwhile one as we have agreed above.
The fact that on average, results show 47% has no relevance to the Manchester Derby pricing for this weekend other than a small reversion that may be necessary to keep numbers round in amateur pricing models.
None of which will make any money since they are fundamentally wrong to begin with.
That's my point.


letsalldance.
thanks for getting back to me
yes I suspected you were looking at the average distribution of results or, average results distributions.
I just wouldn't place too much emphasis on them
Games chosen for this thread which suit those distributions have been chosen (maybe subconsciously) because they suit those distributions
It's coincidental on the Manc derby.
Neither side is an "average" side either, home or away and should not be priced as such
they are both very much above average both home and away
so how do you apply your average prices to that conundrum?
it's a flawed approach is all I was trying to say, has to be more involved

Dec 7, 2012 -- 1:59PM, tobermory wrote:

No Hejik, just the last 3 seasons i'd say Teams may have a season where they seem to be home or away 'specialists' but there would be no reason to expect that to continue ,as it would just be coincidental

Tobermory,

I think it may have been you who stated something I certainly would agree with in that it's less a case of "Home Advantage" and more a case of "Away Disadvantage".
That's a better way of looking at things

Apologies if I am mistaken and it wasn't you

vg post Tinkers

Hejik i believe there is home advantage but it generally favours all teams at that level equally, so that all premier league teams will, over time , get about 60% of their points at home and about 40% away.

In the short term you can find teams with 80% of pts at home and only 20% away, or even 45% at home and 55% away, but the longer the sequence you look at, the more it will align with that 60-40 ratio.

The only example that comes to mind of a team having a big discrepancy over several years was Luton in the 1980s , who consistently had one of the Top 5 Home records while being bottom 6 in the away table . But there was a specific reason for it : their plastic pitch, which either gave them a genuine advantage in their familiarity with it , or gave the away team a mental disadvantage as they didn't like playing on it.

Not sure that the Manchester derby is the right game to use as a template for working out what home advantage is worth.

I think that home advantage is not worth quite as much in a local derby as in a 'normal' match, not by a huge amount, but enough to make a derby a bad guide.

The draw price in any match is the hardest thing to price up as it's related to goals expectancy as well as the relative strengths of the two sides involved.  And there's still no exact science in it as far as I can tell.  Draw prices quite often move in excess of 5%, sometimes as much as 10%, in the hour up to kick off.  This has a serious knock on effect to the pricing of each side.

As a suggestion to letsalldance, in the short term, you could remove this problem by taking the draw out of the equation and playing in the draw no bet market.