Saturday, August 7, 2010

Thinking Outside the Box

It never occurred to them that, if everyone had to think outside the box, maybe it was the box that needed fixing” [Gladwell, What the Dog Saw]

When picking a fantasy team, like when trying to win anything, most participants will search for an ‘edge’ or secret formula to give them a jump-start over their opponents. This may take the form of relying on industry knowledge (“a friend of mine is a scout at Spurs and he assures me Assou Ekotto will retain his starting LB position from last season”), trend analysis (“Rooney scores 69% of his goals against bottom half teams”) or historical performance (“high price midfielders always out deliver their expensive forward counterparts”).  This quest for more information works on the assumption that the common ‘knowledge’ is at best limited and at worst incorrect.

How true is this though? In NFL fantasy we are told to draft running backs in the first round, in baseball anyone who can deliver speed and power is a sure fire early pick and the EPL game is no different in giving us several age old adages upon which we base our team. One of them that interested me is how reliable a measure of success is the number of playing minutes a player generates in the season. Many – including this blog – will often praise the virtues of the reliable player will who play 90 minutes each week and while not win you the league, will never lose one either.  Let’s take a look at the numbers:

The correlation between forward minutes and points is strong (0.772) which is in line with common sense – if you don’t play you will struggle to achieve the main aim of a forward: score goals. What is also worth noting is the strong correlation (0.726) between points scored and player value. 

 This suggests that bargains are hard to come by when it comes to strikers and only a couple of players can be described as anomalous outliers last year (Bent, Tevez and Rodallega all outperformed their value). When it comes to minutes played, the four best ‘overachievers’ are the big four of Drogba, Rooney, Tevez and Torres (to be expected given their individual and team talent).

You can summarize this position as ‘you get what you pay for’. It is going against recent history to expect to find a forward equivalent of a 2008/09 Stephen Ireland or a 2009/10 James Milner. Players like Amr Zaki 2008/09 have done it to a degree but failed to keep it up for a whole season, instead leaving you to try and capture lightning in a bottle.

In summary, the data from last year underlines the need to pay for big name strikers. A look over the message boards at Fantasy Football Scout suggests that many consider this to be an anomaly and are returning to the big midfield names, but until I see otherwise, I will be paying a premium for strikers who can be relied upon not just for minutes but also for goals.

The correlation between midfield points and minutes is much weaker than with forwards, and indeed at 0.535 we can conclude that there is only a weak linkage. 

Admittedly the sample size is quite a lot larger than forwards but still the number of both over and underachievers is noticeable. This is somewhat of an eye opener to me as I have often stuck with midfielders through baron patches on the basis that they are guaranteed to play – which I thought was a useful measure of overall success.  In fact, you are better chasing points a bit and trying to find the form player rather than stubbornly sticking with the player who just isn’t delivering the goods.

The linkage between points and value is once again relatively strong at 0.701 showing once again that you get what you pay for.  Plenty of outliers exist though and this line of best fit is somewhat skewed up by the outstanding returns delivered by Lampard and Fabregas.

 The data shows that while, of course, the top players will always usually deliver good returns, there is value to be had elsewhere. Also, relying merely on a player getting consistent minutes is not enough to guarantee good returns in the future.

The key lesson here is the lack of correlation between value and points – standing at just 0.559. 

Thinking about this for a second, this suggests that all those United, Chelsea and Liverpool players are overvalued and you should continue to pursue players from team such as Fulham and Villa last season. This makes sense due to there being 2 ways to ‘beat the market’ when it comes to defenders. Firstly, a team’s players are undervalued, like with Birmingham last season, whose players could mostly be picked up for 4.5m or less. The second is a player on a premium team who is valued significantly below his teammates but still gets the same access to clean sheets. These are hard to find right away but much success has been enjoyed in previous seasons with players like Johny Evans, Armand Traore, Martin Skrtel and Roy Caroll who have all gained significant runs in top teams despite costing several million less than their teammates.

Now, be careful here, as we are not saying the top defenders to not deliver at all. Indeed, of the top 10 leaders in points only Jody Craddock who scored an unlikely 5 goals was available at a low price. What we are saying though is that, comparatively speaking, there is better value to be found lower down the price lists than with forwards and midfielders.

Not much to say about value against points as there is not a strong correlation (0.619) and the best points delivered by 4m and 4.5m players comes close to that delivered by the most expensive players. 

I emphasized last season that you should not pay for a goalkeeper – instead relying on a rotation system – but even if you simply played someone like Sorensen or Hart every week you were still rewarded with much better value and only 20 points less over the whole season.

As you would expect the correlation between minutes and points is extremely strong but there isn’t much fantasy advice to gain here other than make sure your ‘keepers are playing!

I am sticking with the rotation strategy for this season, and I will post shortly who I think complement each other well to avoid having to play unfavourable matchups as often as possible. 

To conclude then, it seems that the old adage of getting 'reliable minutes' is not always a useful measure of fantasy success and we should look for other means of determining value. While it seems obvious to say that price dictates performance, we can see that this holds truer for forwards than for other positions and so it seems wise to once again spend money up front, so long as you feel the EPL forwards will enjoy another good season.


Ryan said...

Another briliant article Chris! You've confirmed my suspicions about the optimal way to allocate our $100m. Took me 2 fantasy seasons to figure out the cheap gk strategy. Still torn between having at least 1 Big 4 defender (eg. Chealsea and/or United) or just employing 5 cheap-to-mid-range ($5.5 and below) defenders who occasionally attack. Looking forward your "Scanning the radar: Defenders" write-up for some enlightenment.

travelnotes said...

Don't forget, the value data uses prices of players that may have gone up over the course of the season (due to gaining points and transfers in) or down (due to not playing and being transfered out).

To think outside of the box we need to find the cheap players whose value will go up (because they're doing well and being bought by other managers) and keep away from the players whose value will go down.

Playing the market also increases our purchasing power as the season goes on.


Anonymous said...

在你一無所有的時候 是誰在陪伴你 他便是你最重要的人......................................................................

Chris Glover said...

Thanks Ryan. I will try and get my underrated defenders up tomorrow. I am looking at similar strategies to you so it seems we're on the same page.

travelnotes - good point on player values and that is definately a drawback of the analysis. I think the point stands that bargain mids are easier to find though. I will post either just before or just after the week 1 games the players who I think should be targeted to make money.

thanks for reading and posting.