Saturday, August 9, 2014

Fantasy Football - 4: The Value of Money

Lab Record Type Section


This post is part 4 of a series of number crunching posts I intend on doing on some EPL Fantasy Football data I have. I focus on the value of your money in this one.

Part 1 has the main introduction and some other relevant details and assumptions and things like that. It also has the Dream Team of the Year and Weekly Dream Teams which I shall be referring to frequently.

Relevant Summary of Earlier Parts

I have ignored weekly transfers and wildcards for the most part. This post shall attempt to do justice.


Disclaimer 1:
The guys at the Fantasy Premier League have some formula to determine price rise, and fall. Nobody really know their logic. However, there are quite a few sites that try to use the numbers to reverse engineer what the formula might be and these sites are helpful sometimes if you're looking to make (or avoid losing) a quick buck.

The disclaimer is that in this post the logic is simplified: transfers in/out proportional to price rise/fall. Nothing else matters.

Disclaimer 2:
When you sell a player, you usually get only half the profit. This half-profit is again tricky business and is ignored for the purpose of this post.

The Rise and Fall of Valuations

The chart below classifies players into two groups, one that experienced a price rise in that gameweek, and one that experienced a price fall in that gameweek, and it reports the percent of players which experience a change in price, and the magnitude of the change in price across the gameweeks.

Concentrate on the forwards and midfielders for the moment. There's a fair bit of price discovery going on in the first 12 weeks or so with a sharp surge during weeks 3,4, and 5, just after the transfer window closes. By the looks of it, a fairly diverse group of attackers are ditched by various FPL managers around that time, which results in a bit of a price fall for a high percentage of players, and the income is then invested into a much smaller group, resulting in a high price rise for a small percentage of players. If you can preempt this, which is not impossible since the FPL folks keep updating the figures on the number of transfers in and out for each player, you can make a nice bit of profit.

There are patches of high activity during the rest of the season, probably stemming from events like Aguero returning, rereturning, rerereturning from injury, Ramsey getting injured again, Adam Johnson playing like a man possessed, double gameweeks, etc. but these are probably not systematic.

Towards the end of the season, around week 34, is when heavy transfer activity starts again in what are probably last ditch attempts to rescue the season. It is ridiculously easy to get transfers wrong and I won't be surprised if this is of no use at all. More on this in part 5.

The value of defenders and goalkeepers on a whole seem to be a more stable bunch as compared to their colleagues from the other half of the pitch. You can find winners here too, like Coleman, or losers, like Ashley Cole, but that needs transfer smarts and like I mentioned, it's very easy to get transfers wrong. Since most of the rise in valuation is going to come from attacking players you can probably expect better returns from heavy investments in those chaps rather than the defense. You might be one of those who scoff at the importance of the money but you will find out later in the post that it is.

Between weeks 10 and 23, if you look hard, you can also see a left to right staircase kind of a recurring pattern in the percent of players with price fall. It stops around week 23 or 24. This probably indicates how the managers rotate their one free transfer per week across the field, but more importantly it hints at the fact that the winter transfer window, like the summer transfer window, also disrupts the regular transfer patterns, and has a lot of attacking players ditched with the investment pouring into a small bunch. The lack of intensity would probably stem from the fact that around this time, managers already know how everyone else who has been playing already is doing. It is only the new arrivals that are the unknowns.

The smell of wildcards emanates from the transfer activity around transfer windows. More on that in part 5.

Riding the Rising Prices

Based on last year's scores and the prices at the end of the season, within a budget of 82, this would be the Overall Dream Team.

Name Position Team Total Points Price
Julian Speroni Goalkeeper CRY 144 4.80
Jose Fonte Defender SOU 147 5.50
Laurent Koscielny Defender ARS 149 5.50
Per Mertesacker Defender ARS 157 6.20
John Terry Defender CHE 172 6.80
Seamus Coleman Defender EVE 174 6.80
Adam Lallana Midfielder SOU 176 7.90
Gnegneri Yaya Touré Midfielder MCI 235 10.30
Jay Rodriguez Forward SOU 152 6.10
Olivier Giroud Forward ARS 182 8.50
Luis Suárez Forward LIV 293 13.20

The total points would be 1981. With Suarez's captaincy would rise to 2274. That's a difference of 129 points from the Dream Team we picked based on prices at the start of the year. Note how it's a 5-2-3 as compared to the Dream Team's 3-4-3, again indicating the higher rise in value of the high returning midfielders and forwards during the season.

To get a proxy for the points value of being on the right side of the price rise, I took the player prices at the end of the season and found the optimal team for various budgets. While not the best logic, if you assume that prices rise slowly over time and one would probably be making transfers during these weeks and would need the extra money, you should be okay with this analysis setting a lower limit -

It's clearly a linear relationship with each extra pound allowing an extra 10-12 points over the season. While not much in itself, if you miss multiple opportunities and lose 3-4 pounds maybe, then you could lose 40-50 points by the end of the season which is quite a bit. The Dream Team's points return without captaincy was 2110 and that is matched at about 92.5m pounds at the end of the season here which is an extra 10.5m pounds from the starting point of 82m. The Dream Team itself cost 92.9m by the end of the season with the midfield being the highest source of price rise.

In fact, I went a step ahead and calculated the different maximum points tallies you could achieve depending on what your available budget is, but taking the plaayer prices from each of the weeks.

The diagonal-ish pattern casually nods towards my loose assumption of a slow linear increase in the value of the players, although the figure on the whole suggests that most of the damage is done in the early weeks itself. With prices as of gameweek 1, the difference between a budget 82 team which scores 2110, and a budget 91 team which scores 2143, the difference is a mere 33 points. This difference rises to 57 points by gameweek 10 prices (2137 - 2080), 89 points at gameweek 18 prices (2123 - 2034) and kind of saturates somewhere around that point because at gameweek 37 prices, the difference is 94 points (2087 - 1981). Note how the change of the tally across weeks is more drastic at the lower budget ranges than the upper ones indicating how the higher performing more expensive players can contribute far more than some price adjusted returns from cheaper players. Just like in the real world, the 99% become poorer and poorer over time at the expense of the 1%.

A slight logical jump, but if you twist these words then you could also interpret this as the -4 penalty for an extra transfer being worth it for any that lies between
1. Where the new player either makes more than an extra 4 points for you
2. If the net profit from change in value of the player being subbed out and the change in that of the player being subbed in is at least 0.3m.

This presents an interesting dilemma. You need to maximise your points, but you also need to ensure that your monetary value is high enough to afford a high tally in the remaining gameweeks. In the ideal world these two would be highly correlated, but they are not. Prices are not related to performance, a price rise is a result of a lot of people investing money in certain players expecting them to score highly over the next few fixtures. More on this part 5. And also, lest you forget, it's very easy to get transfers wrong.

Bonus Section: How Easy it is to Get Transfers Wrong

As a crude starting points I drew two charts of the points that players returned over the next 5 games if they either had net transfers in of greater than 60,000 (which is about 20% of the managers in the world), or had average at least 4 points over the 3 previous games, and compared them against the sort of points the typical player returns when he played over 60 minutes in a game (inclusive of the two sets I mentioned).

The chosen ones return pretty much the same sort of points as the typical player playing over 60 minutes and two signs that I thought would be highly relevant are more or less useless. When to transfer a player in (or out) is probably a question that needs more sophisticated analysis and better data than I have or is deserving of a blog post. I miiiight have a go at it after I'm finished with the basic series. We'll see.


Money matters. Beyond the baseline of 82, each extra pound of profit earned by the end of the season can add another 10-12 points to the score.

Even after the price discovery is done, the well-performing more expensive players are worth more than the extra money you need to invest to get them so get them if you can.

Prices move more for the attacking half and there's some money to be made by trading wisely towards the first 5-10 weeks. But it is also very easy to get transfers wrong.

Other Posts in the Series

Part 1: General Team Composition
Part 2: Picking the Goalkeeper
Post 3: Picking the Defensive Line
> Part 4: The Value of Money
Part 5: The Value of Wildcards
Part 6:  Getting Ready for 2014-15
Part 7: How Much Does the Opposition Matter?

No comments:

Post a Comment