I’ll explain everything to the geeks: DMs and BPS in FPL

In a recent podcast, we recently went over the bonus points system and how it could be improved.

Our thoughts on the matter boil down to one simple word: transparency.

But this article isn’t about that.

Instead, this is more about something I’ve been thinking of as a result of looking at the system, and how it could be changed to give us some new experiences in FPL.

One way of doing it I think it could potentially transform our view of one group of currently defunct players in FPL (defensive midfielders) and add a new dimension for us when picking our teams if the bonus points system was retooled.

n.b. this article is of course speculative in the extreme and one where I “explain everything to the geeks” and do a bit of ‘theorycrafting’.  Enjoy!

The problem

As we know, FPL is about picking the players who score the most points. By extension, this means that there are inevitably players who are not, and almost never will be, owned by FPL managers.

Data snapshots (using csv files from FPL Overlord) reveal that this lack of investment from FPL managers hits one section of the player base hardest: midfielders.

Above, I’ve just done a basic analysis using a large spreadsheet of all players in the game (correct at immediately after the deadline on 3/3/2018), coding up players based on their ownership. I’ve not excluded injured players in the data above as it was worth getting total level data for evaluation.

As you can see from the bit I’ve highlighted, the area that is under-serviced currently is midfield, with seven in ten players featuring in under 1% of teams. This is significantly more (at a 95% confidence level) than the other positions, and points to a potentially unforeseen impact of the current set up of FPL: lots of players are not featuring at all in manager’s thinking.

This has the knock-on effect of narrowing the pool of players that are in the game: just under half of the players from the other positions are active enough to have at least 1% ownership.

To put this another way, if we exclude midfielders 44% of the available players are at least in 1% of sides. But bringing midfielders into the data set takes that number down to just 38%, which means that 62% players are currently not in more than 1% of sides.

The BPS system doesn’t help

For this section, I’ve asked for help from our man Prokoptas to crunch the numbers on BPS to see if the system does indeed disfavour defensive midfielders. Guess what? It does. Here, he explains why…

The role of the BPS scoring system in disproportionately narrowing the player pool at the midfield position can be observed in a number of ways.  Most simply, when analysing data from the current season (in which all players with at least 100 minutes of gametime are considered), we can observe a lower degree of correlation between BPS and points for midfielders than any other position (R2=0.88; by comparison, the coefficients of correlation at other positions are 0.97 (GK), 0.92 (DEF), and 0.96 (FWD)).

This suggests that BPS does a worse job of translating on-field actions into fantasy points for the midfield position than it does elsewhere.

Indeed, analysis of the residuals suggests that defensive midfielders earn too few points for their actions during high-activity performances (which feature a high number of tackles, successful passes, etc.) for these to frequently translate into actual fantasy points through the BPS scoring system.

Bonus points are also less equitably distributed within midfielders, particularly towards the middle and lower echelons of the position. Again analysing all players with at least 100 minutes played, we can see that the percentage of midfielders who earn bonus points is lower than at any other position for each of our defined thresholds. 38% of qualifying midfielders have failed to earn a single bonus point so far this year – a far higher proportion than at other positions (GK=10%, DEF=30%, FWD=20%).

Further, nearly two-thirds of all playing midfielders (65%) have earned 3 or fewer bonus points so far this season – compared to just 38% of Goalkeepers, 44% of forwards and 59% of defenders. As a result, a Pearson’s chi-squared test of independence shows that the distribution of bonus points within midfielders differs significantly from other positions (p<0.01).

(Thanks Ed!)

A solution to the issue

At the moment, as we’ve mentioned a lot recently, the bonus points system works by allocating points to players based on their in-game actions.

These OPTA-recorded actions fall into four broad key categories (that I’ve made up!) – Gametime, Attacking, Defending and Passing.

Below, I’ve broken down the list (roughly ordered on the actual site) to fit these key categories, and provided highlights (orange = offensive midfielders / blue = defensive midfielders) for what types of midfielders currently benefit in the BPS.

By focusing in on the positive actions, we see that there are many ways that offensive midfielders are currently favoured: there are seven factors that an offensive midfielder is likely to benefit from in terms of accruing BPS compared to just three for a defensive one.

In practice, this means that defensive midfielders are close to non-entities in FPL (outside of being a cheap enabler or perhaps an Etienne Capoue on a blistering yet greatly overperforming 2-3 Gameweek period).

It doesn’t have to be this way though.

What if the BPS system was slightly adjusted to bring DMs into the game? By that I mean introducing “defensive” BPS-worthy actions which are midfielder-only:

  • A successful tackle for a midfielder could be rewarded with BPS rather than as a net value at the end of the game
  • Clearances, blocks and interceptions (CBI) could be strung out to a category each instead of being lumped together.
  • Recoveries – a vital part of the job of a defensively-minded midfielder (who doesn’t play for Arsenal) – are worth 1 BPS, could this be re-weighted?
  • Clean sheets should also be of benefit to midfielders in the BPS – the more defensive ones may have done just as much to ensure it as the goalkeepers and defence.

What if these were given a new weighting to bring DMs into the game?

N’golo Kante, for instance, is averaging 9 recoveries, 2 interceptions and 1 clearance per game this season. What if he began to be rewarded for each of these actions in the same way creating a big chance was, or somewhere in the middle?

If these defensive actions were given 2bps each instead of 1 – as a simple example of a tweak that could be made – he could be sitting on 24BPS per game for that average. If we add on appearance time, passing and perhaps a new BPS factor for clean sheets, then we start to see him maybe enter the running for bonus points at 30 BPS+ on a routine basis.

Pricing would remain near-enough the same, with managers forced to choose between more explosive yet less consistent options at higher prices versus more consistent yet less explosive options at lower prices. Hopefully this would mean the game wouldn’t be broken by such a change.

Consequences

I hope you can immediately see what this idea could do if defensive midfielders suddenly were in contention for bonus every game: imagine an asset that offers you guaranteed ~4 points per week (especially for a defensive midfielder like Kante who keeps a clean sheet on a regular basis). It would certainly introduce an interesting new angle for playing FPL.

If upweighting BPS potential for Defensive Midfielders, some of the consequences could be:

  • More variety in teams. As the ownership data right at the top showed, 62% of players in the game at the moment are selected by under 1% of managers, with more than half of them outright defunct. Bringing a few of these players in from the cold could drastically widen the pool of players worth the consideration of FPL managers and lead to greater choice, creating more variety between our squads.
  • Defensive midfielders enter the template. We would see there being a whole new set of decisions and strategies opening up with the new pool of players available to us. Managers might be seen electing to run with two or more defensive midfielders for a cheaper price if they are a near guarantor of points.
  • The 3rd, 4th and maybe 5th midfield positions get interesting. Managers would have trade-off the between the drip feed of points from defensive midfielders against the more explosive (if less reliable offence options). Outside of the Salah-type auto-includes, we may see greater divergence in strategies, with some managers perhaps electing for steady points in the 3rd, 4th and maybe even 5th midfield positions to prioritise cash elsewhere.
  • Shaking the game up. FPL hasn’t been solved or anything, nor have we really settled on which formation is the best (I suspect this varies with the meta) but it might be good to add a new angle to the core game. We’ve had the chips recently, and then before that vice captain choices, which have been more cosmetic than something that really impacts the mechanisms behind the game – surely that’s due some development now as the modern game has changed?

All in all, though I suspect of course it won’t happen, the suggestion above kills two birds with one stone: it would reinvigorate the game by bringing a group of players into our thinking, as well as tweaking the bonus system to reflect the integral role of the defensive midfielder in modern football.

If you take away anything: only 38% of players are in more than 1% of teams. Re-tooling the BPS system could supercharge the prospects of Defensive Midfielders and also give us new decisions to make in FPL: it can get a bit stale (as we’ve realised recently) if everyone is running highly similar sides.

Appendix: examples from other games

It’s worth noting that a couple of comparators that have a different approach than FPL to bonus:

Sky has a multi-tier, simplified bonus system where players get points for more basic in-match accomplishments: saves made, tackles won, complete passes and shots on target. The more of these achieved by a player (after a minimum threshold is reached) dictates whether they fall into tier 1 or tier 2 of bonus

MLS Fantasy as bonus for defensive or attacking plays. Actions such as a key pass or chance created earn bonus, as do points for defensive ones. There are 6 categories for defensive bonus in MLS, catering for the number of players in defensive midfield, defence and GK positions.

The key features of both systems are that they are simple and position-neutral. In both systems, bonus seems to be given without constraint (e.g. no 3-2-1). If OFPL didn’t want to do something complex and instead preferred to simplify the system so that entry-level and engaged players alike understood the system with perfect clarity, this could also be a way to go.

Thanks to Ed for the stats analysis and Con and Dave for proof reading.