The BPS conundrum: abolish, replace, or modify?

As I mentioned yesterday in my 5 FPL Wishes for Next Season, I think a major revamp of the bonus points allocation in our game, and more particularly of the 'Bonus Points System' (BPS) rating scale currently used to achieve this, is needed urgently. [On top of everything else, it's very annoying that the 'ratings units' used in the BPS itself are also caled 'points'. It's really clunky to have to speak of 'Bonus Points' points. And it also causes confusion sometimes as to what people mean by 'bonus points' - the extra points actually awarded to players, or the BPS scores that determine those. I have fallen into the mental habit of usually referring to BPS 'points' as 'credits' instead; I wish FPL would follow suit.]

The sources of dissatisfaction with the current system are threefold:  a) It lacks transparency (too little information is shared about how the BPS totals are calculated);  b) It has been much abused by the FPL Gnomes this season (often the BPS scores have been adjusted after the event - apparently to produce less contentious outcomes in a few instances);  and c) Its results often appear unfair, inconsistent, and contrary to common sense (players who've had very good games - often, indeed, been generally acknowledged as the 'Man of the Match' - sometimes get strangely overlooked for FPL bonus points,... or at any rate given only a token 1 or 2 extra points, rather than the 3 points they seem to deserve).

Many long-standing FPL managers are now so disaffected with the BPS that they're grumbling it might be better to do away with it altogether.

I can sympathise with that view. Like the bonus chips (which I've grumped about elsewhere), they seem superfluous to the basic gameplay - merely an additional randomizing factor that tends to make the game even more unpredictable and less meritocratic.

At least, with the current BPS the award of bonus points is reasonably predictable for certain players over a long run of games; so, from that point of view, it could be considered 'fair', as it's not too difficult to take account of when making selections. But over a short run of games, or in a single Gameweek, it can be vexingly opaque, capricious, random. 

And it can potentially have a very big impact. While my weekly returns of bonus points don't seem to have a huge variance (almost never any less than 3 or 4, but rarely much more than 12, and mostly around 8 or 10), over a season I can easily stray 100 points either side of my 300-point median; and that's pretty much all - sometimes more than... - my typical season-to-season points total fluctuation. Bonus points and BPS might actually decide the whole shebang!

So, the bonus points are a big deal. And, at the moment, the way they are distributed is causing a lot of resentment in the FPL community.

However, I have a sentimental regard for tradition. And 'bonus points' - in pretty much the form they are now, I believe - have been around ever since FPL's inception in 2002. So, I'd be loathe to give them up completely, after being so long a core part of the game. (Apart from anything else, that would make it extremely difficult to make any meaningful comparisons between present and historical data in the game. This is one of the many gripes I have against this season's absurd novelty, the 'Assistant Manager' Chip: many people earned 30-50 points from it - more than you typically get from the other two bonus chips combined; a lucky few got even more from it; and it could conceivably have yielded 80+, maybe even close to 100 points. That is a really huge - and distorting - addition to the game's points potential for the season.)

Other critics favour replacing the current BPS with a simpler - hopefully fairer - means of deciding the weekly bonus points allocations. There are indeed a number of stats-compiling companies who offer ready-made player ratings (the current BPS is based on stats licensed from Opta; although, curiously, assists and own goals are adjudicated with the assistance of Stats Perform instead). And the Premier League itself is now making the official 'Man of the Match' awards 'democratic' by inviting fan votes through social media (though this is quite new, and hasn't been that well publicised as yet; I don't know what kind of numbers are participating).

While a ready-made player rating system could give more satisfactory results (if you pick the right one!) than the current one (which goes through the clunky additional step of filtering third-party data through a weighting template of FPL's own devising - I think that's where the problem really lies), there would be bound to be considerable teething troubles with any new rating system applied in the game. And I doubt if a new ratings provider would be immune to my misgivings about 'transparency' - since all of these stats companies seek to keep an awful lot of their process secret. 

Using these new online 'Man of the Match' polls is more immediately tempting to me as an alternative. But the problem with leaving the rating process to the subjective judgement of individuals (even very knowledgeable football professionals, as used to happen with the EPL 'MotM' awards in the past; or very large numbers of people, as we now have) is that there's a risk of the results being skewed by personal biases - especially, now, the loyalty of large fan groups. You've noticed how the BBC's 'Goal of the Month' competition, also decided by a fan vote, is very rarely won by a player from a less fashionable club (and indeed, even among the most popular clubs, a player is far more likely to win the accolade if his club was playing on the day the vote was held, especially if that was one of the 'games of the day' featured early in the show)? Heck, with a mass-participation game like FPL, there's a serious danger that groups of Fantasy managers would organise 'Man of the Match' voting in favour of the most popular captaincy picks for the gameweek. I do quite like the idea that all the popular votes could be tallied to identify the handful of most impressive players in each match in a rank order, to decide the award of FPL bonus points; but in practice, I think there would be too much scope for 'manipulation' of the results.

Another option sometimes suggested is to replace the current bonus points with new categories of points awards for specific game actions. We've seen something of this in Fantasy games for the big international tournaments: the last Fantasy World Cup introduced additional points for a certain number of 'ball recoveries'; actions like tackles or duels won, 'key passes', and 'big chances created' might be other possibilities for inclusion in such a revised scoring scheme. While I quite enjoy having to adapt to such novel wrinkles in a once-every-four-years Fantasy tournament, I feel it would be too much of an upheaval in our well-established annual competition of FPL, Again, it would produce much higher potential points scores for each gameweek, and across the whole season, rendering all earlier seasons incompatible for FPL performance comparison.

So, reluctant though I am to admit this as the only viable solution, I feel that we probably have to make do with the current bonus points format - decided by the dreaded BPS.

How, then, might we address the three areas of difficulty I outlined at the start of this piece?

a)  Transparency

Probably a lot of the problem here arises from the fact that Opta, the provider of the underlying game statistics used to tally the BPS player scores, is reluctant to share much of its data - or almost anything of the process it uses to compile that data. (All other stats compilers are much the same in this, I would imagine.)  Partly, they want to steer people towards premium subscriptions for richer data; partly they want to protect their IP, to prevent upstart businesses from too easily copying what they do; and also, probably, they don't want it to be too easy for people to check up on their accuracy and consistency by attempting to replicate their stat-compiling process, even over a small sample size.

However, this could be an area where sticking with the incumbent data-provider (rather than instead buying an off-the-shelf 'player rating' stat from a rival company) will give FPL some useful leverage: they ought to be a powerful enough client that they can persuade Opta to allow the release of more data than they might ideally like to. What I'd like to see is the full background stats BPS is supposedly based on - for every player. But if Opta is digging in its heels against that, I'd probably settle for being able to see the detailed breakdown for the 'Top Ten' BPS scorers usually listed for each match; or even just for those few players who ultimately receive bonus points. And there surely shouldn't be any problem about FPL publishing the BPS total for every player??  (In an ideal world, I'd also like to see detailed explanations of how each of the relevant game actions is defined, and at least some explanatory examples - each week! - of how potentially contentious incidents have been classified, and why. But let's work towards that slowly, eh? We probably can't get everything we want, all at once....)

b)  Surreptitious adjustment of BPS scores (after matches are over)

That seems to rest with FPL rather than Opta. But either way, it shouldn't happen - not without an open acknowledgement, and an apologetic statement explaining what happened. Most of the BPS data is updated almost live - so you can actually check on who's in the running for bonus points while games are still in play. You must expect that some things might get tweaked up to an hour or two after the game ends. But this season we've seen quite major points adjustments occurring a day or two later; if that occurs, we need to be told why.

c)  Appropriate Results

The main thing we want to see is the bonus points for each game more consistently, accurately, and predictably reflecting the commonsense assessment of player performances. 

And I think this could be achieved just with some tweaking of the current BPS scoring. 

The key problem with it is that it massively over-rewards certain game actions, while under-rewarding and even excessively penalising others - with the net effect that the bonus points tend to go mostly to the players who are already earning FPL points in the game: those who've made an attacking contribution, or defensive players who've managed to keep a clean sheet. It's a classic case of double recovery, and that is fundamentally unfair - particularly as a lot of significant game actions don't get any recognition in the main FPL points system, ('Pre-assists' are my particular pet peeve: the pass before the actual assist is very often the one that actually makes the goal; yet it earns no recognition, in either direct points or BPS credit.)  Midfield playmakers who quietly dictate the tempo of the whole game, or 'engine-room' lynchpins who break up every opposition attempt to progress the ball through central areas... are the kinds of players almost invariably overlooked by BPS - although proper football fans recognise them as the true 'Man of the Match'. And just last week (Gameweek 37) we saw an instance - sadly, not at all an unncommon one - where some exceptional goalkeeping performances from the likes of Sels, Leno and Kinsky went unrewarded. That's what we need to change.

Other game contributions get only negligible BPS recognition at present. A defender blocking a shot may be as important as a goal - but he only gets a tiny fraction of the BPS credit (for two of them!) that an attacker does for scoring a goal - essentially nothing. Defenders have never fared all that well under the BPS (unless the match is very low-scoring, they're bound to be eclipsed by all the players who contributed to the goals); and I pointed out early on this season that a small change in the BPS scoring would make it even harder for them to win bonus points this year (defenders and keepers are now more heavily penalised under BPS for conceding a goal, which makes it extremely difficult for them to get into bonus point contention if they fail to keep a clean sheet).

I'd be tempted, in fact, not to give any BPS credit for actions that are already credited in the main points system. However, that might lead to eccentric outcomes where a multiple goalscorer was overlooked for bonus points - which would also seem unfair under any commonsense view of things. So, I think we'd have to keep BPS credit for goals, assists, clean sheets, etc., but massively reduce it from the levels it's at now; while increasing the range and value of other game actions credited in BPS.

I don't see why scoring a goal should have such a massive weighting in BPS, or why it should differ for different players, different goals. There's a case for giving defenders (and keepers!) more game points for a goal, because it's so much rarer an occurrence for them (and they are under-rewarded by the overall points system, compared to attacking players). But as part of the overall 'game contribution' assessed under BPS, one goal is surely the same as another. 

I've always found it particularly baffling and exasperating that BPS awards an additional 3 'credits' for the 'goal that wins a match' - but offers no definition for that. Is it the last goal scored? Or (more probably) the last goal that moves a side into the lead? Either way, it's nonsense; it's really a matter of chance which player may get to contribute the 'most important' goal. And in any case, the truly decisive goal is the one that turns the momentum of the game - often, not one that establishes or extends a lead, but one that ties the score again, or even one that gets a side back in the match after falling a long way behind. I can see no reasonable argument for giving extra credit only to one of the game's later goals.

I'm doubtful about rewarding clean sheets under BPS as well; certainly not with a massive 12 'credits', as is currently the case. A goal can come out of nothing, out of pure fluke (or a bad refereeing decision...). It's probably undesirable that defenders and keepers get such a huge lift from a clean sheet under the main points system (but they need it, because they get no points for anything else, and usually only a fairly remote chance of big bonus points); and again, 'double recovery' seems inappropriate to me - if they're getting 4 points for the clean sheet already, they don't need a huge BPS boost too. For me, there's not usually any difference in quality of defensive performance between a team that concedes 0 goals and a team that concedes 1 (or 2, sometimes....).

Weighting the BPS so heavily in favour of game actions that are already rewarded is plainly wrong-headed and unfair. If those 'credit' items are still to stand, they need to be massively dialled down, I would say. Whereas, credits for other important game actions like 'key passes', 'tackles won', and 'fouls won' should be significantly increased. That could produce BPS results that accord more closely with actual player performances.

It will be a complex task to get this rebalancing of the 'Bonus Points System' right, but - I believe it can be done.