At this time of the season, managers start thinking about end of season rank targets and we see community posts on this topic. Often, these are above their current rank, sometimes well above.
“Currently at 44k, looking to push into the top 10k…”
Getting a green arrow
According to LiveFPL.net, Gameweek 30 had the following averages:
- Top 10k: 61.25 points
- Overall: 47.25 points
- Elite: 60.88 points
These are fairly typical values. There was no inversion against the template, yet a few differentials came good. It was a good week to examine.
LiveFPL shows a ‘safety score’ for each team. This is the score required to achieve a green arrow. You can see this for any team by entering its FPL ID number. Here is a selection of pre-Gameweek 30 ranks and the corresponding safety scores.
| Pre Gameweek 30 rank | Safety score |
| 8 | 65 |
| 103 | 65 |
| 175 | 64 |
| 329 | 65 |
| 1,958 | 63 |
| 2,844 | 64 |
| 8,672 | 64 |
| 11,758 | 63 |
| 20,585 | 62 |
| 135,387 | 62 |
| 142,599 | 61 |
| 299,528 | 60 |
| 563,742 | 59 |
| 666,624 | 58 |
| 1,006,299 | 57 |
There was a steady decline in the safety score as the rank increases, although it isn’t completely linear. With the top 10k average being 61.25 after hits, you could be forgiven for thinking that scoring 62 would see you hold rank. However, that was not the case. Top 10k managers needed to beat this average to hold rank.
The reason for this is that top performing managers have a chasing herd of managers behind them. Every Gameweek, some of this herd will score particularly well. This results in churn in the upper echelons of rank. In Gameweek 30, it was managers owning Alexander Isak (£7.9m) that scored above 70 and made the biggest gains.
Needing to beat the top 10k average doesn’t stop there though. A manager ranked 20,585 required a score of 62 to hold rank and that safety score extended all the way down to managers ranked around 135k. I checked this at regular rank intervals all the way from 20k to 135k.
Here is the same exercise on Gameweek 31. It is good to check ideas against more than one data set.
Gameweek 31 averages
- Top 10k: 60.04 points
- Overall: 46.22 points
- Elite: 61.45 points
This was another week with an average top 10k score near the fairly typical value of 60.
| Pre Gameweek 31 rank | Safety score |
| 6 | 69 |
| 7 | 68 |
| 17 | 65 |
| 83 | 66 |
| 1,385 | 65 |
| 20,422 | 64 |
| 30,383 | 63 |
| 40,027 | 62 |
| 142,933 | 63 |
| 272,869 | 62 |
| 945,107 | 61 |
| 1,072,030 | 60 |
| 1,782,826 | 59 |
| 1,842,067 | 58 |
| 2,414,120 | 57 |
This time, the safety score fell below the top 10k average at a much lower pre-Gameweek rank than in Gameweek 30. It was around the one million mark! To attribute this solely to the top 10k template captain Mohamed Salah (£13.4m) failing is over simplifying it. None of the player base did well on captaincy.
Taking the two Gameweeks together, the main takeaway is: You must beat the top 10k average to maintain rank, even at much lower ranks!
How many points must you gain?
At the bottom of the Live FPL rank page is a table of points totals at selected ranks. Here is the position after Gameweek 31.
| Live Rank | Points | Points behind leader | Points per week |
| 1 | 2,110 | ||
| 100 | 2,032 | 78 | 11.1 |
| 1k | 1,992 | 118 | 16.9 |
| 5k | 1,962 | 148 | 21.1 |
| 10k | 1,947 | 163 | 23.3 |
| Live Rank | Points | Points behind 10k | Points per week |
| 10k | 1,947 | ||
| 25k | 1,926 | 21 | 3 |
| 50k | 1,907 | 40 | 5.7 |
| 100k | 1,884 | 63 | 9 |
| 200k | 1,857 | 90 | 12.9 |
| 300k | 1,838 | 109 | 15.6 |
| 400k | 1,822 | 125 | 17.9 |
| 500k | 1,810 | 137 | 19.6 |
There is a wider points spread within the top 10k than from rank 10k to 500k. The upper echelons are like the right-hand end of a normal distribution: a very thin tail of a few managers with extremely high scores. This is like a wealth distribution, with a minority of multi-millionaires and billionaires.
To win FPL, a manager ranked 1k must make up 118 points in the remaining seven Gameweeks (16.9 points per Gameweek) and hope that nobody above them gets close to achieving this. Whilst these tables don’t account for differences in chips remaining, that is still extremely unlikely. That gap increased from 112 points at the end of Gameweek 30.
Managers ranked 100 to 1k not only have at least 78 points to gain on the leader, they have a huge herd of managers behind them, some of whom will outperform. Maintaining rank at this level is incredibly difficult because the distribution starts to fatten out: there are 4,000 managers within 30 points of managers sitting at 1k. For managers sitting at 5k, there are 5,000 managers sitting within 15 points of them, as the distribution starts to widen.
To finish in the top 10k, a manager at 50k must make up 40 points (5.7 points per Gameweek) and hope that not too many managers above them achieve likewise.
Modelling the probability of this is tricky. Complications include:
- How widely spread points are across different Gameweek ranks.
- How the double Gameweeks affect the points spread (you would assume a wider points spread but by how much?)
- Whether trends in player or team form emerge (dependent events) or whether there is inconsistency (more like a series of independent events).
My conclusion is that eroding this 40-point gap steadily over the remaining Gameweeks is very unlikely. To simplify it greatly, having seven consecutive Gameweeks of correctly calling a coin flip (a series of independent events) has a 0.78% chance. A coin flip analogy has some validity because your Gameweek ranks must be within the top half of the number of still engaged managers every week to make up those 40 points in a steady fashion. I will spare you those details but if anything, that uses an under-estimate of the number of engaged managers.
Performances by midtable teams and inconsistent players can be seen as something like a series of independent events, so somewhat suited to being modelled on a consecutive coin flip basis. In any case, where players or teams find a run of form, you would need other managers to not jump on those bandwagons, or you would need new advantages elsewhere (new independent events).
More likely then, a few knock-out weeks with differentials coming good is required. I think many managers intuitively realise this but the above (greatly simplified) steady erosion analysis confirms it. Add in the first point regarding the need to outperform to maintain rank and you’ve got a case for managing your expectations and going easy on yourself.
