fantasy premier league tools – Joe’s transfer planner’s team season review 2019/20

Contributors, Fantasy Premier League

Here’s our fantasy premier league tools article where we take a season review of the FPL team that was run by Joseph Crilley’s transfer planner and points projection tool and finished with a 32k overall rank.

fantasy premier league tools – Joe’s transfer planner’s team season review 2019/20

A reminder of the additional Patreon content

$2 plus Vat per calendar month

* Weekly podcasts with myself and the FFGeek Contributors

* All the articles on www.fantasyfootballgeek.co.uk advert free

$3 plus Vat per Calender month (plus the above)

* access to the FFGeek Contributor tracker team incl live transfers (52k OR in 19/20)

*access to Joseph Crilleys transfer planner and points projection tool

* access to the the Patreon Planner team runs from Joe’s planner projection tool incl live transfers (32k OR 19/20)

* Imaginary wildcard articles

*Bandwagon and Sinking Ships articles

*Slack Channel access

*access to private FPL leagues

$10 plus Vat per calender month (plus the above)

* Entry to a watts app group with myself including weekly RMTs and general FPL discussion (limited to 10 people)

No Commitment beyond 1 Calender month




Introduction

You can follow Joe on twitter here

For those unaware, the Fantasy Football Geek website introduced a new project last season, our Patreon site. This featured an interactive transfer planner which included points projections and fixture rankings as part of an automated and easy-to-use spreadsheet. Once this had been built, FFGeek suggested that I create an FPL team that based all of its decisions solely around the points projection model including starting XI, captain and transfers.

To be honest, we had no idea how the team would perform given that this was our first attempt at creating points projections, let alone having a team based on them. We would have been satisfied with anything inside the top 250k, but the model far exceeded our expectations and wound up with an impressive overall rank of 32k! For some context this was 9th in our league of 21 Contributor teams and was comfortably ahead of the rank for my personal team. Naturally, this led to FFGeek and I wondering what had specifically gone well for the model during the season and whether there were areas that could be improved.

Therefore, we did some analysis (mainly comparing against the Contributors analysis and 10 top FPL managers  who finished the last 3 seasons inside the top 10k OR) and here are the results!

The team will continue this season to be posted only on the FFGeek Patreon site

Captaincy

Selecting the captain each week was simply decided by placing the armband on the player who had the highest projection in the model’s squad. Given the projections are primarily based on underlying stats and FPL points per game (PPG), it was no surprise to see Salah leading the way for captaincy selection. As show in the table, the general tactic was to favour the big hitters such as Salah, De Bruyne and Sterling with the very occasional punt taken. This strategy certainly seemed to pay off as the model totalled 556 points for captaincy which was only bettered by one FFGeek Contributor and two of 10 top FPL managers! Given that I had such a poor season with the armband selection (431 captain points), there are clearly some lessons for me to learn from the model.

fantasy premier league tools

Transfers

Another area where the model had success was immediate points gained from transfers in with a total of 260 points. To again make a comparison, there was only one contributor and one of the 10 top managers who could beat this statistic. So why did the model perform well this area?

An obvious reason is that the model is very opposed to taking hits (only 12 points all season) as it rarely deems this worthwhile. Additionally, the method for calculating the projections does mean that it takes a while for poor form to catch up on a player, especially in the case of a premium. For example, Sterling remained in the team between GW2-GW13 despite only scoring 36 points in this time. This highlights an attribute that the model has which many of us do not have (including myself) – patience. Naturally, this doesn’t always work out as shown by the Sterling example. However, it does limit hits and I can recall hauls from over the season when patience has been a positive:

– Tielemans 12 points in GW4 after three blanks

– Gunn 21 points in GW3-GW5 after two blanks

– Pukki 12 points in GW11 after five blanks

– Pope 18 points in GW12-GW13 after three blanks

– Vardy 11 points in GW29 after six blanks

– Mahrez 9 points in GW38 after two blanks

What’s more Joes planner teams immediate transfer points of 260 was only beaten by 2 of 21 of the FFGeek Contributors and 1 of the 10 top FPL managers.  That was achieved despite having one of the lowest amounts of total transfers due to the low number of hits taken.

The final reason can be attributed to using our points projections which seemingly make some strong predictions. Typically, I would suggest that using this model will result in selecting the ‘obvious’ pick and as many of us know, it is often the obvious pick for a reason. Combined with the fact that ownership is not a factor, the model will rarely look to be too clever with its transfers and overthink things, something I’m certainly guilty of! Additionally, it will isolate good fixture runs which can regularly provide the spark for a player to start bringing in points.

In summary, by avoiding hits, saving transfers for mainly for times when players are unavailable and using the points projections, the model was able to perform effectively in area which is never easy to do so.




Bench problems (or lack of)

Now this is an area that I hadn’t particularly thought about until FFGeek sent over the stats, but it was certainly a strong suit for the model. It actually had the lowest points left on the bench compared to all of the contributors and the top 10 managers, so it is worth looking into. The fundamental reason for this is that rotation is not taken into consideration when building a team from points projections as it is in their nature to assume that a player will start more often than not.

Interestingly, I could only find one gameweek where we were unable to field 11 players so there was method behind the madness. The performances of Lundstram and Cantwell at the base price for their respective positions did help, especially in the first half of the season. Therefore, the strategy of targeting a strong starting XI by saving money on the bench in order to improve total points projection seemingly had some merit.

Also the Planner was very efficient on defender bench placement with only 4 times a clean sheet being left on the bench which was very low compared to others.

*On a quick sidenote, this stat would have been even better had I not forgotten to leave Ricardo Pereira on the bench in GW7 – goal and clean sheet for him that week!*

Team value

Now to focus on something that was less of a success as the final team value was 101.7mil and it was never higher than 103.3mil. The obvious reason here is that I would usually do the transfers on Friday with little attention paid to price changes. However, I would imagine that the aforementioned patience would also be a factor. It is difficult to quantify whether waiting to do transfers benefited the model due to more available information or meant that team value was a significant hindrance. Nonetheless, I may look to be a little more aggressive with the timing of transfers for the upcoming season.

Clean Sheets

The Planner team was good at predicting  clean sheets with a total of 66 clean sheets which was better than all but 2 of the 10 top FPL managers and all but 6 of the 21 FFGeek Contributors.  It was also very efficient in predicting which defenders to start and which to play with only 4 clean sheets being left on the bench.  Double figures clean sheets left on the bench by other FPL managers including the Geek who regularly used bookies odds was quite common

Start of season

It took a while for the model to get going last year shown by ranks of 827k and 631k after GW6 and GW10 respectively. This will partly be due to the fact that the projections are based on last season’s data at GW1 and it can take time for the current season form to have impact. Also, this was the first season for the projections so the model can learn from experiences of last season and should be better placed for picking a GW1 squad this time around.

Chips

Let’s finish on a positive note – it’s fair to say that the final rank was boosted by the various available chips. The first wildcard helped to recover from the slow start while the second wildcard and free hit were used to cement the position inside the top 50k during the final weeks.

The major success stories here though were the triple captain and bench boost. The former was used on Salah during Liverpool’s double gameweek to return 48 points and a gameweek rank of 22k. The latter was even better after we used the free transfers to build a bench boost squad for GW30+, which registered a whopping 125 points for a gameweek rank of 20k!




Summary

It was an all-round success for the model in 19/20 and an overall rank of 32k is highly impressive for a first season. The highlights were captaincy picks, chip usage and transfer success which showcases the strong upside of using the projections provided by the planner tool. With a stronger start and more aggressive transfers, we will be targeting an even better campaign!

Based on this season you should get great use out of the Planner team not only in the general sense with it’s high overall rank but also by using it as a point of reference for Captain picks, transfer picks clean sheets and bench decisions all of which were very successful this season.

Ahead of the new season, I will be writing a new article to go alongside a Youtube video to provide a visual demonstration of everything that this planner can offer for those interested. The Planner team based on the projections will then be available to see on the Patreon site with updates for every gameweek including GW1. The first version of the Team Planner and Points Projection Tool and the FPL team resulting should be out before the end of August on the Patreon site.  Hope to see you there!

 

 Other posts

Player Rankings Defenders

Player Rankings Midfielders

Player Rankings forwards

The Underlying Stats team of the season

The best attacking defenders

The FPL site is live

The first draft FFGeek team

The FFGeek Contributors first draft team

For more fantasy premier league tools follow  FFGeek on   twitter and facebook and instagram for article updates and news

You may also like to read one of these too: