Introduction to new patreons

The Transfer Algorithm is perhaps a bit difficult to understand for those who havent been following them over the years at the Transfer Hub. Here is a short version of what it is, and how to use it: 

The Transfer Algorithm load tonnes of player data for every player, and combine those data into expected FPL points for their past performances. In addition to the FPL points they actually have managed to score, these data combine to give the strength of each player going forward. Recent performances count a lot, past performances count less.  Combine this strength with the expected minutes for the player going forward and the upcoming fixture list, the Transfer Hub calculates the point-expectation for the player.

All players are then subject to the money-limiting vector field (captaincy material considered) . The vector field create a list of players by a BCV rating (BCV stand for Budget-Constrained-Value). The ranking of players is trying to tell which players will be the best to buy/hold, and who will be the priotiy to get rid of. It's here important to note that BCV is assuming you need to field the player almost every week. Hence, cheap rotation assets will be somewhat undervalued by BCV.

Well, that's the case for midfielders and forwards at least. The quality of these projections should be top class among all the various sources on the internet. Defenders though, I gave up years ago. Hence, the defenders ratings published are created by importing third party data from elsewhere, and importing it into the money limiting vector field to create a BCV rating. To much frustration, I must say third party projections got variable quality. I'd therefor advice to only use the defenders ratings as a looser guideline. To take my self as an example, I follow the Transfer Algorithms conclusion > 95% of the time for midfielders and forwards, but it's not uncommon for me to deviate from results regarding defenders.

A practical example can be as follows:

Let's say your worst players by BCV (not counting the bench junk) is for instance Kane and Harrison, and you got 0.5 ITB and 1 free transfer. You already own Toney, so he isnt an option.

Simple version:
In principle easy, sell low BCV players for high BCV players. The best alternative among forwards would be Kane->Antonio, improving BCV by 0,05. The best alternative in midfield would be Harrison ->Sarr improving BCV by 0,07. The midfield transfer is then rated as the generally better move.

But as the next GW point expectations are larger for your current assets, saving the transfer is definetly the best idea to potentially do the transfer the next GW.

Nerd version:
Too be completely precise, you also need to take into account money involved. A +0,10 BCV transfer involving players at 10 mill. each make for 0,10 BCV x 10 mill. = 1, while a +0,10 BCV transfer involving cheaper players at 5 mill each make for 0,10 BCV x 5 mill = 0,5. Doing the BCV improvements are more important for the expensive players.

In addition we have weak forces like price movements impacting play - at least at the start of the season. Price falls (below selling value) are much worse than the benefit of a price rise. What matter here is the selling value of a player and the future benefit of more funds to play with. Earlier mathematical studies Ive done show the incremental value of the last 0.1 mill. is about 2 points over 38 GWs, decreasing linearly as GWs go by. Transposed to BCV, this is about 0,03 BCV for a price rise, and -0.06 BCV for a price fall below selling value at the start of the season. Half way through the season, this is hence 0.015 BCV and -0.03 BCV respectively.

Another weak force is leaving dead money in the bank for a week. 1 million in the bank correspond to about 0.016 BCV.

Let's go back to our example above, and assume Harrison seem to be falling in price, while Antonio seem to be going up. To calculate the best moves here we can set it up like this: 

Resulting in saving the transfer being the optimal move.

But it's not that easy either. In addition you will have to look into short term elements like the next fixture and hence short term point expectations. That means, you will probably be able to time the transfer a little better than playing strictly by the Transfer Algorithm's BCV value, using the flexibility of saving transfers and timing moves. That said, being stuck with short term solutions very often lead to frustration, as issues arise down the line. Sacrificing short term edges for the long term shape of your squad should be held as a general principle.


Other practical values:
The BCV value of a 4 point hit is not fixed and unsolvable, as it is so dependent on team shape and the likelyhood you will be able to fix the problem for free down the line. As a practical guideline, Id say the value is in the range 0.15-0.30 BCV. This means hits in only very rare cases will be the optimal move if you havent messed up big time in your planning.

As for the option of saving a transfer, a BCV value of about 0.10 BCV is for practical matters often about right. This will often lead to saving being your best option. As for hits, it's not really a fixed value, so take it as a good guideline.

Here is a little more information about the building blocks behind this spread sheet present: https://www.patreon.com/posts/20740418. (Old article, so design has changed)

By becoming a patron, you'll instantly unlock access to 103 exclusive posts
203
Images
1
Link
50
Writings
By becoming a patron, you'll instantly unlock access to 103 exclusive posts
203
Images
1
Link
50
Writings