Part 1: Fantasy Player Price Change Predictions - How the calculations are achieved
These instructions must be read in conjunction with the notes at the bottom of the price prediction page.
The first step is to calculate the net transfers balance for each player and produce an initial table of absolute differences between the target and threshold values. The thresholds are calculated based on the rules given at the bottom of the price prediction table. It is important to note that the calculated transfer balance will be inaccurate as it contains transfers, from a number of sources, that in reality do not contribute towards the effective balance.
The next step is to attempt to remove the ineffective transfers. To do this we take the top and bottom twenty players from the table together with any additional players within the table who have high half hourly transfer rates. This is typically a further twenty players. A more detailed analysis is then undertaken for each of these sixty players.
The probability that each transfer is an 'effective' transfer (i.e. contributes to the overall transfer balance) is calculated for each of these sixty players - the calculations are described in detail here. These results are fed back into the original table and a weighted adjustment is made to the remaining players. This process is typically repeated every half hour. The calculated probabilities are reused until either a further recalulation is initiated or a players primary data characteristics are reset.
The consequences of this approach are that the players in the middle of the table will have less accurate target values and at times will 'jump' in value as the accuracy is adjusted. However, since these players are typically at least 50% away from a price change this approach is deemed acceptable.
A further consequences is that the accuracy of the probability calculation are dependent on the overall change in number of transfers made for each player during the last half an hour. These volumes can be quite small for players slowly trending towards a price drop and consequently players of this nature will have an inherent increased inaccuracy in their calculated target values. (Resolving this inaccuracy is currently being investigated and is the main factor affecting the accuracy of the price drops.) A similar inaccuracy exists for the slow rises - but this is easily managable.
Unlike other sites, this site does not and will never ask for access to your fantasy team password to help facilitate the price change predictions. At first glance having ready access to the transfers made by other teams should in theory improve the accuracy of the predictions. However, this approach, (if used), is flawed as it is not based on a random sample of teams. Rather it is based on a sample of teams who are predisposed to give away their password and team control. The actual teams to be analysed must be an appropriately sized random sample taken from the overall population of fantasy teams. Only time will tell if this assumption is correct.
A further change to the price predictions table is made at the end (and also at the commencement) of each gameweek when the system reassesses the effective baseline value of all players and adjusts each players value accordingly. Clearly, these readjustments will have a major impact on the key risers and fallers at a gameweek start. If all goes OK the re-adjustments are typically completed two hours after the gameweek starts, i.e. if the game week commences at 11:30 am then typically the repositioning will have settled by 1:30 pm.
Confused - Well now you know what it like for the fplstatistics support team!!!
Donate to this website
The continued operation of this website together with the supporting analyses are funded through the 'Adverts' and the 'Donate' facility.