It is essential to understand the methods of any model or analysis to correctly apply the associated output. I want to take some time to walk people through how I do most of my work without giving away too many secrets behind the process I created. Honestly, I feel anyone could have done the analysis I did. Nothing is revolutionary or overly complicated; I just viewed it differently and focused on functional graphic representations of the data.
Step one in understanding my methods is to look at the assumptions. All my algorithms and player-specific qualifiers are based on some fundamental assumptions. All other scoring methods are variants of my analysis are based on these assumptions and calculated relationships. The assumptions are:
- 12 team League
- 1 QB, 2 RB, 2 WR, 1 TE, 1 Flex (TE/RB/WR), 1 DST, and 1 Kicker
- Standard Scoring (STD); Decimal Scoring
- No points per receptions
- 10 yards rushing per point
- 24 yards passing per point
- Rushing TD 6 Points
- Passing TD is 4 points
Step two is to understand the backbone of my trade value model. I found that most trade value charts have a hard time with cross-positional values and don’t have a ton of data to calibrate their values. I started my journey by going through hundreds of trades on the weekly Reddit Player Value threads. My goal was to find similar, even-valued trades focused on one-for-one positional swaps. I wanted to calibrate the value of WRs, TEs, and QBs in terms of RBs. The basis for my analysis is that running backs are THE currency in 1 QB leagues. They are by far the rarest and greatest positional need, so I wanted to use RB values as the basis of everything. I looked for one-for-one trades because trades with lopsided players (example: three-for-one) make the math much more difficult. From there I was able to write relationships for WRs in terms of RB values and the same for TE and QB. I also use WR to TE and WR to QB relationships to calibrate weekly data, but with a much lower weight factor. From this historical data, I wrote a series of functions that I use to predict future Trade Values. I can then aggregate weekly Rest-of-Season (ROS) ranks from Fantasypros and Harris Football to get reasonable position rankings. I then will average the results with existing trade value data (CBS Trade value Charts) to normalize with more data points that correspond to a weight. I round all player values to 0.5 to help with math and to account for significant digits. This is the basis for my standard ranks. To generate PPR and 0.5 PPR rankings I generated a PPR multiplier database. Each player in the league is assigned a PPR multiplier between 1 and ~1.4 (highest I have seen) that is a function of their targets, catches, team pass attempts, etc. QBs and DSTS always have a value of 1 and most RBs stay close to a value of 1 due to low passing volume for that position. The 0.5 PPR values are an average of the STD and PPR values.
Calibration is key for my model, so I regularly look at one-for-one positional trades in the weekly Reddit Value Threads to adjust the RB vs Other position functions as needed based on injury or starting scarcity. Recently, we have seen WRs drop in value when compared to RBs and high-end TE value increase due to scarcity. It will be interesting to watch the trends over the coming years.
The 2021 season is my first season where I was able to incorporate 2-QB or Superflex trade values. The main issue was that it is very difficult to find in-season weekly ROS rankings for Superflex. So I had to generate my own 2-QB relationships to apply in-season and calibrate as needed. To do this, I looked at historical values of QBs in Superflex leagues and wrote a function to calculate the value in terms of the rolling average of RBs, WRs, and TEs in nearby value. This allows me to predict future 2-QB values based on linear ranks and positional scarcity. I am still working to refine and calibrate this analysis since it is so new.
I will also be rolling out 6-point per touchdown and 3 WR rankings for my Paterons. To generate these trade value charts, I looked at historical data to generate individual QB multipliers for 4 PT vs 6 PT leagues. This was easy to do for veterans since there was available data but meant I had to estimate Rookie values based on analytical comparative players. That means there will likely be more swings in Rookie values as the season progresses and we get more actionable data. Three wide receiver analysis was a little easier since it was a function of positional scarcity and not really player-specific. I apply a tiered multiplier to inflate the scarcity of the wide receiver position in terms of RBs.