The Weekly Gut Check - Vol. 40 Consistency Revisited - Part I: Crank Scores 8/2/05

Crank Scores 2004: QBs | RBs | WRs | TEs '03 & '04 Combined: QBs | RBs | WRs | TEs

Standard deviation doesn’t take into account what fantasy owners really need know: how consistently a player averages at least the desired amount of fantasy points per game. Fantasy owners actually desire the “inconsistency” that occurs because a player had several performances significantly above the desire fantasy points per game average. Priest Holmes didn’t score less than 12 fantasy points in a game last season, but his standard deviation is among the highest of all backs. Forgive The Gut Check, but he’d gladly take this kind of inconsistency—especially with a fantasy point per game average of nearly 25 points!

The Gut Check ranks consistency in a different manner. He tracks how often a player reached a desired baseline on a weekly basis over the course time. Determining a weekly fantasy point baseline and calculating the percentage of times a player reaches that performance level during a given range of games or seasons is a more effective starting point. This type of analysis doesn’t penalize players for performing significantly above their normal level of play when that “average” performance is still impressive. The key to effectively creating worthwhile baselines involves some research into historical average values for every skill position.

For those of you that didn’t see last year’s column, Yours Truly provided an in depth explanation of his consistency formula. This year The Gut Check is taking his consistency formula to the next level: in addition to providing a series of percentages and leaving it up to the individual to figure out how he wishes to use the information, Yours Truly has developed a consistency rating score (Crank Score) for those owners looking for more of a finished product.

The Gut Check’s method of determining consistency automatically sets tiers for individual performances. The Gut Check has created database queries to determine total fantasy points for each player within their position for the past 54 years. For now, the queries are set for standard scoring leagues:

• .1 pt per yard rushing/receiving
  
• .05 pt per yard passing
  
• 6 pts per rushing/receiving td
  
• 4 pts per passing td

This allows Yours Truly to see where tiers occur within a position during any range of time. Setting tiers isn’t a clear-cut process, but it’s not difficult to determine with a little practice. Generally, The Gut Check doesn’t incorporate more than a 2-3 year range of average values. These queries automatically average the baselines for the Gut Check’s consistency tiers:

• Elite Level: These are the top fantasy performers at their position. The percentage for the Elite category indicates how likely a player over the selected range of time performed each week among:
  
  
  • The top 2 RBs
    
  • The top 2 QBs
    
  • The top 3 WRs
    
  • The top TE
    
    
  For example, Kansas City RB Larry Johnson ranked at a positional-best 62.5% Elite Rating for 2004. This means Johnson was 62.5% likely to have a performance last year where at worst, only one back could have been better. Of course, Johnson only played 8 games in 2004, but for 5 of those games only one back in a fantasy league might have put up more points. This is definitely a validation of fantasy owners choosing to draft Johnson higher than any back up in 2005. According to this piece of data, Priest Holmes owners would be wise to select Johnson early enough to guarantee they have an elite runner throughout the season, regardless of Holmes’ health.
  
  
• #1 Level: These are the performers that score enough points in a given week to be considered a #1 starter on a roster. For instance, in a 12-team league, the #1 Rating is the player’s historical likelihood of performing each week among:
  
  
  • The top 12 RBs
    
  • The top 12 QBs
    
  • The top 12 WRs
    
  • The top 12 TEs
    
    
  These are the players most owners hope they are getting when they draft a player in early-to-mid rounds of serpentine formats. A great example of understanding the value of this information is to compare Edgerrin James and Curtis Martin in 2004. Martin scored 17.39 FPG to James’ 16.07 FPG, but James had an 81.25% #1 RB Rating to Martin’s 56.25% #1 RB Rating. Although Martin scored more points over the course of the season, James performances were more indicative of a top RB in a 12-team league. Later on, The Gut Check will demonstrate how drafting for consistency can positively impact a team.
  
  
• #2 Level: Depending on the lineup options of one’s league, the #2 Rating is a solid, fantasy starter (RB, WR, & Flex-QB positions) or a quality, fantasy backup (QB and TE). For a 12-team league these are the players’ likelihood of scoring within the top 24 of their position. In addition the #2 Rating for a position helps an owner see whether the point swings one might find with standard deviation are more often on the negative side of performance. Comparing Edgerrin James with Curtis Martin demonstrates that James’ 93.75% #2 RB rating made him a virtual lock to perform at the very least as the second back in a fantasy lineup throughout the season. In contrast, Martin only had a 75% #2 RB rating—still a good score, but that’s 3 fewer games than James. Three games in many leagues is the difference between a first-round playoff loser and the champion.
  
  
• #3, #4, or #5 Level: These are generally backup positions, although the #3 Rating often applies to starting receivers or flex-Rbs in many leagues. These scores aid an owner’s chance to acquire consistent depth for bye weeks or injuries.
  
  
• Sub-Par Level: This rating tells the fantasy owner how often a player performs lower than the average weekly score of a viable fantasy backup. In other words, how capable is the player of such a poor performance it could cost an owner a game. Let’s compare Warrick Dunn and Ahman Green from 2004. Green was considered a top 5 RB and Dunn a #2 RB, at best. Performance-wise, Green averaged 12.79 FBG to Dunn’s 12.13. Yet when it comes to clunkers, Dunn only had one sub-par game all-season and Green had three. Although this is just one year of data, the point is to show how Sub-par Ratings can be a very helpful way to separate players with a historically close level of performance in a more traditional sense.

Here are the fantasy point baselines for each position in a 12-team, standard scoring system mentioned earlier for both 2003-2004 and 2004:

RB Baselines x Year: Year Elite #1 #2 #3 #4 Subpar 2003 21.56 13.00 8.88 6.31 4.13 4.00 2004 18.75 12.31 9.81 6.25 4.00 4.00 2003-04 18.75 12.31 9.81 6.25 4.00 4.00

WR Baselines x Year: Year Elite #1 #2 #3 #4 #5 Subpar 2003 12.25 10.13 7.19 6.13 5.06 4.44 4.38 2004 12.88 10.31 8.63 6.56 4.94 4.19 4.19 2003-04 12.56 10.22 7.91 6.34 5.00 4.31 4.28

QB Baselines x Year: Year Elite #1 #2 Subpar 2003 20.75 17.63 11.06 10.94 2004 26.75 17.50 12.94 12.81 2003-04 23.75 17.56 12.00 11.88

TE Baselines x Year: Year Elite #1 #2 #3 Subpar 2003 9.50 3.75 2.5 1.69 1.63 2004 10.88 4.31 3.00 1.88 1.75 2003-04 10.19 4.03 2.75 1.78 1.69

So how do these baselines help a fantasy owner? The Gut Check feels consistency is a good complement for average value or value based drafting—traditionally selecting players based the highest amount of projected fantasy points. For example, let’s compare two teams in a 12-team league with a lineup of 1 QB/2RB/2WR and a standard scoring system. Team A has players with higher F Pts/G averages than Team B, but Team B’s players at their respective positions are more consistent. Which team has the inherent advantage over the course of a season in head-to-head leagues?

The tables show the team, fantasy points per game, percentage of time the player scored the minimum number of fantasy points to be among the top-12 scorers at their respective position for that week (#1 baseline), and how many games this player scored at this baseline:

Team A Pos Player F Pts/G #1 Baseline Games QB Kerry Collins 18.74 42% 6 RB1 Curtis Martin 17.39 56% 9 RB2 Domanick Davis 17.44 67% 10 WR1 Javon Walker 13.14 56% 9 WR2 Darrell Jackson 10.12 31% 5   Totals 76.8 50% 39

Team B Pos Player F Pts/G #1 Baseline Games QB Tim Rattay 17.11 55% 5 RB1 Corey Dillon 16.79 73% 11 RB2 Edgerrin James 16.07 81% 13 WR1 Joe Horn 12.87 68% 11 WR2 Ashlie Lelie 9.43 50% 8   Totals 72.3 65% 48

Without viewing any stats, most owners would look at the players on Team A and believe it is better than Team B. When using Fantasy Points Per Game as the sole-supporting stat of comparison, team A appears better—they score an additional 4.5 points per game, or 72 additional points over the course of a 16-game season.

The problem is Team A’s players performed as top-12 players at their respective positions less often than Team B’s squad. In fact, Team B’s players performed like top-12 players at their position for 9 more games combined over the course of the season! Breaking this information down by player further highlights the difference.

• Curtis Martin had a slightly better FPG average than Dillon over the course of 16 games, but Dillon scored the average baseline for a top-12 back in 2004 of at least 12.31 FPG more often than Martin. Although Martin had some bigger games, Dillon had more games an owner could count on as an RB1. It’s true Martin could carry a team to victory during one of his big weeks, but an owner that has a roster filled with players with Dillon’s reliability is tougher to beat on a weekly basis.
  
  
• Edgerrin James and Domanick Davis played the same amount of games, but James had three more games than Davis where he performed to the minimum expectation of a starting RB. Davis scored nearly 1.5 more points per game than James. In many leagues this difference in FPG would have placed him in a tier above James on many draft boards. But James’ had fewer dips in his performance, and that means delivering similar totals with significantly more consistency. As The Gut check mentioned before, three games (the difference in consistency between James and Davis) can separate owners from an also-ran to a playoff team, or a contender from a champion.
  
  
• Most owners would select Darrell Jackson over Ashlie Lelie last season, but the Broncos receiver actually had three more games where he performed like a #1 WR in comparison to Jackson—an established fantasy performer.
  
  
• The comparison between these two quarterbacks highlights the value of consistency in a slightly different aspect. Kerry Collins was nearly 2 points per game better than Tim Rattay. Although the 49er performed like a #1 QB one game less often than Collins, he played five fewer games. In reality, Collins was more valuable because he played more games, but if Rattay started the same number of games as Collins he’d clearly be the better choice if his consistency stats remained the same.

There is a lot to discover when examining each of the separate tiers, but The Gut Check has developed a formula to incorporate these levels of performance with fantasy points per game. The result is an overall consistency score for each position. For the sake of calling it something, let’s refer to it as The Crank Score (Consistency-Rank Score):

Crank RB Score = FPG+(FPG x Elite%)+((FPG x #1%)*2)+(FPG x #2%)-((FPG x Sub-Par%)*2)/FPG

Crank QB & TE Score = FPG+(FPG x Elite%)+((FPG x #1%)*2)-((FPG x Sub-Par%)*2)/FPG

Crank WR Score = FPG+(FPG x Elite%)+((FPG x #1%)*2)+(FPG x #2%)+(FPG x#3%)-((FPG x Sub-Par%)*2)/FPG

The formula is designed not to have a maximum score, because fantasy points per game drives much of the calculation. Yet consistency levels of performance illustrate where a slightly lower-scoring player could be a better fantasy performer in leagues with head to head match ups. The formula gives extra weight to #1-quality performances and heavily penalizes Sub-Par performances. Therefore, the players with the best Crank Scores:

• Churn out a high FPG average
  
• Consistently score like a #1-Quality fantasy player more than their peers
  
• Rarely have poor games that can kill an owner’s weekly performance

Here are some examples of this formula calculated for 2004 performances.

How can an owner apply this information to a draft list? The Crank Scores are calculated in a similar format across positions, so it inherently creates a natural function to tier players. Value Based Drafting, Average Draft Position, or Stud Running Back Theory can serve as great draft day preparation tools, but they are ultimately guidelines. The most successful and experienced owner understands when or when not to veer from the guideline. Remember we’re creating projections—plans for an anticipated event. Reality won’t always be a part of the picture.

Crank Scores are no different. A great example of what Yours Truly is talking about is to compile a draft list with tiers solely on 2004’s Crank Scores. One should immediately notice that several factors aren’t accounted for in such a list: injuries, off season player movement, and rookies. Despite these missing factors, the list is a good starting point.

First, pay attention to the tiers. A valuable guideline for drafting with a tiered list is to place the priority on selecting the best player from the highest tier within a position with the least amount of players in that tier. For example, if The Gut Check were in a 12-team league with standard scoring and holding the 12th and 13th picks, a tiered list is a great tool. Most owners select running backs in the first two rounds. Therefore, Yours Truly would have a great shot at his choice of either Culpepper or Manning in the top tier of quarterbacks.

But for the sake of exploring different ways to use tiers, let’s assume The Gut Check has the 12th and 13th picks and the following players are gone:

• Holmes
  
• Tomlinson
  
• Culpepper
  
• Manning
  
• Barber
  
• Alexander
  
• Moss
  
• Dillon
  
• Davis
  
• James

In this case, the draft list would dictate the Gut Check to select Donovan McNabb and follow up with Terrell Owens (assuming Muhammad wouldn’t truly be this high on a draft list based on his move to Chicago). Yet, here’s where projections and reality collide. It’s important to know your league. If running backs are flying off the board and will continue to do so after you select a QB and WR, odds are that your team will be in trouble. This is why these theories are best viewed as guidelines rather than gospel.

The Gut Check would examine his list and take the chance on McNabb falling to him in round three. If not, he’ll have his pick of Bulger, Green, Plummer or Favre in rounds 4-6. This is why Average Draft Position along side these figures—or any other projection information is helpful (the more specific to your league system, the better). Therefore, Yours Truly would select Owens and Curtis Martin based off this list.

So how does one incorporate the rookies or players rebounding from down years due to injury into Crank scores? Ahman Green is a good example. Here are his scores from 2003-2004 and for both years separately:

What about rookies? The Gut Check can query the average fantasy points per game and Crank score for rookies and limited the players to those that started a specific number of games in their first season. These figures can provide reasonable projections for rookies expected to see the field in their first season. Next week, The Gut Check will combine all this information and unveil his first annual Crank Score Draft List: an actual 2005 cheat sheet based on seasons 2003-2004 and projected Crank Scores compiled with specific historical data.