The value of a kicker: points kicked above average and Elo Ratings

Photo courtesy: Scott Grant/

Instead of starting with odds I want to talk about evaluating the value of kickers kicking field goals.

If you are familiar with the great work of Derek Taylor (formerly of TSN and now the voice of the Saskatchewan Roughriders and host of the Sports Cage on 620 CKRM), this evaluation will be familiar in many ways.

Derek Taylor is well known for tracking statistics the CFL does not officially track. Among these are insightful knowledge like QB pressures created or given up on the offensive and defensive lines.

However, one of most well-known statistics tracked by Derek Taylor is related to place kickers.

He regularly tweets out his Troy Westwood Memorial Kicker Rankings.

These rankings diverge from typical CFL or TSN kicker statistics talked about during your average game as they attempt to give credit to place kickers who make field goals from lower percentage distances.

The methodology used compares the distance of each attempt to the average field goal percentage from that distance. (Derek Taylor uses field goal data back to 2005.)

If the place kicker is successful, then he gets credit for the points he scores above the average field goal made from that distance.

In other words, three points for a made field goal minus three multiplied by the odds a kick is on average made from that distance.

For example, a kicker from a distance with 50 percent success rate is worth 1.5 points if made or minus 1.5 points if missed.

I enjoy this as an alternative statistic for two reasons.

  1. It gives credit to a kicker who make their kicks while penalizing those who do not.
  2. It also attempts to capture the benefit of long distance kick risk-taking and success.

As with any stat, there are negatives.

  1. Some kickers won’t be asked to make deep kicks. Sometimes this is about a coach who doesn’t trust their kicker or it will be about a coach who wants to manage risk. Long missed kicks can become return touchdowns against. If you are in control of the game already, then this is a risk not worth taking for many coaches.
  2. Kickers will often unfairly be asked to make kicks they are not normally asked to in an attempt to win a game. For example, a last play 60-yard attempt versus a hail mary.

Just like any statistic this statistic must be properly considered within these relevant contexts.

Despite these weaknesses, the statistic is very good for indicating which place kicker has been most valuable to their team. In other words, when asked to kick from any spot, how much value has the kicker delivered on.

The one danger with comparing a kick from a particular distance to the average from that distance is that field goal kicking is highly variable. Kicks can be missed at any distance. When looking at the average there is no guarantee that adding a yard distance makes the observed average decrease.

Here is a chart of field goal attempt distances and field goal percentage made for the years 2013 to 2018 each as a separate series. Derek Taylor was good enough to provide his version of kicking data from 2005-2018 which allowed me to compare and combine it with my play by play extracted data for that period. In this bubble chart the size of the bubble is the relative amount of kicks from a distance.

CFL Field Goal % By Distance 2013-18
CFL Field Goal % By Distance 2013-18

There is a lot of noise in this chart. There is regularly a number of 50-plus yard kicks made a year in the CFL, but the sample size is really low. When we look at the larger distances with larger samples, we get a confirmation of what we’d likely all agree on. That agreement is that field goal percentage decreases as the field goal attempt distance increases.

Now it is probably easier to look at the data combined across all six years. Below is the resulting chart.

Combined CFL Field Goal % By Distance 2013-18
Combined CFL Field Goal % By Distance 2013-18

We can see that across the high sample distances after about 20 yards field goal kicking rates seem to consistently decrease.

However, it is important to note, if we were to judge each kick against the average seen in this chart, then a 50-plus yard attempt would look to vary rather widely yard to yard. Is a 50-yard field goal which has average 80 percent actually easier than a 49-yard field goal averaging 65 percent?

Simply using the average does seem to match our expectation of how the field goal odds should consistently decrease as the distance of the attempt increases.

There are possible factors to why field goal attempt distance may not match with field goals made. Part of this may due to how coaches defend different field goal distances.

Within the 50 teams will press knowing the odds of a made field goal, and often losing the game, are high. This pressure likely results in more field goal misses. However, above 50, many coaches sit back and hope the lower odds of that distance will work in their favour and the kicker will miss by themselves. This lets the defending team avoid penalties on the play that might make the kick shorter after being applied, but actually makes the kickers job easier from that distance.

One solution to comparing to the average across distances is to try and fit a line to represent how the odds behave as the distance increases.

The generally way this is done is through regression. Basically, the concept is to look at the graph and guess what shape of function seems to match the numbers. Then use a process which adjust weights of that function to fit it as close as possible to the data.

For field goal kicking one function that works rather well is known as the logistic function. It often fits well when we have an event with two outcomes (field goal made, field goal missed).

This two outcome type of event is generally called a Bernoulli response variable. For now we will be treating only field goal attempt distance (FGA) as our input variable. (In more complex field goal models other input variables will often include whether the kick is in a dome, the altitude of the attempt, and the weather conditions).

The chart below shows a logit model fitted curve to the field goal averages for the years 2013-2018.

Fitted Curve Combined CFL Field Goal % By Distance 2013-18
Fitted Curve Combined CFL Field Goal % By Distance 2013-18

This curve is more regular than the average. As we step up in distance, it always decreases. It also has a steeper drop off as we approach the distance at which kickers struggle. One negative is that we may not agree at longer distances. For example, it predicts a 60-yard field goal can be made 20 percent of the time and a 70-yard field goal six percent of the time. In spite of this, it does nevertheless give consistently more credit for any field goal made as the distance increases.

Looking at 2019 field goal kicking we get the following chart prior to Week 21.

2019 CFL Field Goals Above Average
2019 CFL Field Goals Above Average

FGM-Field Goals Made

FGA-Field Goals Attempted


PTSAA-Points Scored – Points Expected


I prefer to sort by the 40-49 range of field goals. This is the money range.

Most kickers in the 0-39 range are very close in value. Castillo made every one of his 23 kicks and Medlock made all of his 25 but their points above average is only 2.2 and 3.2. Barely one field goal worth of difference over the season.

On the other extreme above 50, usage of a kicker can depend on your team not having a great offence and trying to get any points they can. Some kickers are only asked to make 50-plus kicks if it is to win the game or if the weather is right.

Castillo stands out as the most valuable kicker. Ward is a close second. Ward has had a Paredes like weakness this year to miss in the 0-39 range as a surprise.

Medlock and Hajrullahu both fall into an interesting category where their weak spot is 40-49. Medlock specifically is the best kicker 0-39. Both of them, are below average noticeably in the money range but then are rather consistent above 50. My hypothesis is teams let off pressure at that distance which lets them relax and return to form. Teams likely should pressure both kickers the same way in the 40-49 range when they kick above 50.

Elo Rating odds

In terms of the East and West Semi Finals, Montreal is favoured at 62 percent (67 percent season) and Calgary at 65 percent (64 percent season) through Elo Rating. A good portion of this is due to home-field advantage as both games feature two teams closely ranked.

Week 16 It’s Elo not ELO

Week 17 Ticats lucky number 13

Week 18 Playoff pretenders and contenders.

Week 19 What win leverage numbers reveal about the race for the West Division title

Week 20 Examining West Division scenarios using win leverage numbers

Win Totals

I have been repeating for weeks that Hamilton should reach 15 wins and they did.

3Down contributor Ryan Ballantine won his bet that the Stampeders would reach 12 wins. Look for Jamie Nye over Grey Cup as he will have to use a sippy cup all Grey Cup weekend.

What are the ‘stories’ from the CFL franchise Elo Rating charts?

The riders slid back in front of Calgary last week by beating Edmonton. Calgary let a win slip away from them on the road at Winnipeg in the back to back. Montreal finally passed Edmonton who have continued their decent losing through the harder half of their schedule mostly without Harris. Otherwise, much is unchanged outside of Toronto making a point that Ottawa is the worst team in the CFL.

2019 CFL Pre-SF Franchise Elo Rating Chart
2019 CFL Pre-SF Franchise Elo Rating Chart

What is different in CFL season Elo Rating charts?

As we reach the end of the season the franchise and season Elo Rating charts have started to converge. This happens most every year, except when some major change like a loss of a quarterback or coaching change happens mid-season. Season Elo Rating shows a clearer differentiation between Montreal and Edmonton.

2019 CFL Pre-SF Season Elo Rating Chart
2019 CFL Pre-SF Season Elo Rating Chart

Playoffs Odds

For playoffs odds we can drop Ottawa, Toronto, and B.C.

Hamilton’s Grey Cup appearance odds remain essentially unchanged (moving up just above 80 percent).

Saskatchewan host of the West Final gives them 62 percent odds at appearing in the Grey Cup. Those odds would be considered inflated if Fajardo would fail to start the game due to his oblique injury. The bye will be in the Riders favour to give him a chance to return but oblique injuries can be longer term injuries. Especially for a QB who uses them to throw.

The Semi Final team with the next best odds at a Grey Cup appearance is Calgary at 26 percent. This is followed by Winnipeg and Montreal both at 11.6 percent and Edmonton at 6.7 percent.

2019 CFL Elo Rating Pre-SF Playoff Odds
2019 CFL Elo Rating Pre-SF Playoff Odds

Grey Cup Matchups

Hamilton features heavily given how expected it is that they would make it out of the East over Montreal/Edmonton.

Hamilton is in the Grey Cup in 81 percent of the matchups with the most likely competitors being Saskatchewan/Calgary/Winnipeg based on their current standings’ expectations.

The Riders have a lead over the Stampeders to take the lead in Grey Cup match-up combination expectations, mostly due to hosting the West Final.

A Grey Cup featuring only West divisional teams sits at 6.68 percent as Edmonton continues to slip in odds to pull the upset in the crossover.

2019 CFL Pre-SF Elo Rating Grey Cup Match-Up Odds
2019 CFL Pre-SF Elo Rating Grey Cup Match-Up Odds


Hamilton is clearly the best team across all three ranking methods.

The Riders sit second with a weak rating in Markov Chain ranking. (Season Elo Rating and RPI, which are based more winning rather than score, have the Riders ranked stronger than Markov Chain. The Markov Chain method is all about points for/against opponents. The implication is that the Riders win, but the margin of victory has not been convincing relative to their opponents in those games.)

Calgary and Winnipeg are almost exactly the same in ranking. I’d lean towards Calgary in a match-up due to quarterbacks. But bad weather would favour Winnipeg with their run game. However, it is very east to imagine the home field will dictate the edge in that match-up. Calgary winning against B.C. to secure this hosting will ensure that communication is not an issue for the home game, however the extremely cold weather forecast favours a run game as developed and consistent as Winnipeg’s.

Montreal is clearly next. They don’t have a clear strong point, but also not clear weak points.

Edmonton is the final playoff team. They have wins as see in Elo Rating, but they’ve really only beaten bad teams which hurt their Markov Chain ranking and RPI ranking.

Of the non-playoff teams B.C. has showed they can score but not beat good teams. Toronto’s D can’t stop anyone, but at least they have an offence. Ottawa clearly lags behind all other teams in both O and D.

2019 CFL Pre-SF Rankings
2019 CFL Pre-SF Rankings

The implied tiers of the average of the season ranking methods are:

  1. Hamilton
  2. Saskatchewan
  3. Calgary/Winnipeg
  4. Montreal
  5. Edmonton
  6. B.C.
  7. Toronto
  8. Ottawa
Hudson is a Ph.D. graduate and instructor in computer science at the University of Calgary. He is a fan of football in all its forms