Hi folks. I’m writing a new weekly piece this year that is a different way to look at expected final record as the season progresses.
As you see it commonly posted, win expectation is typically cited as a number of specific win/loss results pegged for each remaining game or a single total of wins. We all know no game is either a sure win or a sure loss, but I’m going to attach a probability of a win to each remaining game as I re-project the Ravens’ final win total after each game.
Furthermore, I’ll attempt to break down this final win expectation on a running basis by explaining the “whys”.
Let’s try an example. The formula after week 1 will be:
(Total Expected Wins after week 1) = (Total Expected Wins prior to week 1) + (Change Due to Week 1 Result) + (Change in Expectation Due to Talent/Scheme) + (Change in Expectation Due to Injuries) + (Change in Expectation Due to Opponent Events).
Let’s examine those pieces individually:
Total Expected Wins (Prior Week): Self explanatory. For week 1, I need a starting win expectation for the season (see below).
Change Due to Week’s W/L Result: This is a simple, backward-looking calculation (not an assumption) obtained by taking actual wins for the prior week (1, 0, or 0.5 for a tie) and subtracting the expected wins from the previous game. For example, if I say the week before that the Ravens have a 70% chance to beat the Bengals at home and they do so, my estimation of expected wins for this component will increase by exactly 0.3 wins for the season.
Change in Expectation Due to Talent or Scheme: This is an assumption I’ll make each week. Examples of why it might change:
–The offense, defense, or special teams is playing at a different level than what was previously baked into the assumptions.
–A new scheme has been effective.
–Talent to replace expected roster deficiencies has emerged or declined.
This is the key lever with which to change expectation. It should also be obvious that the more time remaining in the season, the more impact a rising star (for example) will have on the win total for the remainder of the year.
Change in Expectation Due to Injuries: This is an assumption and could be positive or negative. It includes both injuries sustained that week and any changes to injury prognosis. An injury-free week should significantly increase the expectation of games won, while a key injury can have a significant negative impact (duh). Since we are frequently dealing with incomplete information here, injury prognosis is a significant assumption. Opponent injuries are not considered here, only those accruing to the Ravens.
Change in Expectation Due to Opponent Events: Injuries and changes in talent/scheme for opponents still remaining on the Ravens schedule are included in this factor.
Each week I’ll give my rationale for the changes in the 3 assumptions (talent/scheme, injuries, and opponent events) together with rationale for the expected win percentage for the coming game.
We’re almost done. We need 2 assumptions to start:
Start of Season Expected wins: Based on projecting each game individually, I currently estimate the Ravens will win 9.5 games this season. Since most of the people who read this are Ravens fans, you’ll probably think this is too low (I hope you’re right). If it is too low, this piece will document the emerging talent, good injury luck, and ability to win close games which led from the original estimate to the final win tally.
Probability of a win at Denver: As I have it now estimated, this game is the least likely win of the season. Playing a road opener against one of the top AFC teams with question marks in the secondary is not ideal. However, the Ravens have the deep defensive line rotation to reduce the impact of altitude. Looking at a cross section of 20+ online bookmakers, the best prices on the Ravens and Broncos reduce to a 33% chance for a Ravens win. That sounds just about right to me, but I’m going to bump it to 35% based the chance the Ravens’ can dominate both sides of the line of scrimmage offset by the warmer weather favoring Denver’s offense and Manning in particular. Again, I’d like to hear why you think 35% is too low. However, if you really feel it should be 45-50%, I’d suggest you wager before posting.
We’ll follow along with the change in expectations by the attached graph.