Weekly Leverage or Are All Fantasy Football Wins Created Equal
Updated: Aug 23, 2022
In my last post, I looked at how a team's playoff odds change depending on their record at different points throughout the season. That got me thinking can we measure how important each week is? A common refrain in the sabermetrics movement is leverage opportunities, moments in games that are more important than others. For instance in baseball, a strikeout with the bases loaded and 2 outs is much more impactful than a strikeout with no one on base and no outs. In football, we see this idea applied to what to do on 4th down, with charts like the one below
What we can see is the odds of winning the game depending on the decision we make and whether that decision is successful or not (as projected by the bot). That got me thinking, could we calculate that for fantasy football? Are there certain weeks that are more important than others?
We are going to use the same dataset as the last post, so ~100,000 fantasy teams combined to calculate the playoff odds for each week, depending on their record. We can then calculate the leverage of any weekly matchup by subtracting the playoff odds next week if you lose vs the playoff odds next week with a win. For example, going into week 1, everyone has an equal chance at the playoffs (40% chance in our dataset). With a win, your odds go up to 53% while a loss drops your odds to 27%, creating a 26% of leverage (relatively high). We can do this same calculation for every week and record combination to see how leverage changes across the season
Line and scatter plot of leverage on a given week. Line represents mean leverage for any week. Scatterplot is each individual record's leverage for a given week. Hover over any individual point to see the record going into that week
So as mentioned in the data, we can see that we start around 26% leverage and that on average this goes down week to week, bottoming out around 8% leverage on average in week 14. However, what is interesting is that this is really driven by a diverging amount of leverage. Teams that without a win or are undefeated see their weekly leverage drop quickly, with teams without a win having no weekly leverage as early as week 6 while undefeated teams' leverage bottoms out around week 7 or 8. This makes sense because your start either clinching or mathematically eliminating teams around then in the extremes.
Far more interesting is what happens to the teams who are around .500combinations. For them, we see the leverage increase with each week. Going into week 13, a 7-5 team has a whopping 53% leverage. A win and you have an 85% chance at making the playoffs, while a loss drops you to 32%. I actually am interested and surprised that it peaks going into week 13 (when there are still 2 more weeks for the results to change) rather than in week 14 (when you get teams who would make it with a win vs miss with a loss).
Overall, we can use this chart to determine the importance of winning on a given week vs preparing for future weeks. Particularly later in the season a 7-2 or 8-1 team can afford to give up a week to prepare for the future. However, if you are around a .500 record or worse, you can't afford to plan for the future. Obviously not surprising advice, but always interesting to see the data underlying it.
One interesting application of this leverage chart is what I am calling competitive leverage. This graph obviously works both ways so if 2 teams, each with a 50% win rate are playing each other, that game has massive leverage for both teams. In contrast, if an undefeated team is playing another undefeated team, that game has relatively little total leverage. Knowing this, you can even try to make moves to strategically eliminate your opponent from the playoffs depending on the leverage that game has on their playoff odds, not just your own.
As with the last post, the same limitations apply that I am not controlling how managers behave at different records, and I am limiting the data to 10 team leagues with 4 playoff teams. The amount of leverage in a given week will change as the league size and number of playoff slots change, but the overall point will remain such that teams on the fringe of the playoffs will have more leverage each week than the extremes.
Overall, we can use this leverage graph to determine how important a given week is. If we find ourselves in a high leverage situation, we may try to trade for a player with a great matchup this week, even if it reduces our rest of the season projections. The first goal is to just make the playoffs, and sometimes we need to prioritize this week in order to get there. The leverage chart provides the data to know when to do so.