Random Win % - Part 7: Getting the Edge on Draft Day With RW%

This is the seventh of a series of posts about the next evolution in valuing players for fantasy football. Just as value based drafting (VBD) corrected many of the errors in drafting by pure points scored, Random Win % (RW%) corrects many areas which VBD does not address easily. The end result - better drafts, more wins, and more championships!

Now that we know how to calculate RW%, we want to find the best way to maximize our total RW% on draft day.  Let's assume that we have the season RW% for each player. How do we use this info to draft the best possible team? 

We might start by drafting the player with the highest RW%. However, we find that just like drafting the player who scores the most points, this isn't always the best strategy. Our old pal VBD to the rescue! By using a baseline for each position and calculating the highest RW% compared to the baseline, we can select the player that will increase our chances of winning the most. I won't go into the different theories of dynamic baselines, etc, yet a search on VBD will give you a wealth of information you can apply.

That concludes my overview of RW%. As the season progresses, I plan to use RW% to review my drafts, discuss in-season performance, and as a starting point for related FF topics. Please let me know in the comments or through email/Google+ of any thoughts, questions, or improvements to this approach. Good luck this coming season!

Random Win % - Part 6: Benefits of the RW% Approach

This is the sixth of a series of posts about the next evolution in valuing players for fantasy football. Just as value based drafting (VBD) corrected many of the errors in drafting by pure points scored, Random Win % (RW%) corrects many areas which VBD does not address easily. The end result - better drafts, more wins, and more championships!

In part 5, we detailed how to calculate RW%. Hopefully, some benefits to this approach are clear, but just in case let's look at what this can do for us compared to VBD:

1. RW% gives us a scaled number that translates to something important - winning games. VBD does not do this. If I tell you that a player has a VBD of 70 and nothing else, you have no idea if that is good or bad. If a player has an RW% of 70% and nothing else, you know that this player is much better than average.
2. RW% allows us to combine multiple weeks of projections intelligently. If Player A has a VBD of 25 in week 1 and 0 in week 2, is he better than Player B who has a VBD of 10 in both weeks? If so, how much? There's no easy way to tell. If Player A has an RW% of 60% in week 1 and 40% in week 2, it's easy to tell that he is 2% more valuable than Player B who has an RW% of 48% in both weeks.
3. RW% can help us to compare the impact of rule changes! Is your league adding a flex position, going to PPR, or expanding by 2 teams? Comparing the before and after RW% scores for each player will allow you to see opportunities to capitalize on the changes while your opponents flail away.
4. RW% provides commissioners a way to evaluate the balance of a league. Does the top RB give a team a 5% edge compared to the top WR or QB, or are QBs way out of whack? An RW% evaluation can help create more balanced leagues without having to wait through multiple seasons to see the effects of rule changes.
5. RW% allows us to fairly compare the effectiveness of different projections. We all know that some projection sets tend to skew high, but it's difficult to tell how much better one set of projections is versus another in different scoring systems, league sizes, etc. RW% allows us to put the projections and actual results on a level playing field. We can then use relatively simple tools like correlations to see who's better - and by how much.
6. Most importantly, RW% maximizes our ability to use our projections to draft a team that can win it all. In my next post, I will review how we can use some VBD concepts to help us rule the draft with RW%!

Random Win % - Part 5: Specifics of the RW% approach

This is the fifth of a series of posts about the next evolution in valuing players for fantasy football. Just as value based drafting (VBD) corrected many of the errors in drafting by pure points scored, Random Win % (RW%) corrects many areas which VBD does not address easily. The end result - better drafts, more wins, and more championships!

In part 4, I outlined the basic approach to determine random win % for a given player. Let's flesh that out a bit and detail how to calculate the RW% for a given player:

1. Gather the projected stats for the week for every player.
2. Using the league's scoring requirements, determine the projected score for each player.
3. Using the league's starting lineup requirements, determine the pool of potential starters at each position.
4. Using the pool of potential starters, create enough fantasy teams to create a meaningful sample (I recommend 200,000, which will get you within 1% of the true RW% with high confidence).
5. Pair the 200,000 teams into 100,000 games of Team A vs Team B.
6. For each game, for each unique position, calculate the number of projected points the rest of Team A scores and subtract this total from Team B's total (Team B total - total of other Team A players).
7. For each potential starter, count up how many games that player's projected point total is more than the calculation in step 6 for his position. Note: you can also think of this as the number of games in which Team A would have outscored Team B if that player was in Team A's lineup.
8. Divide the number of games won (from step 7) by the total number of games to calculate the players RW%.

Our end result is that regardless of league size, scoring system, bench size, or position requirements, we can assign every player at any position a score from 0% to 100%. It's a bit of work, but in my next post, I will review the benefits we get for our trouble.

Random Win % - Part 4: Determining Win Percentages

This is the fourth of a series of posts about the next evolution in valuing players for fantasy football. Just as value based drafting (VBD) corrected many of the errors in drafting by pure points scored, Random Win % (RW%) corrects many areas which VBD does not address easily. The end result - better drafts, more wins, and more championships!

In part 3, I outlined how poker hands are evaluated in Texas Hold Em. We can use a similar approach to evaluate fantasy football performances. Consider a matchup between two teams, Team 1 and Team 2. Both teams score 100 with their current lineups. Team 1's quarterback X scored 20 points and the rest of the team scored 80 points. If we replace quarterback X with QB Y that scores 30 points, Team 1 would win. If they had started QB Z that scored 10 points, Team 1 would lose. Depending on how the rest of the team performs and the opponent, Team 1 might not have even needed to start a QB to win. Or, it might take a QB scoring 150 points to win the game!

Let's take this a little further. Suppose we looked at a million fantasy football games. For each game, we determine how many points a QB would need to score for Team 1 to win. If QB X scored more points than that, we credit him with a win. We could then compute the % of games that QB X makes Team 1 a winner. Call this QB X's Random Win %. We can compute this for any other QB, or any other position for that matter.

So what have we accomplished? We now have a measure that gives every player, regardless of position, a single value directly tied to the most important part of fantasy football - how much this player help you win games! In my next post, I will dive deeper into this measure and detail how exactly to calculate RW%.

Random Win % - Part 3: Why "Random"?

This is the third of a series of posts about the next evolution in valuing players for fantasy football. Just as value based drafting (VBD) corrected many of the errors in drafting by pure points scored, Random Win % (RW%) corrects many areas which VBD does not address easily. The end result - better drafts, more wins, and more championships!

What makes one fantasy football team better than another? We've all been in games where one team was a huge underdog yet ended up winning. This is also a theme in poker games like Texas Hold Em, as anyone who has lost an all-in bet with pocket Aces knows. There are a lot of similarities between fantasy football and Texas Hold Em. Both games start with a known quantity and then introduce randomness. The end result is that you can start out as an overwhelming favorite yet end up a loser.

One of the key tools of a good Texas Hold Em player is an understanding of the relative strength of starting hands. While 2 Aces is a better hand than 2 Kings, how much better is it? To answer this question, we have to know how likely our hand will beat our opponent's hand after all the cards are shown.

Here's the problem: the number of possible games is massive. To work around this, poker odds calculators simulate millions of hands and count how often one hand beats another. Due to the power of random sampling, we can get very good estimates without having to calculate every possibility.

I apologize for the digression into poker, but the concepts are critical to RW%. In my next post, we will use these concepts in fantasy football to definitively answer the question "how much better is an awesome player compared to a scrub?"

Random Win % - Part 2: Total Points vs Head To Head

This is the second of a series of posts about the next evolution in valuing players for fantasy football. Just as value based drafting (VBD) corrected many of the errors in drafting by pure points scored, Random Win % (RW%) corrects many areas which VBD does not address easily. The end result - better drafts, more wins, and more championships!

Virtually all fantasy football projections today are done for the full season. Nearly all leagues consist of weekly head-to-head games. Clearly, we have a gap here! While it is debatable how much this impacts the quality of projections, this clearly affects how we value past performance.

To address this, RW% uses weekly stats as its fundamental building block. While it may not seem that big of a deal at first, it will become more important as we dive deeper into the approach. In part 3, we will talk about why the approach is called "random win %" and why it is critical to improving our understanding of fantasy football.

Random Win % - Introduction

This is the first of a series of posts about the next evolution in valuing players for fantasy football. Just as value based drafting (VBD) corrected many of the errors in drafting by pure points scored, Random Win % (RW%) corrects many areas which VBD does not address easily. The end result - better drafts, more wins, and more championships!

As the old saying goes, I'm not here to bury VBD, but to praise it. The core principle of VBD  is that it's not points scored that is essential, but points scored compared to peers. That's as true as it was in 1996. But VBD does have gaps that we would like to fix.

• VBD is usually done on a total points basis, but nearly all FF is played head-to-head
• VBD does not help us understand how likely we are to win games based on projections
• VBD does not distinguish between consistency and inconsistency

Random Win % addresses these gaps. In my next post, we will start to dive in to the ideas that led to RW%.