It's getting closer to August, we have a couple new #bestball tournaments to play with, and I'm a mix of bored and sleepy. Let's talk about CONDITIONAL ROSTER OPTIMIZATION in best ball, so you can become bored and sleepy too.
About 5 years ago, bestball became my "thing," simply because the idea of conditional optimal roster construction didn't seem to exist. I was getting into a new game, and I wanted to know how to beat it.
I found an article on @RotoViz that used a clever monte carlo simulation to determine the "optimal" roster construction for "MFL10s" (e.g. 2QB/6RB/7WR/3TE/2DEF). It was a good article, but after reading it, I was left with what felt like an obvious question...
"What if..." What if I start my draft with 3 WR picks in a row? What if I start with two tight ends? That "optimal" construction must change, right? And so began an unhealthy obsession with "conditional" roster construction
The answer, of course, is that optimal construction *does* change based on the picks you make early in the draft. I often think of it as one of those balance scales. 1 first round RB does not equal 1 12th round RB. This seems obvious, but I hadn't seen it addressed
I was able to illustrate this, kind of, in a couple of articles (rotoviz.com/2017/04/evolut…) through an algorithm that generated optimal rosters for every team in a 12-team draft, with prior knowledge of how their picks would perform
It became clear that where you picked your players (how much "draft capital" you spent on positions) impacted how many you should be drafting. Here's a pretty straightforward decision tree that came out of that very not-straightforward analysis at the time 
When I was running the @bestballcommand service, we had a channel in our Slack environment called "Draft Review" where subscribers would post the boards from their recent #bestball drafts, and I would give feedback. Most of my feedback would be abt relative strengths of positions
It took only a minute or two to size up each position. Without even knowing the player names, seeing where they were drafted gave me a sense of the roster's "balance." The whole thing is a balancing act. Quality vs. Quantity
Finally, I was able to build the "Roster Construction Explorers" that you can find today on @RotoViz. Millions of draft picks aggregated to illustrate these intellectually fairly simple ideas about balance.
How many tight ends should I draft in an MFL10/BB10? ... Definitely 2 or 3...
WHAT IF, I took my first tight end before the third round and my second tight end by the eighth round? Draft TWO!
The number one thing about draft strategy, to me, is that you should not go in with "a" plan... you should go in with 20+ plans. At least 5 of them are eliminated with each pick that you make, as the balance of the roster takes shape.
The real plan is flexibility, and balance. By keeping your roster balanced across positions, through quality, quantity, or a combination thereof, you put yourself ahead of the field. And then if you're really good at picking individual players, your opponents don't stand a chance
and even if you AREN'T really good a picking players, you DO stand a chance. That's what has made this so interesting to me. I am not trying to beat you. I'm trying to beat the game.
and the game is very beatable. @FF_Contrarian has been using my tools and data to lay it all out for you for over a year now. all you need to do is read. and if you want to play with the data yourself, it's right here: rotoviz.com/fanballrce/
The thing I really love about roster construction is that it looks like such a small edge. Look at those Tight End tables I just posted. at most a 1% edge in win rate? drafting Lamar Jackson gets me a 13% edge. so who cares? A lot of people think this way
The thing they miss is that there are dozens of these edges, and they ADD UP. Here is a table Shawn included in a review of this year's #MFL10ofDeath. It has nothing to do w specific players I picked. Just a collection of roster construction decisions. 2x the avg win expectation
and guess what, I might have also picked this year's Lamar Jackson or CMC
If I did, it's over. If I didn't... well, 65% of teams with CMC lost last year, and I don't think they lost to teams that didn't know how to build a roster.
