Man + Machine: First Impressions

This article is part of our Man + Machine series.

This is a fantastic tool and resource for daily players.

I wanted to leave my lineups largely to the mercy of the machine to identify strengths and weaknesses of the optimizer. After all, as daily and seasonal fantasy players we all have strengths and weakness, too.

And while it hasn't been a lengthy or in-depth study with oodles of results, there are a number of immediate observations to share which should aid your use of the optimizer going forward. Mine, too.

First, though, here is how I approached the first three days:

1. I clicked the show me the optimal lineup button and entered the roster exactly.
2. I identified four to seven "locks," locked them into my lineup and then hit the optimal lineup button and entered the roster exactly.
3. I entered the lineup from Step 2, removed the players I didn't like and locked in the ones I did and hit the optimal lineup button again. I did this until I was happy with each player.

As you can see, the human element is quite limited.

With that said, there was the occasional late swap because a player wasn't in the lineup. Wednesday, I had to swap out Miguel Montero out at FanDuel, and I did not use the optimizer to find his replacement. Salvador Perez, James McCann and Yadier Molina were the players I replaced him with based on the remaining cap space for each lineup.

At DraftKings on Wednesday, I also had to swap out Andrew Susac in two lineups. I started Yadier Molina and Brian McCann without using the optimizer.

Also, on Tuesday I locked Ryan Vogelsong into a FanDuel lineup to see how the optimizer would stack the hitters. The results of the Vogelsong lineup have in asterisk in the results chart below.

Lastly, on the same day, I didn't use players from the games between Detroit-Pittsburgh and Atlanta-Washington because of the potential for unfavorable weather.

I just add those notes for transparency.

Here are the locks I used each day:

Monday:Lance McCullers, Prince Fielder, Mike Trout, Carlos Correa and George Springer at FanDuel. At DraftKings, I added Josh Donaldson and Mike Leake to those five.

Tuesday:Chris Sale, Josh Phegley, Adrian Gonzalez, Kris Bryant and Mark Canha at FanDuel. At DraftKings, I did not use Canha as a lock, and I used Ryan Vogelsong along with the other four.

Wednesday:Jon Lester, Jason Kipnis, Yasiel Puig and Adam Lind at FanDuel. At DraftKings, I added Carlos Carrasco.

There wasn't significant research involved in the lineup staples from Monday and Tuesday. I penned a Fantasy Aces article here yesterday, so the Wednesday locks were a little more supported by statistics. Otherwise, the first two days I went purely on surface hunches.

Here are the results:

June 30 did not go well. My locks didn't perform, especially Vogelsong. However, the other two days show an increase of points for each lineup as I went through the steps, which is the progression I hoped for.

Choosing Lester over Carrasco was the only reason the FanDuel lineups didn't have that result July 1.

Again for transparency, I did not fill out the chart or track my results closely until sitting down to write this column. So while I expected and hoped my lineups would beat the machine, I did not change the course of lineup assembly except where previously noted.

Optimizer Pros

There are endless positives to list. I'll stick to this one example, though: the optimizer has been in love with Shin-Soo Choo, so when Mitch Moreland popped into my lineup during Step 3 while finishing my final lineup Monday, I locked him in.

Moreland was only $600 more expensive than Choo, and he had the same handedness advantage against Bud Norris. Moreland has also been fairly hot of late, or at least he boasts home-run upside. He left the yard twice that game. Choo also hit a home run, and almost hit another.

Picking the right players isn't always easy. My locks didn't always pan out, and neither did the optimizers. However, between the two of us, I'm confident we can collaborate to identify favorable match ups.

I'll continue to look for specific examples going forward, but I encourage you to check it out for yourself and find your own.

Optimizer Cons

Batter vs. Pitcher

The optimizer doesn't automatically rule out batters and hitters facing each other. It is an easy fix, and something I'll alter going forward, but I have just let it ride so far to minimize my interference in the selection process.

Starting a hitter against your pitcher is like selling your car for gas money.

Star Struck

You'll want to become accustomed to seeing Mike Trout, Paul Goldschmidt, Stephen Vogt and others almost daily. The optimizer doesn't mind spending up to lock in the top hitters available at the expense of other positions, so just be careful with that.

Spending up isn't always spending wiser, after all.

What's Next?

I'm going to take a more hands-on approach with the optimizer and begin entering lineups into cash contests. Thus far, I've just been creating free head-to-head bouts to track the results. Now that I have a better feel for the machine, I'm looking forward to using the lineups I create with it against the field.

I'll continue to track results and experiment, too. Plus, if there are strategies or ways to utilize the optimizer you're curious about and want me to tinker with, please let me know in the comments. These trials are aimed to help our shared experience with the optimizer and hopefully improve it and our use of it.