Sadly, I did not make it to NY for the Draftkings Fantasy Baseball Champs. Call it a TKO... after 8 hours at the airport on Thursday, two flight cancellations (one after boarding), and no available flights on Friday.
This is a two part blog.
Part one describes what Advanced Sports Analytics (ASA) will be like going forward. Part two describes my thinking behind the three line-ups I chose for the Fant Baseball Champs.
The goal of ASA was to move the discussion in fantasy sports away from mean expectations and towards correlations and distributions. It was targeted especially for serious tournament players.
I don’t want to lose the discussion in too much detail (or time,, as I’m making line-ups), but here are some examples of how tournament players could use ASA data to add value over other available fantasy data….
You’re a tournament NBA player
You have in front of you the betting lines, and some excellent projections of expected mean fant points per player. Even if you have perfect projections, you have four problems: 1) everyone else has decent mean projections and ownership correlates with mean projection (thus in high dollar tournaments, good projections alone won’t make money for you). ASA doesn’t solve this problem. 2) in a tournament, you need to avoid guys who have high expected fant pts per dollar but a tight distribution of exp fant points (you need wide distributions… you need the possibility of very big scores and thus winning). ASA specializes in data on the distribution of fant points. 3. You don’t know the corr of fant performances across players on the same team (to win a tournament, you need to seek pos correlations and avoid neg correlations). ASA solves this problem. 4. You don’t know how fant performances of players change by a) the actual point total in the game being higher (or lower) than the Vegas expected point total or b) the actual point spread in the game being higher (or lower) than the Vegas projected spread. ASA specializes in this type of data. If you want to position yourself for, say, a GSW blowout win, ASA data would tell you exactly which GSW bench players to target.
OK, I was going to do a NFL and MLB example, but you get the idea.
Two crucial points: 1. to collect this data, ASA cost exceeded ASA subscription revenue. 2. Mason Grimshaw, who wrote all the necessary scripts and had ALL of the technical expertise needed to manage the obscene amounts of data generated, took a real job.
So what I’ve done is open up all of the past data to everyone. The paywall is removed, but data will no longer be updated.
If someone who does have technical skill is interested in keeping the data updated as a hobby project, shoot me an email at firstname.lastname@example.org. Can’t promise responses. Also, you need to be wizardly in your programming skills, like Mason, and willing to work for free (at least in the short term).
I am going to try to keep up this blog for a little while. So ASA will shift from mostly data with paywall, to mostly blog with no paywall. The data will still be there, but it won’t be updated.
So now part 2… DK FBWC lu’s.
I have three entries.
It’s an interesting short slate… 6 games… all those starting bw 7.05 and 7.15.
Some possibility for weather in BOS vs BAL.
BOS-BAL is the second part of a double header.
I looked at data on this… intuitively one would think that the second half of a double header should be avoided. Pinch hit risk seems to go up, and guys get tired (though this affects pitching and defense as well). Roughly speaking, the data suggests that you can ignore this intuition and play the second half of a back-to-back with confidence. This situation might be different though bc a) there’s weather, which could stretch the game and b) the game last night was unusual, with 31 runs scored.
I’m super high on Bauer. CHW has been great to attack all year, and Bauer has a perfect pitch fit. Right hander with lots of variety… sets up perfectly.
I find it hard to gauge bauer ownership in this spot. I’m guessing 60%, though honestly I have no idea.
Even at this ownership, I don’t consider him a good fade. At this point, I’m planning to have him in 2/3 or 3/3 lu’s. Subject to change as word count increases.
You have to pair him up with someone on the cheap, and I think Straily is the least-bad option.
Marlins might be the team I know best (value of this knowledge=0; maybe it all changes tonight.
I can analyze Straily for you very easily…
Quality= slightly below avg.
Sign me up.
Also we have confirmation from Vegas…. MIA is the favorite.
Their 6-7-8-9 looks like this
Bauer plus Straily, though it should be chalky, seems so good that I think it’s best to err towards playing it. There are plenty of minefields elsewhere. And there are enough ways to go with hitting that we can seek our variance on the hitting side. It’s worth noting that I don’t see much of a ceiling in any pitcher outside of Bauer.
At C , I think Realmuto is underprojected by a tad by most sites.
I think the Vegas line of 8 runs MIA -115 is prob right.
We have the weak bottom of the lu above plus
I think it’s Realmuto-castro-dietrich doing most of the work among these 8. I think that’s a reasonable mini-stack that fits well with Bauer and gives you corr with Straily. Also figures to be quite unique.
This stack couldn’t include any HOU guys. The value in HOU is bregman-correa-gattis , power righties. You need gattis there, and you want the correlation, so the inclusion of Realmuto means all these guys have to be excluded.
NOTE: as pricing gets tighter… and right now DK pricing is very tight… correlation becomes much more important. Winning is like hitting a parlay, and hitting a correlated parlay is much, much easier than hitting a non-correlated one.
So first lu is our non-bauer lu. It’s CLE power w our Realmuto-dietrich-castro stack, along w straily and duffy.
Duffy seems slightly underprojected to me. At any rate, the K upside is def there. I think he’s a good play tonight.
High value all around. MIA stack uniqueness w straily correlation, and the high value CLE 5way stack (that’s hard to get to tonight, and therefore unique). Can win by CLE going off, or CLE being avg (bc avg is still good positionally… an avg perf by lindor or Ramirez is still very good for their position) and MIA going off.
LU2 is somewhat behavioral… I’ve noticed that people, in live finals, are a) more conservative than usual and b) less inclined to pivot to new info. BAL vs BOS has been doubtful bc of weather all day despite having the highest proj total and good projections for indiv players. I decided to run a game stack there, as the weather proj improved a bit last minute.
The bottom of the CLE order offers, in my opinion, the best upside on the slate for the price. The bust probability is high, but the upside easily has the potential to win the tournament and allows you to fit in Bauer (w correlation).
I was able to roll my preferred combo of bauer and straily w the bottom of CLE’s order and some bats that I thought were high value one-offs….
Hope you enjoyed.