Best of BP: How To Win The Main Event (Maybe)

This article was originally published on June 8, 2023.

---

Those that are in the top 10 of the NFBC Main Event are akin to the chosen ones; it is such a hard thing to do, that only a very few can claim to have done it intermittently, (much less frequently).

In case you might not be aware, the Main Event is a fantasy baseball competition with traditional 5×5 Roto categories and scoring, for two-catcher leagues of 15 teams. Not only do teams compete against the other 14 teams in their league for the title (and prize) for said league, but also compete for an overall championship among all 795 teams, distributed in 53 leagues as of this year, with the largest prize in the industry–a couple of hundred thousand dollars–on the line. As you might expect, competition is fierce and relentless from the beginning (during drafts) to the end (during FAAB periods). 

Lots of people every year start playing high-stakes fantasy baseball on that platform and other important ones, like Fantrax, so they have become ubiquitous in the industry. I had the curiosity of checking out how are the top teams at this moment achieving their success so we can have a glimpse of what it takes to be a front-runner in this kind of competition. Let me tell you, that’s not easy at all.

As of the moment of this writing, these are the teams and the contribution in each category from their players, in the top 10 overall:

Overall	TEAM	R	HR	RBI	SB	AVG	K	W	SV	ERA	WHIP
1	Reez Beez	420	120	414	62	.2633	583	35	25	3.333	1.133
2	Bronx Yankees ME2	414	109	382	64	.2637	565	38	26	3.16	1.131
3	Da Gildz Main Time WonTime!	449	126	457	63	.2515	538	39	29	3.369	1.181
4	Kurland – ME	395	116	427	85	.2621	583	38	24	3.389	1.236
5	Bronx Yankees ME1	418	117	405	53	.2591	549	40	29	3.416	1.074
6	Christenson	412	121	397	73	.2508	603	41	24	3.476	1.181
7	BK1NG	441	142	461	49	.2672	577	39	16	3.484	1.13
8	Hallie Scruggs	413	108	383	54	.2598	588	44	27	2.892	1.048
9	Martin_ME_1	416	117	383	83	.2694	553	36	24	3.781	1.224
10	SULTANS of SMACK	406	119	413	73	.2708	560	39	12	3.665	1.187

And this chart summarizes how each of their categories ranks in its own league:

Ov	TEAM	R	HR	RBI	SB	AVG	K	W	SV	ERA	WHIP
1	Reez Beez	1	2	2	5	1	1	4	5	1	1
2	Bronx Yankees ME2	3	4	6	5	1	2	1	4	1	2
3	Da Gildz Main Time WonTime!	1	1	1	7	9	3	3	1	1	2
4	Kurland – ME	5	3	2	3	4	1	2	5	1	5
5	Bronx Yankees ME1	2	2	2	9	3	3	2	3	2	1
6	Christenson	4	2	2	4	7	2	1	7	1	1
7	BK1NG	1	1	1	11	2	1	2	13	1	1
8	Hallie Scruggs	4	3	4	8	4	1	1	4	1	1
9	Martin_ME_1	3	3	6	1	1	2	5	5	5	8
10	SULTANS of SMACK	4	1	1	4	1	1	2	13	3	4

My biggest questions were: which are the categories with a better correlation to a good position, and which are those that introduce the biggest variance in the team’s performance, for this group? With the important caveat that this is a really small sample of an extremely biased population and that some math is coming your way, let’s try to get some answers.

To achieve this, there are some simple statistical chops that we can apply, starting with a direct correlation calculation for each variable–in this case, the categories versus the overall rank. After running a quick-and-dirty Python script for it, we can get the following plot with the results:

The longer the bar, the bigger the correlation; if it is to the right side the correlation is positive and it means that the better you do in that category, the better rank you achieve. If it goes to the left, the relationship is the opposite: the worse you do there, the better you rank.

We can see that batting average, ERA, and wins are some of the biggest contributors with a direct correlation, while saves tops them all with a negative one. The first part is understandable, but how can it be possible that the worse you do in saves the better you rank? That doesn’t make sense, right? 

Well, yes and no. As the old adage says, correlation doesn’t imply causation, so we need to try to find out what’s happening here. For that, the next step was to check the normalized distribution of the density for each category to try to see if there was something else happening:

Now this is interesting: bolded, we have the biggest correlations and, dashed, the saves among them. If we calculate the area under each curve, we find that the latter has the biggest one by a good margin. Combined with the biggest leaning to the right, this could be interpreted that what happens with saves is that they introduce the largest variance to the overall result, which makes sense as we know saves is probably the hardest-to-predict category in roto.

What all of this means is that these teams have gotten a solid baseline on most of their categories and that batting average, ERA, and wins, to a bigger extent, are improving their chances but saves are moving them around, for better or for worse. 

Does this mean drafting stud closers will help you diminish this variance and propel you in the rankings? Possibly, yes, but what I read from this is that for some of the best fantasy players in the world, nailing down the saves is as hard as for almost everyone else. It might be a good idea to always keep this in mind when drafting and make sure that we don’t try to solve for that one category in detriment of the others.