Preseason 2020 Blog Series Part 2: How do End of Season results compare to Projected Draft results?

We’ve had some news in the league since the last blog, Matt has decided that he can no longer make a commitment to the fantasy league anymore and has retired from Floored. I have reached out to longtime south Florida resident and friend of the familes, Ryan Olsen, to enter the league and he accpepted the invitation. Ryan played baseball and went to school with Michael and Max, was coached on at least one of those baseball teams by Paul, took some Poker money from Josh, and has met Arthur and Dave. With all of these roots to the league as well as an extensive fantasy sport background, Ryan will be a great addition to the league. For his professional background, Ryan went to University of Florida earning his law degree and is currently living in Detroit.

For this blog I wanted to learn something from years passed. The last few years I have gotten way more into projections. It’s pretty cool that baseball is so projectable in nature. You think Carlos Correa is a top 10 player? Well, his projections don’t really support that. You think Michael Brantley is washed up? Well, think again. Another thing I’ve learned is that projections are only a starting point. From the last two years that I have data and tracked results, a players’ projection is going to be accurate to around +/- 20 to 25%. As in, if the projections say he will hit 20 Home Runs, he is a pretty strong best to end up between 15 and 25. A big range, you say? Well that’s a lot better than just looking a list of players, sticking your wetted finger up in the air, and seeing which way the wind is blowing. The fun and skill of the game is finding breakout players, avoiding busts, making trades, and adding free agents, which are beyond the realm of projections, but projections are as great a place as any to start player evaluation.

The last two years I took a screenshot of what the projections said would be the results of the league at the draft, and we also have the data for where things ended up. Let’s take a look:

2018:

Based on the projected stats of the players drafted, the power rankings were going to look like this (remember the lower the number the better).

2018	TOTAL	hit	pitch
Dean	4.071	4.333	3.600
Michael	4.071	4.444	3.400
Dave	4.214	4.444	3.800
Cory	4.500	3.889	5.600
Max	5.929	5.333	7.000
Paul	6.214	6.333	6.000
Arthur	6.286	6.667	5.600
Matt	6.357	6.000	7.000
BJ	6.571	6.556	6.600
Keith	6.643	6.889	6.200

But by the end of the year:

2018 End of Season Ranks
TOTAL			HITTING			PITCHING
1	Michael	3.64	1	Cory	3.38	1	Brian/Josh	1.83
2	Cory	3.93	2	Dean	3.75	2	Dave	2.67
3	Brian/Josh	4.14	3	Michael	3.75	3	Michael	3.50
4	Dave	5.00	4	Matt	4.25	4	Cory	4.67
5	Dean	5.00	5	Paul	5.25	5	Arthur	5.17
6	Paul	5.29	6	Arthur	5.50	6	Paul	5.33
7	Arthur	5.36	7	Brian/Josh	5.88	7	Dean	6.67
8	Matt	5.50	8	Dave	6.75	8	Matt	7.17
9	Keith	7.86	9	Keith	7.63	9	Keith	8.17
10	Max	8.36	10	Max	8.38	10	Max	8.33

Michael stayed at the top of the rankings, but as you can see team BJ rose from 9^th projected to 3^rd. Right away we can see that these draft projections are not a be all end all. Lets check out how 2019 went:

2019	TOTAL	h	p
Cory	4.143	3.500	5.000
Dean	4.214	4.250	4.167
Dave	4.857	5.625	3.833
Michael	5.071	5.250	4.833
Paul	5.143	6.000	4.000
Matt	5.500	4.625	6.667
BJ	6.000	5.625	6.500
Max	6.143	4.500	8.333
ARTHUR	6.143	8.375	3.167
Keith	7.429	7.000	8.000

And by the end of the year:

2019 End of Season Ranks
TOTAL			HITTING			PITCHING
1	Michael	3.93	1	Dean	3.25	1	Michael	3.67
2	Max	4.21	2	Max	3.38	2	Brian/Josh	4.00
3	Dean	4.36	3	Dave	3.50	3	Paul	4.17
4	Dave	4.43	4	Michael	4.13	4	Arthur	4.50
5	Paul	4.50	5	Paul	4.75	5	Max	5.33
6	Brian/Josh	5.79	6	Cory	6.63	6	Cory	5.50
7	Arthur	6.07	7	Keith	6.75	7	Dave	5.67
8	Cory	6.14	8	Brian/Josh	7.13	8	Dean	5.83
9	Matt	7.00	9	Matt	7.13	9	Matt	6.83
10	Keith	7.50	10	Arthur	7.25	10	Keith	8.50

It was Max that made the big move last year, from 8^th to 2^nd, Cory fell from 1^st to 8^th while Keith was a beacon of consistency projecting 10^th and ending there as well.

So let’s figure out what happened. Running this same data but on a category by category basis, here is how things changed from beginning of year projection based on players slotted to be started by their teams to end of year results based on what teams actually recorded. Here is what 2018 looked like.

2018	Hitting	R	H	HR	RBI	SB	BB	K	AVG
DELTAS	AVERAGE	12.5%	12.5%	17.3%	18.0%	22.9%	6.7%	8.3%	3.2%
	MAX	6.9%	10.3%	10.0%	11.5%	12.2%	12.3%	16.2%	0.1%
	MIN	15.3%	11.9%	14.7%	20.9%	0.0%	6.1%	8.4%	4.7%
Absolute deltas	AVERAGE	12.5%	12.5%	17.3%	18.0%	22.9%	6.7%	8.3%	3.2%
	MAX	6.9%	10.3%	10.0%	11.5%	12.2%	12.3%	16.2%	0.1%
	MIN	15.3%	11.9%	14.7%	20.9%	0.0%	6.1%	8.4%	4.7%

	Pitching	W	SV	ER	K	ERA	WHIP
DELTAS	AVERAGE	-7.7%	-0.2%	1.7%	0.1%	2.6%	1.0%
	MAX	-15.2%	8.8%	6.0%	-2.7%	-0.8%	3.6%
	MIN	-1.9%	18.0%	3.6%	-1.1%	11.4%	-9.9%
Absolute deltas	AVERAGE	7.7%	0.2%	1.7%	0.1%	2.6%	1.0%
	MAX	15.2%	8.8%	6.0%	2.7%	0.8%	3.6%
	MIN	1.9%	18.0%	3.6%	1.1%	11.4%	9.9%

The positive numbers in the DELTAS rows means that the projected stats were higher than the actual stats. The Absolute deltas row is just the absolute value of the DELTAs rows to show how projectable the stat was without caring if the value went up or down.

The first possible reason for most of the offensive stats being over-projected is that when I summed up each teams projected stats, I included one bench hitter assuming that they could be substituted in for off days and injuries. (This is a good time to note that you should ignore what Yahoo shows as the standings in the draft room, they show the projected stats from everyone on your team, so when someone has 5 bench hitters, all their extra stats won't count if they can’t fit the player in their lineup). It is hard to say how much of an impact this actually had and it is likely a team by team difference. Michael almost never has an empty active spot if a bench player is playing, but Max or Matt on the other hand, well, they may have had a few.

For the pitching, the projections for the counting Starting Pitching stats (W, ER, K) mostly being under projected is going to be due to streamed pitchers. Each team would have thrown way more innings than projected as people streamed to win their matchup week to week.

It should be noted that the ratios projectsions (AVG, ERA, WHIP) were almost spot on for the league averages, showing that projections are most valuable in a per At Bat or per Inning Pitched evaluation. This is right in line with previous projection analysis done: a player’s skill level is projectable, however their playing time is likely to vary widely based on a number of factors like injuries and promotions/demotions.

Let’s see how 2019 looked:

2019		AB*	R	HR	RBI	SB	BB	K	AVG	SLG
DELTAS	AVERAGE	4.1%	-1.9%	-7.5%	-2.3%	22.3%	4.3%	0.2%	-0.1%	-2.9%
	MAX	3.5%	-2.6%	-5.6%	-4.3%	28.7%	6.0%	7.6%	0.7%	-5.3%
	MIN	9.8%	0.0%	-13.8%	-6.3%	48.8%	3.9%	-10.9%	-1.7%	-4.2%
Absolute deltas	AVERAGE	4.1%	1.9%	7.5%	2.3%	22.3%	4.3%	0.2%	0.1%	2.9%
	MAX	3.5%	2.6%	5.6%	4.3%	28.7%	6.0%	7.6%	0.7%	5.3%
	MIN	9.8%	0.0%	13.8%	6.3%	48.8%	3.9%	10.9%	1.7%	4.2%

2019		IP*	W	K	ERA	WHIP	QS	SV+H
DELTAS	AVERAGE	-19.6%	-17.8%	-18.1%	-22.2%	-6.9%	-27.8%	31.0%
	MAX	-19.5%	-18.4%	-20.4%	-28.3%	-7.0%	-23.4%	26.6%
	MIN	-1.0%	-10.7%	-2.6%	-22.0%	-6.4%	-16.1%	11.8%
Absolute deltas	AVERAGE	19.6%	17.8%	18.1%	22.2%	6.9%	27.8%	31.0%
	MAX	19.5%	18.4%	20.4%	28.3%	7.0%	23.4%	26.6%
	MIN	1.0%	10.7%	2.6%	22.0%	6.4%	16.1%	11.8%

I was forward thinking enough in 2019 to track At Bats and Innings Pitched at the draft and in the results. What you see here is very similar results. The streamed pitching impact is much greater here because of the rule changes we made to emphasize innings pitched and hence starting pitching. You’ll also notice that this really hurt ERAs, but did not hurt WHIPs as much. This implies that below-replacement-level starting pitchers’ WHIPs are rather close to their above-replacement-level counterparts.

On the offensive side the total At Bats came out a bit closer to projected. I don’t have a good explanation for this. We would have had fewer bench hitters available because we were so busy streaming pichers, it would have made sense that we would have had even fewer At Bats against the projection. You’ll also notice that Slugging Percentage was highly projectable as well.

You’ll notice that Runs, Home Runs, and RBIs went up from projected even with a dip in At Bats. Last year the offensive explosion in baseball was prevalent in our fantasy league too. The final notable observation is how bad we are at projecting Stolen Bases. For two years in a row that was the least projectable offensive category. Every year, proven stolen base players either lose playing time, are given a stop light by their team for one reason or another, or young speed comes into the league that we didn’t see coming. But also, analytics are showing how low in value a stolen base is to winning a game relative to the value of the person being on base able to be driven in, so that has driven down SB totals the last two years too.

So what is the takeaway from all this? Projections are very helpful when it comes to projecting how good a player will be as seen from their efficiency stats: Batting Average, Slugging Percentage, ERA, and WHIP. From our past research we know that the per At Bat and per Innings Pitched stats for the couting stats carry similar projectability (home runs per at bat, strikeouts per innings pitched, etc). So, when you’re figuring out who to draft, take a look at their At Bat and Innings Pitched projections with a questioning eye while you bank their ratios projected and you’ll be set up for success on draft day. But also, don’t get discouraged if you come out projected lowly on draft day, because there's nowhere to go but up.