Best Puck Manifesto - Volume 2

Welcome back to part two of the Best Puck Manifesto, focusing on the fantasy hockey season-long best ball tournament on Underdog Fantasy. If you haven’t yet, now’s a good time to check out Volume 1 and subscribe to the newsletter for immediate updates whenever I post a new piece!

Let’s pick up where we left off.

The same two suggestions from V1 if you’re looking to improve your game:

1. Hop into the Morning Skate Podcast Discord, with lots of discussion and hundreds if not thousands more of the 15k Best Puck Classic entries hanging out in one spot. It’s the place to be to discuss the latest news, especially when we get into training camp, and is the best way to get a hold of us outside of streaming hours.

2. Which brings us to the streams - which will primarily be hosted on DJ’s YouTube channel, with a weekly audio version going up on the MSP podcast feed up through the beginning of the season. Sub to both so you don’t miss a second! There’s lots of relevant discussion about structure and strategy in last year’s content as well, helpfully labeled with UD or Underdog in the title.

Best Puck Manifesto, pt. 2: Manipulating ADP to Gain an Advantage

We learned last time that there is a clear advantage to be gained by not falling behind the proverbial ADP 8-ball in net, by investing in the goaltending position when drafting with your picks. We also learned that based on ADP Capital, a custom metric I created to closely mimic the actual scoring and experience of the Best Puck draft landscape, the regular season and playoffs told rather different stories about how you should use your draft choices to fill out the C and W positions, with the regular season advancement rates dominated by Ws and playoff success largely driven by high-investment C teams.

But where you draft a player is only a part of the battle, of course. After all, if you spend the first selection on Ryan Suter, rather than Connor McDavid, Nathan MacKinnon, or (controversially?) Auston Matthews, you’re clearly not taking advantage of your ADP Capital, or the value that your picks inherently possess.

ADP Capital Refresher

Much like the actual NFL or NHL draft, the delta, or slope of the change, between picks is not linear - the elite talent separates at a much greater level, thus fitting a curve takes a bit more precision:

While it doesn’t really matter what we use based on the best-fit pattern aligning whether we use just reg season, playoffs, or full year data, I have implemented the blue curve, or total points for the full season, to assign each individual draft pick an ADP Capital value:

draft_capital = 
945.3161 + log(overall_pick_number) * -148.3622

For simplicity, all ADP Capital values are rounded to the nearest whole number. For those re-creating in Excel, note that this is equivalent to the “ln()” function, the natural log rather than the “log()” function in Excel, which is the base-10 log.

Finally, here’s a grab of the ranges for each of the buckets outlined last week, if that’s of interest to anyone in conjunction with the above formula.

Updating Expectations for ADP Analysis

As noted above, one possible flaw in this sort of analysis is that every draft offers up a multitude of decisions to make with each listed pick. If I’m in a lobby with eleven aliens from outer space who love Underdog and the color teal, maybe they’ll draft 22 Sharks players in the first two rounds and leave me to scoop Connor McDavid and David Pastrnak. This would clearly “break” the draft capital model, which is based on the actual pick, and not where the player should go based on current ADP, which you can find in the applet while drafting.

This screenshot of a recent draft of mine reflects a common sentiment: “why should I be credited with the 52nd pick in this analysis when I got a player who based on the field should go at pick 42?”

Let’s apply this to all drafts, but ignore positional allocation (as in theory, value is value is value). How much “value” can you realistically gain in a draft, based on 23-24 data?

Based on the capital metric, those in the top bucket average 153 “points” gained per draft. This is the equivalent of moving from pick 1.08 to 1.03 (without the subsequent cost of reversing the order at the 1-2 turn!) or your 16th rounder into a late 6th round selection. Likewise, those at the very bottom of the scale lose 251 points, the equivalent of burning an early 10th round selection entirely or taking 1.08 and dropping all the way to pick 24, or 2.12.

These are not draft-defining differences, of course, but every edge helps. Let’s take a look at how these groupings fared last year:

Clearly, you can gain an edge by simply drafting players that slide past their ADP - you can compare the two graphs above and see that while the first four buckets capture teams who in-aggregate beat their draft slot, the first six buckets all post above average advancement rates - though slightly weaker as you progress from 1 to 6. Players who go entirely off the reservation, on average, lose in dramatic fashion.

This advantage doesn’t only exist in the regular season, it also translates to playoffs, via the same process for estimating playoff expected value that I outlined last week:

Even after accounting for the teams that actually make the playoffs (showing they must have done something right when drafting), the teams that gobble up the draft’s fallers, or at least aren’t constantly reaching for selections, perform much better. Interestingly, the fifth bucket performs slightly better, which is basically “clicking on guys at ADP” based on the spread of results. I theorize this might be where elements such as stacking and correct positional allocation come into play - while you’re allowing the room to come to you, you’re taking the initiative to keep your team on track for the most part by completing mini stacks or ensuring that you don’t fall too far behind at W or G, for example.

Putting this all together, here is the modeled impact on full-season Best Puck results based on how much value you get in your draft room:

We see far better results for teams that capture ADP value in their draft rooms than those that do not. The difference is a team-based value of ~$11.50 for the top two buckets (keeping in mind that the entry fee in 23-24 was $10, and the total payout per-team was ~$8.57 after rake and a small portion of the prize pool that was mistakenly allocated to non-playoff teams) and equity as low as $5.61 and $3.85 for the bottom two buckets!

Closing Line ADP

To this point, we’ve focused only on dynamics within your specific draft room. The picks you deploy and how you fare relative to current ADP on those selections both have varied impacts on your results. But the next point of consideration is that, in theory, ADP gets sharper as more drafts are completed and the season draws nearer. There is more information in the market, both from a player/player projection standpoint (you’re getting more and more opinions on how players compare to one another from drafters, from content providers, and even from projection sources) and from a non-fantasy information standpoint (things like injury clarity, role and deployment, or coach-speak), all of which will refine ADP to drive improvement as time progresses.

As a matter of fact, it seems quite noteworthy that three of the top 18 drafts in terms of generating Closing Line Value (CLV) on ADP advanced all the way to the Best Puck Finals! Sorting by the same “current ADP” value metric as used above, it takes us until the 92nd best draft (out of 14,100) to find one that made the playoffs.

So, in theory and based on these anecdotes, CLV is more important than within-draft ADP gains. Let’s compare:

CLV ADP is slightly easier to gain/lose than within-draft ADP, with both the high and low ends expanding about 30 capital points beyond the range of the current ADP chart from above. Instead of climbing from 1.08 to 1.03 as we did before, you move from about 1.10 to 1.03 in this case.

26% advancement (expectation of 16.7%) is quite good, and the pattern decreases as you lose more value far more dramatically than with current ADP. That’s a good signal that we should prioritize closing value!

Once you make the playoffs, similar to current ADP, closing line ADP tells a major story. You simply benefit from having combinations of good players (with strong ADPs) that the lower buckets don’t have access to, since the good players they did draft (which obviously helped them make the playoffs!) came at a greater cost, and a greater opportunity cost.

Put this all together, and we see a dramatic story based on Total EV:

Your $10 entry in Bucket 1 of CLV is worth $15, and you’ve turned your $10 into a paltry $2.31 based on these results if you fall into the 10th and final bucket. Well then.

Let’s take a step back and compare this to NFL, where a very similar process yields the following results:

Multiplying my numbers by 2.5 to reflect a $10 vs. $25 entry fee in 2023-24 contests, NHL looks like Bucket 1 is worth ~ $37.50 and Bucket 10 ~ $6.00.

As much as CLV matters in NFL, it matters a little bit more in NHL, the bulk of the difference coming from the increased playoff equity that NHL CLV offered, a phenomenon we do not see in NFL data.

Checking CLV for Unintended Correlations

And, well, this is a bit unnerving, personally. I believe it does make sense that playoff weeks, especially in a league with projectable superstardom and less sensitivity to game script than NFL, accelerate the advantage of ADP Value as you meet other teams in your playoff pods with the “league winners” that helped you advance in the first place. On the other hand, not much is changing from the dead of the summer to opening night. In Best Puck, we’re only drafting the 4-5 best players, at most, on a given team, and the vast majority of players that enter our draft consideration are virtually locked into L1, PP1 roles. Injuries in the preseason are far less likely, especially compared to the NFL where season-ending injuries occur quite regularly and dramatically shift the fantasy landscape. No such environment exists in NHL, so we shouldn’t expect CLV to have such a dramatic impact, right?

In any case, it’s undeniable that only a certain subset of players will offer you CLV, and as Leone notes in the BBM Manifesto, we don’t control CLV nearly as much as other facets of our draft capital allocation or elements yet to be covered such as stacking.

So let’s run some checks on our assumptions that this CLV analysis is picking up on the value of CLV, and not, say, the value of a certain grouping of players who just so happened to offer a lot of CLV.

Starting with the top risers, we find that a number of league-winner types show up at the top of the list. Of the nine players with a rise in CLV ADP of at least 100 ADP capital points, seven were positive contributors and six were skeleton keys to advancement. Only Timo Meier and Tristan Jarry of this group were considered disappointments.

Looking at last season’s true league winners, there’s a definite pattern in that only three of these players fell in ADP value, and all three were centers who fell as the field sharpened up to the C/W disparity investigated in part 1. There were no players who were falling due to anything to do with their on-ice play, while the rest of these players generally trended upward throughout draft season!

In general, the field drafted quite well, accelerating their selections of the eventual league winners at a much higher rate than other players.

Clearly, having access to these eventual success stories before they rose to peak ADP allowed you to fit in more studs with those strong performers, a positive feedback loop that led to the boosted equity amounts we viewed earlier.

It’s worth considering what the causation of this trend is. Ultimately, what you should take away from this section needs to be strongly informed by your belief in “our” ability to predict player success better than the field.

Meaning, does a sharp portion of the field have a substantial edge in player takes? 

If so, then this feedback loop is entirely expected, with the strong player takes driving both CLV and financial return in a way that completely disconnects CLV from having a provable link to profitability. If all the best players are properly priced, then there is no further benefit to achieving CLV value.

Alternatively, the more random these performances on a player-level become, the less these player takes matter. The fact that 23-24 just so happened to have so many late risers become the Players You Had to Have simply necessitated that those who drafted these players early, before they had risen, had a substantial advantage. In this case, CLV does drive the financial reward, and there’s more skill involved in predicting where the steam will go than which players will be successful for fantasy scoring. If we have no control over player performance, then holding a portfolio of players at the cheapest prices in a myriad of combinations will drive the best results.

Final Takeaways

Where I stand on this matter is that the player takes matter the most. I don’t think that’s a 100% certain fact, however. I believe that while 23-24 might have been a bit of an outlier in that what we expected to happen largely did happen, particularly among the players who were grossly mispriced to begin draft season, we should not be relying on market movement to make our decisions. Rather, I will advocate for dialing in on each player’s talent, circumstances, and certainty of role to unpack who we should be overweight and underweight on when drafting as a part of a well-balanced portfolio of teams with proper construction.

To crystalize my stance on this matter, here’s a scatterplot of every player representing a dot plotted by their Indexed Equity Value (0 being perfectly average) on the Y and their rise/fall in ADP Capital on the X:

While we reviewed the very high end and found tremendous correlation, in the aggregate there’s a very slight relationship between CLV rise/fall and their ultimate profitability. While CLV does matter, I believe the specific season of 23-24 might overstate its importance in a dramatic fashion by tying together some of these player-driven results and presenting them as CLV-driven.

Next up, in volume 3, we’ll unpack stacking dynamics in Best Puck before spending the rest of August and September diving into 2024-25 and players I am heavily invested in, those I am fading, and a multitude of other insights to help chase the $25k top prize on Underdog.

In the meantime, make sure you’re popping in the Discord from time to time, where I’m dropping various bits and pieces that don’t make it into the newsletter. Here are a few examples from the last week or so:

Thanks for reading! Subscribe to the newsletter and stay tuned to Discord and my Xwitter to make sure you don’t miss a thing!