It’s that time of year, when the days are getting warmer and longer, the birds are chirping, and playoff hockey is ready to be played. This preview is looking at the first 6 series of round one, with picks for each opening game. Then on Sunday, there will be a part 2, discussing the remaining series, along with picks for the upcoming game twos. I also posted a 6,000 word preview last season that dissected betting results from the previous 2 playoffs (when home favorites -1.5 goals dominated) only to have visitors come storming back with a vengeance (including 6-2 in game ones). I’m not repeating the format, but did dissect that report for any useful information.
Before we go any further, it’s time for my regularly scheduled obligatory *DISCLAIMER* it needs be noted that I’m not betting with real money. These are all fictional wagers in a spreadsheet. My mission is to engage in a mass betting campaign, picking a winner of every single game, every over/under, because it provides a complete dataset for macroeconomic analysis, which can be shared with you, shedding light on what worked and what failed. I’m also tracking the results of betting every outcome, to help me (and you) uncover previously unknown or newly emerging profit vectors. What started as a thought experiment has evolved into much more.
My picks for Dallas/Vegas, LA/Edmonton are all done and will likely be mentioned here. I took a moment to fill out a bracket on NHL.com attempting to perfectly predict how the playoffs will unfold. One key note: I picked the Panthers in the bracket, but Tampa was my series pick. The only reason Tampa was my series pick was the +150 price being offered. If it was even money on both sides, Florida would be my pick, but it doesn’t matter because either team will lose to Toronto in the next round. The stupid page where you fill out brackets is formatted so you can’t zoom out and screenshot, so a top and bottom had to be “glued” together. The Dallas Stars emerged victorious.
My playoff spreadsheet from last year was updated with some new statistics, including standings and game logs for 2014 to 2023 (including expected goals data from Natural Stat Trick) excluding 2020 because the covid bubble data has very little use here. My database includes game-by-game betting lines/results from the last 3 playoffs, which contains 146 first round games. Just like last spring, I’m logging a pick for every series beforehand. Last season my record was 7-8, going 6-3 on favorites and 1-5 on underdogs. Had I bet $100 on each pick, I would have lost $298, returning $0.82 on the dollar. Whereas betting the individual games produced a 9% profit thanks mostly to Vegas, Edmonton, and Dallas.
Those were the wagons that pulled me into profitability (Edmonton was -220 on the series line, Vegas -170, Dallas -140), splitting my bets between moneyline and puckline in the individual games generated a far better return than the series line. My other challenge last season was picking teams to win the series in the correct number of games, but went 1 for 15, returning $0.66 cents on the dollar. The picking exact outcome challenge will be back on my plate again, but am not logging every permutation, just my picks. My only hit last season was Edmonton to beat LA in 6 games.
Best Bets 2021 to 2023
Last 3 Seasons
Last Season
Of all the round one categories, none produced a higher betting profit in 2021 & 2022 than home favorites -1.5 goals. Looking back at my First Round Betting Report last season, the dominant theme was favorites -1.5 goals (with home teams favored in more than 70% of games). Then visitors slapped me in the face and made all my diligent preparation look foolish. Frankly at this point, predicting which version of history will repeat is not a challenge I’m undertaking this season. My betting models select wagers based on profitability in the last 3 playoffs, so big picture will be reflected individual games, filtered through my models, using the stats that previously separated winners from losers.
In the first round of 2021 and 2022, betting $100 on every home favorite -1.5 goals, you banked $1,771; whereas the same wager on home fave moneyline only yielded $360. That all flipped in 2023, with favorites and home teams (which are often the same thing in playoffs) getting hammered by visitors. It’s going to be interesting with my models built using all 3 seasons. Last season I shared the 5 best bets from the previous 2 playoffs regardless of seeding, and 3 of them were good bets in 2023. Home favorite -1.5 goals in game two, overs in game three, road moneyline in game one. My new models definitely like road moneylines.
Game-by-Game Probability Calculator
One new feature added this spring is used by some of my new models to make selections, including one that uses nothing but this. It attempts to predict probability of victory for either side based on seeding differential, the game of the series, and high-seed win difference. This is completely untested on new games, and may not work at all, but the model I’m calling “the Oracle” will let me know how effective these are. The count you see below is the number of historical games used to estimate each probability (basically it is a count of each possible game state in a series).
The Nine Cup Wraiths
Nine new models were created that were built using only playoff games, but had dozens of input variables available to each depending on what had the biggest difference between winners and losers. These are all in my main spreadsheet, but a different worksheet. They’re not in the same tracking as my main models, so their updated performance data won’t be readily available when needed (whereas their bet amounts are included in pick graphics). I’m actually visually monitoring each of them game-by-game just out of curiosity, so I’ll know which are succeeding or failing. If any are terrible off the jump, they may get removed from future pick graphics.
1) Orthodox Zig Zag
Betting the loser of the previous game to win the next is a very popular playoff betting strategy, nicknamed “zig zagging”. This is a strategy I’ve attempted adopting to various degrees in previous playoffs, but it never proved to be a winning strategy. Oddsmakers are fully aware of the strategy and charge very expensive prices to ride that train. I had Chat GPT analyze my “game notes” playoff comments to differentiate between notable differences in my bet justifications on winners vs losers. Zig zag was mentioned disproportionately in the losing bets. That doesn’t mean I’m any less determined to make it work.
The Orthodox Zig Zag will never bet on a team that won the previous game, strictly adhering to the religious doctrine. There are a few instances when it will abstain, making it the only 1 of 9 with an abstention option. That being said, this could be considered metaphorically to be a scholar in “intelligent design”, using science to prove the creator’s righteousness. The design flaw here was the degree of difficulty milking profitable angles strictly adhering to doctrine. That’s why Ziggy needs to know when to fold’em. Still, the amount of tax that oddsmakers charge on the loser of the previous game limits your options.
This had 28 different angles, averaging 9.3 historical observations per group. That’s a big problem. This was the first model off the factory floor. It bet visitors moneyline in all 8 opening games. It never bets anything else in the first game of any series.
2) Next Gen Zig Zag
The so-called “next gen” zagger does not blindly follow the faith, but tries to lean in that direction wherever possible. It has the freedom to bet winners of the previous games in instances where it’s profitable to do so, but the creator maybe got a little too cute splicing out winning angles, finishing with 38 bins averaging 6.9 games per bin. Yeah, that’s a fun number, but less so with a decimal between the two. All that splicing wasn’t entirely necessary, but it did generate a whopping 63% return. This was by far the most profitable model on the historical sample, but possibly built on a house of cards. Let’s just see how this works before dismissing this model as a religious extremist.
This one also bet the visitor in every game one, but for a 1 or 2 seed, it bets minimum on V-1.5 goals, otherwise a max bet on visitor moneyline. Game ones, road teams are substantially more profitable than home teams.
3) Shorting Value 3.0
My beloved Shorting Value 2.0 model does have “playoff escape clauses” built into some compartments, which was done in every single instance when one of the angles lost money in playoffs. My crystal ball was already glaring into post-season when that retrofitting project was undertaken. That being said, the playoff specific version of this model only uses the last 5 road or home games each team has played against other playoff teams, whereas SV2.0 uses the last 10 home-road games and includes all opponents. At this stage in the overall construction project, sample size concern was top of my mind. This had 20 angles, averaging 13 games each. Shut up, that’s a lucky number.
For game one, this bet 8 home teams with a relatively even split between home ML, +1.5, and -1.5.
4) Game Sum 3.0
This is the exact same concept as my non-playoff Game Sum model, but that one does not have special playoff commands, nor was built specifically verifying playoff viability. No idea if that concept will translate, but we do have a version of that concept that does (using last 5 games instead of last 30 days). You just add up what’s been more profitable VML, V+1.5, V-1.5, HML, H+1.5, H-1.5 in the sample, then there are specific commands depending on what’s been hitting lately. Granted, it doesn’t always bet the positive value, it just takes the most profitable output based on the most profitable input. This had 18 angles and 14.5 average games in each.
This one also bet every home team in game one, and 7 of them were -1.5 goals. Half of its round one compartments are visitor wagers, so it was not program with home bias. It just worked out that way here, probably because home teams were very profitable in the last 2 weeks.
5) Expected Goals Macro Daddy
By this point, over/splicing was a major concern and minimalizing the cuts in the data was the primary objective of my newest expected goals model (which uses 5v5 instead of all strengths) and only has 5 compartments (3 are VML, 2 are H-1.5). It does not care about round or game. For the 8 game ones, it has 7 VMLs and 5 of those are max bets. The Macro Daddy will be big pimpin’ if VMLs repeat last year’s 6-2 record in opening games of the opening round. Aside from that, there is not much else to say about this one.
6) The Angle Minimalist
The title is what it does, except it does splice the data more often than Macro Daddy. The building process was simple, take all the big dogs, and cut it in half using the statistic with the biggest difference (in standard deviations) between winners and losers. Repeat for big favorites, mild faves, etc. for 8 total compartments with an average of 33 games in each. For the weekend, it has 3 VMLs and 5 home pucklines (it does not like home moneylines).
7) Betting Save Percentages
This model just takes the home and road teams save percentage (on the road or at home depending on venue) and has 6 compartments based on over or under .900 (example, home > .900, road <.900). The only potential flaw is that it’s heavily skewed to home teams (80% of the games within picked home team), so if it’s another dominant year for visitors, this one will struggle. Though it did split game ones 4-4 between V-H.
8) The Oracle
This is the model that bets entirely based on the aforementioned game-by-game probability calculator, with 8 different compartments (5 betting home, 3 road). There is not much to say aside from that, because the efficacy will be entirely based on the accuracy of the calculator. If the calc proves useless, then so too will this model be garbage.
9) Betting Expected Win %
The final “model” on the new squad is similar to the last one, but fewer angles and a different input. Instead of an estimated win probability, it uses the last 5 road games of the visitor and last 5 home games for the host, and uses that to measure win probability.
9B) Expected Goals Value Voodoo
Yes, there is a 10th member of the team, but this one is not a “model” rather a staggered algorithm. Basically, I took expected GF% and subtracted the implied probability of the ML. Then I sorted them from largest to smallest looking for a pattern. There was a clear pattern flipping back and forth between home-road every so often, which was quantified and weaponized, but I have no idea why that worked or if it’s replicable. This is purely an experiment and does not hold membership in the 9.
Over/Under
There is not too much enthusiasm in the over/under section of this preview given my advisory team struggling in the last few weeks. To make matters worse, average goals in the playoffs is much smaller than regular season, most notably because fewer penalties tend to be called. So, my existing models that are still using regular season data have potentially flawed inputs. While my 4 new OU models (which I’m not discussing quite yet) are built on very small sample sizes (with a concerned effort to minimize angles on my part). I’m sharing over/under picks for every game, but my confidence is low. Tail with caution. Below is over/under betting stats from round one of the last 2 playoffs.
Overs went 55-42 and was easily the best bet in both playoffs combined. In the first 8 games, I have 7 overs (Isles/Canes under 5.5 is my only under). You can actually go vote on your favorite over if you browse my Twitter history. Tampa vs Florida and Colorado vs Winnipeg are currently the two leaders in the poll.
New York Islanders at Carolina Hurricanes
It should be disclaimed from the onset of this analysis that I posted a large loss betting all the games between these teams in the regular season, betting the home team too often when visitors prevailed. It did come as a surprise to see my pick was the Islanders in 3 of 4 games, but it was betting them to win on home ice that did the most damage. There was not a moment when I even considered making the Islanders my series pick, but my models may convince me to bet NYI for some individual games depending on circumstances. 3 of their 4 games comfortably hit the over. My only concern being the Isles have a brutal penalty kill, but they tend to call fewer penalties in the playoffs.
Canes are -350 to win the series and I’m picking them to win in exactly 6 games at +350. Having said that, looking at the game one line +184 feels like a nice number for a team that’s won 8 of their last 10 games. There have been 9 “big dogs” in game one, round one, of the last 3 playoffs, and betting $100 on each (all visitors obviously) yielded $420 of profit. I’m almost talking myself into Isles game one before even consulting my models (pause to consult models) but they are 68% on Carolina. My top models prefer moneyline, the Nine prefer puckline -1.5 goals. Given my return is higher on HMLs than H-1.5 lately, that will be my pick at -225. I do think there’s some value on NYI and would not object if you want to take a swing at a big payout.
One note, Carolina -350 currently has zero votes in one of my “which of my series picks would you rather bet” polls. As previously mentioned, this is my only under bet of all the game ones.
Toronto Maple Leafs at Boston Bruins
It should be disclaimed from the onset that I’m historically a big loser betting Toronto to win in round one of the playoffs, and went 1-3 picking their season series. The primary reason for those struggles was betting Toronto to win too often, despite noting in my betting write-ups that Boston owns Toronto. I’m the guy at the roulette table who sees black come up a dozen times and starts hammering red because dammit, red is due. The “game” we’re playing here is far more complex than roulette, but that was my justification for why I’m hammering blue and doubling down. Leafs are +105 to win the series, +520 to win in 7, and +110 to win game one (that has moved to +105 too). Those are my picks.
I’ll admit that come playoff time, I’ll superstitiously bet Leafs to win because they are my least favorite team and have a history of performing poorly when they are my pick (both regular season and playoffs). That’s why they are my Stanley Cup pick every New Year’s Eve, though in my bracket I only had them advancing to the conference final. But all superstition and reverse jinx attempts notwithstanding, the Leafs solicited quite a lot of investment from my betting models, at least the 9 newbies. My traditional models preferred Boston. Also, the Leaf bets in my current polling are soliciting your interest as well (at least the dozen of you who vote in my polls). I’m taking over 5.5 goals, but the under did attract significant interest.
Tampa Bay Lightning at Florida Panthers
The Florida Panthers are a -180 favorite to defeat cross-state rivals Tampa and had you asked me a month ago for my pick in a Florida-Tampa match-up, the answer would have been Florida with little hesitation. But the way both teams played down the stretch has me feeling uneasy about paying that price. Looking at the season series, road teams went 3-0, which is bizarre considering both are traditionally better on home ice. That does cast some doubt on “home ice advantage”, but my models don’t seem terribly concerned. Their lack of concern does not alleviate mine, especially considering my success betting Tampa lately (currently #7 in my full-season profit ranks).
The choice my models presented me was either Florida -1.5 goals, or Tampa ML. Sadly, neither them or me has performed well betting home -1.5 goals in Panther home games or Tampa road games the last 30 days. The fact that I’ve mostly laid off Panthers H-1.5 lately tells me plenty about that pick. May as well just chase the higher payout and bet the road team. If I’m picking Tampa to win game one on the road, may as well take them +150 to win the series (let’s say 7 games at +630). If that was even odds, Florida was my pick, hence why they’re my pick in the bracket. This over/under opened at 6 and I had the over. It moved to 5.5, so I’m raising my bet size on the over. My new playoff OU models LOVE over 5.5.
Washington Capitals at New York Rangers
The Washington Capitals won some big games in the final week of the season to make the playoffs, but they were garbage before that. Rangers are 100% my pick, both in game one and the series. I’ll even pay -450 for a small return. As a Red Wings fan, I’m disappointed this is not Detroit here, but my plan if the Wings had made the playoffs was to bet the Rangers to win in a sweep. They didn’t deserve to be here and frankly neither does Washington. That’s why I’m picking another Rangers sweep, but the reason is because I’m getting +400, where Rangers in 5 is +250. I’m hoping to squeeze a little extra juice from this match.
That previous paragraph was written before even consulting my betting models about game one. Nothing in the data they provided did anything to diminish my Rangers enthusiasm. Granted, the Rangers have not done much to charge that enthusiasm in the last month and I don’t think this team will make it past the next round. They are however doing well in early voting for a potential weekend parlay, but that could be the result of Washington negativity. For over/under, I’m picking over 5.5, which did not have the support of my Small Council. If Shesterkin shows up and Charlie Lindgren continues his strong play from the last few games, I may regret this pick later.
Colorado Avalanche at Winnipeg Jets
Let me disclaim from the onset, I’ll 110% be cheering for the Winnipeg Jets in this match-up. My own personal bias is strongly leaning towards Jets and Colorado’s road struggles (relative to home dominance) is a big red flag, but my gut is telling me that Colorado will prevail. Yes, Hellebuyck is the great equalizer and the Jets also play a grinding playoff style of hockey, but the Avs also have plenty of grinders and have that recent Stanley Cup winning experience. That’s why I’m picking Avs to win the series (at -130) in 7 games (at +500), including the first game as a -105 underdog. Though they have moved from -130 to -135 on the series line, so the public is voting with their wallets.
For game one, my betting models actually had 64% of their money on the Jets, with my existing models disproportionately taking Jets (2/3 of that -1.5 goals). The Avs lured several of my Cupwraiths to their moneyline, but my mind was already made up before seeing that. Yes, the Jets dominated the season series, the Avs have big question marks in goal (the Jets strongest position), so this is by no means a slam dunk for Colorado. If you do decide that’s your chosen path, maybe don’t get nuts with the bet size. My confidence is higher in over 5.5 goals, which was the most beloved over for my large team of advisory models/algorithms. Jets overs were shockingly good down the stretch and Avs goaltending is a concern.
Nashville Predators at Vancouver Canucks
The Vancouver Canucks had a miraculous first half, while the Nashville Predators were equally astounding in the 2nd half. All Vancouver games are on my television, so this is the team I’ve watched the most and much like the Jets, I’ll be cheering for them to win. My betting models strongly prefer Vancouver, both the existing models and the Nine. 57% of all model money is on Vancouver -1.5 goals. I performed very poorly betting VMLs in Nashville road games and Vancouver home games in the last 30 days. All signs point to Vancouver winning this series, yet I’m picking Nashville to win the series at +125, in 7 games at +590. That’s more of a value play, but close enough to take Nashville in the bracket too.
The motivation driving my decision to ignore betting model advice (at least for game one, none of my models predict series) was more a product of big picture reflection. In the last 17 match-ups of #2 vs #7, the first wildcard has won 9, more than 50%. It’s not a coincidence that I’m also taking Tampa, which is also directly correlated to historical 7 upsetting 2s. Granted, the #2 seed has won 3 of the last 4, with the only exception being last season when Seattle upset Colorado. Is it smart to bet based on that alone? Maybe not, but I’m getting +125 to win the series. They are +126 in game one, so I’ll take them there too, along with over 5.5 goals. Doesn’t feel like Demko had enough time to get back in the grove.
Weekend Parlay combinations
There are a few different polls on Twitter right now, so please go vote!
My Picks
Carolina ML, Toronto ML, Tampa ML, which currently parlay to +612. Add Rangers -1.5 goals for +1466.
This is currently being decided in 2 polls (the order above based on early polling) which will have a final posted by 9pm Pacific time. Or just vote who is leading now.
Model Favorite Underdogs
Nashville ML, Toronto ML, Colorado ML, which currently parlay to +784. Add Tampa ML for +2040.
This is based on current polling order. Check Twitter tomorrow morning for updated results.
Model Favorite Home Teams -1.5 Goals
Rangers -1.5 PL, Winnipeg -1.5 PL, Vancouver -1.5 PL, which currently parlay to +1830. Add Florida-1.5 PL for +4803.
This is based on current polling order. Check Twitter tomorrow morning for updated results.
My Best Series Picks
Colorado -135, Toronto +105, Tampa +150 which currently parlay to +792. Add Nashville +130 for +1952.
Currently ordered by Twitter polling, check back later for updated results.