Week 5 NHL Betting Report

Week five of the NHL season has been logged into the history books and contained my best output to date, sharing my picks for every single game on this blog or social media, producing my pick graphics for nearly all of them (minus maybe 2-3). Week 4 was my worst, but that did not discourage me from sharing wagers, perhaps discouraging you from tailing. The reason to continue sharing all my picks for every game is the joy of reporting effective decision making, even when acting against the advice of my models. My over/under picks went 26-17-8, almost entirely tailing my Grand Aggregator, or at least largest model position (GA doesn’t count other aggregators or itself, but the All 20 models you see in pick graphics do).

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. However, my sister has begun betting hockey based on my advice and I’ll be sharing all those picks with you.

• My Weekly Profit: $1,820

• My Season Profit: -$1,458

This weekend I built a worksheet with all my sister’s bets, now up to 131 non-parlay wagers totaling $108 and has won $98 for a $0.90 return per dollar. My return per dollar for the full season is $0.96 with most of those picks shared either on the blog or social media. Ergo: she would have been better off tailing all my picks instead of being recommended favorite wagers…so that’s what we’ve started doing, with bet sizes ranging from 25 cents to a dollar. Those will get scaled up when I’m able to sustain positive returns and recover the lost balance. I know my disclaimer just said these picks aren’t being made with real money, but my sister has tiny wagers on nearly all of them now.

One breaking development teased on social media (Bluesky activity has increased substantially since Tuesday) is the creation of version 1 of some line movement models, even shared a bunch of picks one night (but none since). That project was paused for the weekend to work on betting and fantasy reports, but I’ll be diving back into them after this gets posted. There are 8 models total that sort old games by % similarity of the opening and closing line. In the past when lines get pushed from there to there, what’s the best side to bet at close (noting my historical data isn’t always the precise close, but always at least 24 hours after the opening was recorded).

There is still some work to do before these are polished, functional models. I need to retroactively apply these to games from the last 2 weeks to see how each performs. Half bet ML/PL, half bet over/under, half bet previous 2 years, half only current schedule, half use the whole league, half only look at games involving those teams. The downside is that their picks are all last minute, so once this gets going you need to check my feed for any last-minute picks prior to puck drop. I’d share the picks for the Sunday games about to start, but the macro that sums the historical data needs to be rebuilt because I added some columns to the worksheet. Perhaps I’ll Tweet some of the picks later once repairs are made.

Week 5 Results

If you bet $100 on every underdog moneyline this week, you lost $1,209, or $1,236 on the puckline +1.5 goals. I put a little too much into dogs +1.5 goals this week, but pulled a nice profit betting both dogs and favorites -1.5 goals. Though my best category was betting against back-to-backs -1.5 goals, so that was the primary driver of my puckline -1.5 success. Whereas under 6.5 goals was the primary driver of my over/under success, with 47% of games opening at 6.5 (which is high relative to early weeks). My models continue their strong performance on visitors -1.5 goals, which is good because they love that bet. Habits likely to change in Q2, we’ll see how much.

*Note* “Overall Market Bets” based on betting exactly $100 on every outcome.

The best 3 teams to bet this week were Buffalo, Winnipeg, and Los Angeles, none of whom appear on my best teams list, but two do appear on my worst teams to bet against list. You may be confused why I’m betting a 14-1 team to lose, and it’s because of my value-hedging models, who otherwise are my best performing. This was explained in a Tweet, but if a line is -110 but should be -280 based on records, the hedgers bet the other team assuming oddsmakers had a valid reason to discount the line. The Jets best start in NHL history is making those inputs skewed wildly in a manner rarely observed. Those inputs get borrowed by other models for individual angles. That’s why models hate the Jets. They have to regress eventually, right?

My Team of the Week: New York Rangers, $1,036

The New York Rangers went 2-1 this week and I went 5-1 betting their outcomes, missing an under by 1 goal. Though I’d be surprised if any of you tailed my Sabres surprise 5-1 upset because I had talked myself out of the pick while doing my analysis, but decided to keep my initial pick to save myself the 3 minutes required to make a new graphic. It was my betting models who recommended Buffalo, but I was too cowardly to tail them -1.5 goals at +375. That game might have made the biggest contribution to their positive balance betting dogs -1.5. Igor Shesterkin has a .920 SV% on the season and Jonathan Quick is .964. So don’t get too excited if the back-up starts unless you’re betting Rangers.

The Edmonton Oilers were my second best team of the week, pulling $425 profit betting them to win, and $425 betting their opponents. Vegas was my pick in one of those, logged prior to learning of McDavid’s surprise early return, but the Knights still managed to squeak out a victory. They played well enough in the absence of McDavid that it feels like they’re ready to catch fire, at least that was the justification in my Tweet for Oilers -1.5 goals at +212 vs Vancouver on Saturday. Sadly, I also had the under, but only with a minimum wager and still walked away from the game with a healthy profit (my sister abstained from the under and fared better than me).

My Worst Team of the Week: Tampa Bay Lightning, -$620

The Tampa Bay Lightning went 0-3 this week, all 3 of which I picked them to win. Star goalie Andrei Vasilevskiy is sporting a .938 SV% in his last 6 games, but the team only won two. They are 7-7 overall, 4-2 at home, 3-5 on the road, so it might be time to resurrect the old home/road strategy and stop relying on models for choosing their outcomes, especially models with a lust for visitors -1.5 goals, at least in the first quarter. Several active angles get closed when the quarter wraps, but some will still hammer V-1.5 in later quarters if history dictated it was profitable. St. Louis retains last place in my profit ranks, but Tampa is not far behind.

The Seattle Kraken were my second worst team of the week, sinking them from 21st to 30th in my team rankings. I’m posting a large loss betting them to win, lose, over, and under. Some of this is likely losing to bad teams and beating good teams, given that I’m still undecided if this team is good. One solid vector would be betting Seattle opponents and overs in Grubauer starts (only 1 win in 6 games with an .877 SV%) but he’s currently day-to-day with injury and Daccord looks to be winning primary duties. Rather than adjust my own Kraken-specific strategy, I’ll just hope my models improve recommending their outcomes as they get more data.

Team By Team Profitability Rankings

These profitability rankings are based on the sum of all my bets per team, including where the money was won or lost. Each week my new Profitability Rankings will be based on all the games in the season, not just what happened this week.

Betting Models Results

There has been some movement atop my model leaderboards, with Shorting Goalies earning top seed, both versions (ML/PL and OU). Technically my over/under Grand Aggregator has a better return per dollar and is still the one I’m tailing most often, but not “Hottest Model” picks shared beneath mine in graphics. Both of those are now Shorting Goalies, both “value hedgers” with a more detailed explanation below. I’m proud to see my own results higher up the page, as there’s really no excuse for underperforming my models that badly when they are informing my picks. I just needed a shift more towards pucklines -1.5 goals.

Top Model Week 5: Shorting Goalies

Shorting Goalies was my 2nd best model last season, though this is a much more advanced version. It uses my estimated probability of goalie 1 or goalie 2 starting the game for both teams, which then uses their stats in their last 5 games to estimate the probability of either team covering moneyline/pucklines, then subtracts the implied probability of the betting line. It’s one of my “value hedgers” who looks for negative value. This team should be -350 but the line is -110, which can deceive your mind into perceived value, when there tends to be a good reason to pick the opposite side. This is burning my models with the Jets, but they are lacking data for teams this hot. Jets are short-circuiting my value hedgers who don’t believe it’s real.

My best over/under model was “Last 5 Home/Road” only counting home games for the host and visa versa for visitor (there are several inputs used often by all the models that look only at home-road specific splits). Simply averaging total goals per game for both teams (last 5 GP), then subtracting the betting total. The issue with any models dependent on this home-road data is that some teams still haven’t played 5 road or home games, so it’s still not a complete sample, but should be by the start of the 2nd quarter, and all are were programed looking at quarterly data historically (which tends to have small Q1 samples). That’s also why I’m not attempting to repair broken models. They may repair themselves in Q2. L5HR has more quarterly splicing than many of the others.

Worst Model Week 5: Expected Goals: Home/Road

My 3 expected goals models were the 3 worst of my 20 ML/PL mods, and all 3 are in my 5 worst for the season. So, when you see my “5 worst models” in the pick graphics below, a majority of those are using expected goals as their primary input (but different versions). Corsi is still among the 5 worst, but did have a good week (after my report last week mused about creating a Shorting Corsi model that just bets the opposite). All 3 of the expected goals models are getting destroyed by the Jets, who does not have an expected goals ratio proportional to their win-loss record. Perhaps it’s time to be more mindful of where my worst models put their money. I share it in all the graphics.

My worst over/under model was “Team Profit Flow” (formerly known as Game Sum) just adds up the total profit betting various outcomes with this team, then makes picks based on which bet is the most profitable. It has quite a sizeable lead for last place, down -$8,500 fake dollars while next worst is -$4,000. It might need to be rebranded again into Shorting Profit Flow and pick the opposite of all its bets. I’ve noticed my Grand Aggregator often taking the side that has been less profitable in the last 30 days, and the struggles of this model suggest that what’s been better than simply chasing profit. Maybe profitability should not have out-sized influence on OU bet selection?

Tomorrow’s Picks

Here are my picks for tomorrow, with some parlay polls down at the bottom.

MTL @ BUF:

The Montreal Canadiens are in Buffalo to play the Sabres tomorrow who have won 3 in a row, outscoring the Rangers, Senators and Flames 14-4. My models have more total money on Montreal, including 4 of my 5 best models. I’m just not inclined to follow that advice because they are performing very poorly with visitors in Montreal road and Buffalo home games. The Habs have lost 6 in a row and might be the worst team in the league, so I’ll tail my largest model position, Sabres -1.5 goals. I’m a little concerned this might be Devon Levi (.878 SV%) in net for Buffalo, but it also might be Cayden Primeau (.860 SV%) for Montreal. That’s also why I’m taking over 6.5 goals when my models have 53% on the under. My top models have 63% on the over.

Warning: My worst models largest position is also Buffalo -1.5 goals.

SJ @ PHI:

The San Jose Sharks face New Jersey then travel to Philly for a battle on Monday. Samuel Ersson is back from injury and my likely projected starter tomorrow. 52% of my model money is on Flyers -1.5 goals, despite the expensive +105 price. Philly is only 2 PTS ahead in the standings, so without the back-to-back -265 would be a ridiculous price (fair line estimator says -147). There is an obvious tax on the back-to-back given both teams have won 4 of their last 10. I’m going to take over 6 goals, but my models are net losers betting both over and under with these teams at this total in the last 30 days. Taking the over gives me Fedotov insurance.

DAL @ PIT:

The Dallas Stars are 5-5 in their last 10 games and are 2-4 on the road, hence why my models and I are performing so poorly betting them as visitors in the last 30 days. They have 56% of their money on Stars -1.5 goals, so that’s going to be my pick too, despite not hitting a single V-1.5 in Dallas road or Pittsburgh home games in the last month (24% of my wagers were on V-1.5 in those matches). My return is less bad on VML, but I just had a good week on visitor pucklines, so will take the risk. I’m taking under 6.5 goals, where 65% of total model money is invested, but my best models were more divided.

LA @ CGY:

Warning: The LA Kings currently rank 29th in my profitability rankings, and I’m doing a terrible job betting their road games where they are 5-5. Given that I’ve taken all favorites -1.5 goals on this report this far, may as well stick with my largest model position (48% total). At least my models are performing better than me betting V-1.5 with these teams in the last 30 days. I’ve been going heavier on the moneyline in those matches, but that’s been a big loser overall. Also taking over 5.5, but with low confidence. Models not performing well with these teams when the total opens at 5.5 in the last 30 days.

NSH @ COL:

The Nashville Predators have lost 5 of their last 7 games, but are at least getting good goaltending from Juuse Saros. The Avs are have been slightly better, losing 4 of their last 7, but have won 2 of 3 and only lost 1-0 to Winnipeg in their defeat. Alexandar Georgiev has started the last 2 games and has a SV% above .900 in the last 2 weeks. That doesn’t mean he’s officially no longer slumping, but early indicators are suggesting it’s possible. That would hurt their overs and it just so happens my top 5 over/under models are 100% invested in under 6.5 goals. My largest model position is Preds -1.5 goals, certainly not a bet I have any confidence winning, but taking a swing with a small wager for a big payout.

CAR @ VEG:

The Carolina Hurricanes are 5-2 on the road, visiting Las Vegas where the Golden Knights are 8-0. The fair line estimator thinks Vegas should be a -180 favorite, not a +114 dog, but several of my models are also trained to look for “negative value”, presuming that odds makers have a valid reason to skew the line towards the other side. Of those 8 home wins, only 2 were against teams currently in a playoff spot (Calgary who is on their way down and LA). So, it’s been a soft home schedule for Vegas and possibly a rude awakening facing this beast from the east. That’s why I’ll tail my largest model position, Canes -1.5 goals, but only with a minimum wager out of respect for Vegas. My models also like under 6.5 goals but are performing terribly making that bet when these teams had an opening total at 6.5 in the last 30 days.

Best Parlay Poll

After my strong week, we are going to take another crack at a parlay poll. Click on the image to be taken to the poll.