In this week’s My Model Monday, I compare our 2018 NHL Season Simulation results to the current NHL futures odds to see where our model is differing from betting markets.
Our NHL future probabilities are generated from a Markov Chain Monte Carlo Simulation of the NHL regular season and playoffs which uses team statistics, Elo ratings, and other factors to predict individual game outcomes. After the simulations are complete, the results can be converted into various playoff probabilities for each team (e.g., Win Division, Make Stanley Cup, Win Stanley Cup). For example, if we run 10,000 season simulations and the Tampa Bay Lightning win the Stanley Cup 1,000 times, that gives them a 10% chance of winning the Stanley Cup (or ). NHL futures odds from betting markets can also be converted to implied probabilities for comparison to our simulation results. An easy way to convert betting odds to a probability is to use:
Implied Probability = Total Money Bet / Total Money Paid if Bet Wins
For example, the Tampa Bay Lightning are currently the favorites to win the Stanley Cup at +600 (or 6-1) and let’s say we bet 1 dollar on them. We can convert that to a probability by taking: (or 14%). Remember, if you win your bet, you get back the money that you risked (in this case, 1 dollar) which gets added to the denominator.
Now that everyone understands the math we’re using, here are the comparisons between our simulation probabilities (labeled “284”) and the betting market implied probabilities (labeled “Betting”):
Keep in mind that the betting market implied probabilities will be slightly higher than our probabilities since the sportsbooks are adding in a charge for you to make a bet (aka the “vig” or “juice”). For example, if you sum our probability to win the Stanley Cup across all teams, it equals 100%, whereas the same calculation for betting markets sums to 137%.
Vegas is the only team running away with their division right now, and our simulation agrees with betting markets on their 75% chance of winning the Pacific. Our model is much higher than betting markets on Nashville (-115 in current betting markets), Boston (+250), and Philadelphia (+500) to win their divisions. Conversely, our model is not as confident as the betting markets on Tampa Bay (-300), Toronto (+450), New Jersey (+500), Columbus (+600), San Jose (+750), Calgary (+800), or Los Angeles (+900) winning their divisions. Although Tampa Bay currently leads the Atlantic, our model actually makes Boston the favorite to be on top when the dust settles.
Make Stanley Cup
The differences between our probabilities and the betting markets on making/winning the Stanley Cup are much slimmer than the division odds. The only teams that our probabilities favor more than betting markets are Washington (+800) and Minnesota (+1400). On the negative side, our probabilities do not see any value in bets on Vegas (+300), Pittsburgh (+300), Boston (+400), New Jersey (+1200), or Anaheim (+1400).
Win Stanley Cup
Unfortunately, there is not a standout team to love here, but our probabilities are roughly in consensus with betting markets on Tampa Bay (+600), Vegas (+700), Nashville (+800), and Pittsburgh (+800) as having the best chance to win the Cup this year. Lastly, according to our simulation, you are better off lighting your money on fire than betting on New Jersey (+1600) or the NY Islanders (+2000) to win the cup.
Note: our simulation results are as of February 24, 2018, and will change as the NHL season progresses, as will the betting market odds.
3 thoughts on “My Model Monday: NHL Playoff Futures”
Comments are closed.