Game Predictions – 3/22/2006

I made some minor changes into the output of my game prediction algorithm which hopefully will clear up some of the confusion surrounding the HTOdds column I had. I now have 2 columns listed, Fair HTOdds and Fair RTOdds and these represent what I would consider to be fair odds if you were to bet on one of these teams. For example, the Fair HTOdds for Buffalo is -188 so if I were looking to bet on Buffalo to win I would ideally be looking for odds better than -188. The odds for Buffalo at Bowmans is currently -135 so I would consider this a good risk/reward bet. Conversely, the fair odds for Carolina is +146 while the odds for Carolina at Bowmans is +115. This would not be a good risk/reward bet in my mind because +115 is not better odds than +146. Does this make it any clearer?

One betting strategy, which I think I might try, is to only bet the predicted winner if its odds are close to or better than the calculated fair odds. Otherwise, don’t bet the game at all. Tonight I am betting Buffalo (-135), NY Rangers (-150) and Anaheim (-125). I know Anaheim doesn’t strictly fit that since -125 is worse odds than fair odds of -113 but the odds are fairly close to fair and Anaheim is playing very good right now while Colorado has been pretty inconsistant. That said, Colorado will probably win and I’ll be mad at myself for making the bet.

Home Team Fair
Road Team Fair
Predicted Winner Confidence
Buffalo -188 Carolina 146 Buffalo Strong
NY Rangers -171 Philadelphia 141 NY Rangers Strong
Florida -164 Washington 139 Florida Good
Dallas -209 Minnesota 152 Dallas Strong
Anaheim -113 Colorado 111 Anaheim Some

This article has 21 Comments

  1. Looks good David. I like the adjustments.

    A note from yesterday’s thread:
    You mentioned that systems are better than picking on emotion, and while I agree with that, the best way (IMO) to do this is to combine a system with instinct.

    I track many of the same factors as your algorithms do but I also look at certain intangibles, like a coach being fired (LA) or a star player out (Tanguay). And, of course, it’s important to watch the games. See how a team is playing- if they’re tired, or troubled (like the Canucks have been).

    There’s also certain stats that come up that if tracked daily are probably meaningless, but on occassion have merit. Like tonight, Ryan Miller is 7-0-0 when starting a game for Buffalo after Biron started the previous game. Not a deciding factor, but again, something to consider with everything else. Good luck.

  2. I agree. I think it is important to know the limitations of the algorithm and make slight adjustments to it as you described. This is why I took Anaheim despite the odds being slightly worse than my calculated fair odds. The one shortcoming of my algorithm is it looks mostly at long term trends, not short term ones like coaching changes or injuries. But, in the case of injuries, I often believe that a team plays better, at least for a short period of time, when a key player is injured, and often the team eases up when a key player returns from injury.

  3. Interesting comment about injuries.

    I just also wish to add that I find the long-term trendiness of your system very valuable, since I often spend too much time thinking the short term issues that teams go through. I think your write ups before each set of predictions do a good job at balancing out the two.

  4. Here is an interesting stat for you.

    Teams that have won 2 games in a row have a .541 winning percentage (279-237) in their next game. Teams who have won 3 in a row have a .558 winning percentage (153-121) in their next game. Teams who have won 5 in a row have a .511 winning percentage (45- 43). When you consider that teams who win 5 games in a row are probably among the better teams in the league, a .511 winning percentage isn’t that great. Complacency seems to set in sometime after 3-4 games for many teams.

    In 2003-04 the numbers were substantially worse. The winning percentages were .534, .505 and .458 after 2, 3 and 5 game winning streaks. 2002-04 the numbers were .509, .478 and .500. These numbers are much lower than this year but my guess is that this is due to the lopsided schedule. I think this is probably because this year it is more difficult for weaker teams to upset good teams through clutching and grabbing.

  5. I should add that none of those games after a winning streak numbers for this year (.541, .558 and .511) are better than the home teams winning percentage which is .567 (581-444). You are better off taking the home team than the team who has won 5 in a row as odd as that sounds.

  6. dave:

    thanks for the respone yesterday. what type of general method are you using to calculate your theoretical odds? given your desire to eliminate emotion, i’m guessing its based on some type of system.

    therefore, have you done any backtesting to see how the system would do if you added this additional requirement (matching or beating your hypothetical odds)?

    one thing to note: adding this new requirement will significantly reduce the # of plays you have every day, although will likely increase the win%. however, depending on how the numbers work, it is obviously possible that a greater # of plays with a lower win% is overall more profitable.

    you obviously know all that, but wanted to add some value to the disucssion.


  7. If you look at my power rankings (I’ll post this weeks later today), I use that as a basis for calculating my fair odds, adjusted for other factors like home team, teams playing back to back games, etc.

    I haven’t done any back testing but would probably be a useful idea. I only need to find where I can get historical betting lines as well as some time to crunch the numbers.

    Yes, you could be more profitable by betting on a greater number of games with a lower win percentage, but if you are betting on a smaller number of games you could theoretically bet more money on each game. Each night instead of betting 8 games at $5 a game you could bet 4 games at $10 a game. In this case it is highly likely that if the 4 games has a higher win percentage than the 8 games that you will make more profit from the 4 games (average over time).

  8. Hey Dave. How sure are you on the Sabres? They’ve burned me twice recently but I like them tonight.


  9. I probably miss something but the sentence “You are better off taking the home team than the team who has won 5 in a row as odd as that sounds.” doesn’t seem to me proved by just the fact that home advantage is .567 and team wins 6th time in a row just .5111 of times.

    I suppose we need to compare odds to check which ‘strategy’ is better.

  10. John: I think Buffalo has a good chance of winning for a couple of reasons. First, they are at home. Second, Carolina played last night and since Gerber was in net I’d say there is a decent chance that Ward will get the start tonight. If he does, that would benefit Buffalo significantly. But then, I thought Buffalo would beat Atlanta but they weren’t even close to winning that game.

    K: Yes, we if we were betting we would need to see what the odds are to determine which team to bet on. But if we are just predicting a winner, on average it would make more sense to predict the home team than the team which has won their last 5 games. I am guessing that most people would instinctively choose the team with the 5 game winning streak though. That is the point I was trying to make.

  11. Personally i like Carolina tonight.. they are 2-0 vs Buffalo this season thus far. Carolina has also last 2 games in a row. Carolina is 10-4-2 after a loss… 33-8-5 on weekdays… Recchi has been cold last 6 games getting 0 points.. should snap out of that slump hopefully today. I also like Colorado tonight as well… knowing fully well Anaheim is playing very well and Tanguay is out with an injury. Colorado is also 2-0 vs Anheim this season. This game is just a hunch based on instinct. Basically i just dont like having all 5 home teams winning tonight. I like to have at least 1 away team winning. Dont get me wrong.. all home team could win tonight.. its just a guess.

    David.. i like your system for the most part dont get me wrong. Its a good guideline to help select winners if ones betting on NHL games…no doubt. But as we all know the system is not bullet proof. No system is obviously. To much is left to chance. Its sports…anything can happen.

    Certain things that i look for when betting is such things as… what goalie is starting? to me personally…if a back up is starting I always feel that teams chances of winning drop slightly. Yet its hard to know who is starting beacuse usually it is a game time announcment. I also like to play in factors such as the season series thus far, or a winning or losing streak. You were correct in saying when for example a team like Ottawa has lost say 4 game in a row… CHANCES are they wont lose again.. playing the odds. Also picking a team like Ottawa to win every night is bound to be incorrect.. no NHL team will ever be perfect. Ottawa loses just like any other team.. not as often but they do. Hard part is guessing which games they will lose.. guessing which games they will win is the easy part. Im not sure if your system plays these factors into effect.

    Predicting on stats and stats alone is fine, but usually as time has shown.. stats are not always correct due to factors in games no one can control.

    What I would love to see David is maybe having a column for your algorithm predictions and a seperate catergory for your own personally predictions. Whether they are the same or not. I notice at times your system picks a winner but in your analysis you stat youd pick the opposite team.. and at times you are correct..maybe your gut instinct is correct alot of times, as is the case for other teams. Just something to think the site for the most part.. keep up the good work.

  12. Mr. Hockey: I agree totally. I have never claimed that my system should be used without external consideration. The use of the backup goalie is a perfect example and some teams will be affected more significantly than others. When I was doing my experiment of betting $5 on every game I frequently pointed out that that probably wasn’t the ideal betting strategy. But, I will claim that it is an excellent basis for a betting strategy and deviations should only be made when a good reason is available to alter the bet.

    “Predicting on stats and stats alone is fine, but usually as time has shown..stats are not always correct due to factors in games no one can control”

    Yes, of course, otherwise I would have a 100% success rate. But a human predictor can’t account for those factors either. No one knows when a goalie is going to stand on his head and stop 38 of 39 shots preserving a 2-1 win for his team. No one knows when a goalie is going to have an off night. No one knows how a team might react to an in-game injury or some bad refereeing. No one can predict these things. A human can’t, nor can an algorithm. Again, it is all about maximizing the risk/reward ratio and not just saying “I don’t think all 5 home teams will win tonight so I have to take a road team”. You are right in saying that quite likely a road team or two will win tonight, but predicting which one(s) does is the real challenge and does the risk/reward (especially when betting) make sense to attempt to try to predict which one will.

    As for my predictions, I could make some but it is all about having the time to research all those extra factors and aside from the use of back up goalies or a select few key injuries I am not convinced that any of them hold a lot of weight in the grand scheme of things. But, when I have time and do find some games that quite likely go the other way I will definitely make those comments known in the analysis. And going forward I will probably make it known what bets I am making, when I do make bets, which should give you some extra information and insight into my thinking. Unfortunately this website doesn’t pay the bills so I have to do it on a ‘as time permits’ basis.

  13. Yes David I totally agree with you on many points. The reason I say that I dont think all 5 home teams will win tonight is based on a few factors. I play Proline very very often and at times I have a hunch on certain games but I dont follow thru with it and thus later that night I found out my instinct was correct. That bothers me most. I notice that its always games that I dont think twice about that I tend to lose. Like the “sure bets” you know what I mean. In recent months I have completely changed my strategy around in a big way… it works for me..ive hit a few payouts on Proline following this guide.

    -I make my selections, and before I send in the ticket.. i make a few “last second” changes, based on a hunch or instinct or human factor. It does burn me at times but for the most part Ive been pretty successful with it. Perect example if you track back on this site to last Tuesday March 14. 2006.. i initially made my predictions than right before game time I came back on this site and changed a few games I was thinking about… as it showed, the changes I made worked out great because i won $440.91 on Proline Pools 8/8 that night. I only use this example because its actually proof I did make my last second chances. It was merily fluke I hit 8/8 of course, but never the less. Its not a great system but I have to go with whats been working for me personally. Before this change to my system.. i was losing Proline night in and night out. So the system works for me. Whatever works for you. And I totally agree with limiting your risk, but than again… sometimes a good risk pays out huge in your favour. Those are the best days.

  14. Also, I noticed that ever since you said your algorithm has included teams playing back to back games you are yet to pick a pick who has played back to back nights to win. Why is that?

    NHL Records of the 2nd night of a back to back game:

    Anaheim 6-3-1
    Atlanta 8-6-0
    Boston 4-6-3
    Buffalo 7-4-1
    Calgary 4-4-2
    Carolina 4-5-2
    Chicago 3-9-1
    Colorado 4-7-1
    Columbus 6-9-0
    Dallas 9-3-1
    Detroit 9-2-2
    Edmonton 4-7-0
    Florida 4-8-3
    Los Angeles 5-5-1
    Minnesota 4-8-1
    Montreal 5-5-1
    Nashville 8-3-3
    New Jersey 3-9-1
    NY Isanders 4-3-0
    NY Rangers 6-1-2
    Ottawa 8-3-1
    Philadelphia 7-7-0
    Phoenix 5-4-2
    Pittsburgh 4-9-2
    San Jose 4-4-2
    St. Louis 1-12-3
    Tampa Bay 8-6-3
    Toronto 11-4-1
    Vancouver 3-7-0
    Washington 3-9-1

  15. Again..last second changes to my predictions before I hand in my Proline Pools ticket:

    -Buffalo (Carolina has Ward, Weight out with injuries, and playing back to back nights) BOXED GAME
    -Florida (6-0 vs Washington this season)
    -Philadelphia (Series tied 3-3, away team winning all games) BOXED GAME
    -Dallas (1st game vs Minnesota in Dallas this series)
    -Anaheim (Ducks are playing very well as of late)

  16. Strong favorite: FLO
    Good favorite: BUF, NYR, DAL, ANA

    Yesterday was a good day for me winning $2.99. 6 of 10 win in 60 min.

  17. Mr Hockey, see, what is the day in the hell for Dave’s algorithm is one in the paradise for mine and contra versa.

  18. haha i totally did not understand that last comment Keif.

    Note: I BOXED PHI/NYR and COL/ANA tonight with BUFF, DAL, and FLA winning

  19. I have no luck with Buffalo, every time it seems they should win, they lose; then when I bet against them they win, I think I may start to stay away from them for awhile. I was pretty confident on Philly to win tonight considering their record at Msg Lately, though with Jagr you never know what could happen. Also Fla was a sure bet to win in my opinion.

    Also, I am seeing people complaining about little things on this site. I think its pretty good considering its free and they do a pretty good job on stastistical breakdowns of each teams before they play eachother along with personal opinions regarding injuries, playing back to back games, etc.
    I don’t always agree with some of the pics but there is good reason and backup on the picks they do have and I did win one ticket last week just picking the games this site
    picked and I won. Keep up the good work, its a good fast website to look at before you want to go bet

Comments are closed.