tirris daily nhl predictions

tirri · Post by **tirri** » 24 Feb 2013, 13:45

Hello friends

as a hobby for about 2 years ive been building a statistical prediction model for nhl results and betting.

i have no real qualifications apart from following the nhl for 10 years, making the model for years, having a statistics/matchematics/economics degree, making my masters thesis on rank based exponential smoothing and predicting stock volatility, and working with the same exact model in my previous workplace that i now use to model nhl results.

depending on the seasons, over the course of a full season the model gives a profit of about 30-50% tax free. this profit comes from only using the model and not making any manual modifications whatsoever. it will be easy to get better profits if you take into account very recent injuries or back up goalies playing. the profits are also calculated from average odds of about 10 bookies, so it would have been possible to find a bookie with better odds every time. this means that if you put money on about 6 different bookies and always make the bet with the best possible odds, the profits will easily be higher.

im simplistically talking about "the model" when in fact well over a 100 models have been tested and now 15 models are being run daily and out of those a kind of summary is extracted to make betting decisions.

so please join me in making easy money with what i dare say is one of the most sophisticated betting system ever made.

here are today's predictions:

the only worthwhile team today is boston against florida, with the odds of 1,7 for winning in regulation or OT/SO. you will no doubt find better odds than that but with at least 1,7 it will be worth it
other matches worth mentioning are pittsburg against tampa and chicago against columbus, but they arent bet-worthy

thats it for today folks!

tirri · Post by **tirri** » 24 Feb 2013, 14:20

edit:
it seems that bostons odds are rapidly decreasing and the bookie i placed the bet on now has boston at 1,62, so most likely you wont find much better odds than 1,7 after all.

i ran my models about 8 hours ago in the morning and already placed the bets then and obviously the model picked up bostons great odds. the odds now dropping is a proof in itself that there was value to be found there.

i might have to start making my predictions post first thing in the morning so you wont miss on these amazing opportunities

Myrk · Post by **Myrk** » 24 Feb 2013, 17:10

Sounds interesting, keep em coming tirri

Pogue · Post by **Pogue** » 24 Feb 2013, 19:14

Finland sucks at hockey.

par73 · Post by **par73** » 25 Feb 2013, 04:28

boston will roll faces

tirri · Post by **tirri** » 25 Feb 2013, 06:18

boston with a convincing win and other predictions hit home too. but dont be fooled as they were all heavy favorites anyway. follow my predictions longer and make your own conclusions!

sadly i wont have time to run the model right now but in 5-8 hours and then i might already have missed the best odds, but i will update this thread later anyway

Results after day one: 70% profit

tirri · Post by **tirri** » 25 Feb 2013, 13:47

monday's predictions:

there are no prime candidates for betting today, although a few teams were really close and i will probably make a few bets just for fun. but just to be clear, none of these matches is really bet-worthy

philadelphia against toronto with 1,68 odds for regulation or ot/so win (phillys predicted probability to win is about 60%)
los angeles against anaheim with 1,62 odds for reg/ot/so win (las predicted probability to win is about 64%)
chicago against edmonton with 1,5 odds for reg/ot/so win (chicagos predicted probability to win is about 69%)

stay away from dallas v nashville and ottawa v montreal

tirri · Post by **tirri** » 26 Feb 2013, 09:29

mondays results
2/3 fun bets isnt bad but all in all this wasnt a very exciting day

Tuesdays predictions

Finally some interesting action, as for the first time the model suggests something else than heavy favourites. im not sure how much i like these predictions but i will probably follow them anyway. these matches are worth betting:

columbus against dallas, 1,9 odds.
phoenix against vancouver, 2,6 odds (road underdog, very interesting to see if this one goes right!)

and finally not bet-worthy but still interesting
florida against pittsburg, 2,4 odds

enculator · Post by **enculator** » 26 Feb 2013, 10:39

Hello fellow myther,
Can you please predict the next euromillion numbers, I'd like to be very rich.
Best regards,
Enculator

tirri · Post by **tirri** » 26 Feb 2013, 11:31

this is better than lottery. this is a kind of an iq test for you fellow mythers and im afraid most of you will fail. i aced all my statistics courses to become a masterful statistician, then worked with prediction models for years, then studied my object of passion for 2 years and im now giving away my expert knowledge for free.

im interested in how many mythers have the brains to take advantage of this special occasion. think of how much a service like this would normally cost

grim · Post by **grim** » 26 Feb 2013, 11:58

Tirri is beginning to use his job as the supreme statistics supervisor of the bureu of supreme statistics of Finland for evil.

tirri · Post by **tirri** » 26 Feb 2013, 17:34

quick check to see how my predictions are doing
phoenix' odds have gone down from 2,6 to 2,5
columbus has gone up from 1,9 to 1,94
and finally florida has dropped dramatically from 2,4 to 2,2

im still standing by all those recommendations, even the columbus one. only with the odds i told earlier of course, i dont think florida at 2,2 for example is good at all

tirri · Post by **tirri** » 27 Feb 2013, 07:22

tuesdays results

even though columbus lost in overtime the results werent too shabby with phoenix' nice 2,6 odds, giving us a total profit of 30% for today
the fun bet i made on florida at 2,4 also sweetened the day.
notice how i dodged some bullets on heavy home favorites like tamba losing against buffalo with 1,6 odds and rangers losing to winnipeg with 1,54 odds

all in all the model is still going strong!!

i cant run the model yet and i probably wont have time until tonight. im afraid good odds might be gone by then so maybe no real bets today

tirri · Post by **tirri** » 27 Feb 2013, 10:33

i had time after all.

wednesdays predictions
montreal against toronto at 1,88. it's a shame im putting this just now becfause i believe montreals odds were even higher

not bet-worthy but still quite good: los angeles against detroit, 1,62 odds

dac · Post by **dac** » 27 Feb 2013, 16:47

these predictions have been amazing so far. tirri you should share this model, i'd love to make a killing on it. I'll even find a way to get you a signed teemu selanne jersey! and yes I can make that happen, i go to ducks games quite frequently. what finnish man can turn down anything teemu?

Hadzenegger · Post by **Hadzenegger** » 27 Feb 2013, 18:24

yeah mad props, im impressed

tirri · Post by **tirri** » 28 Feb 2013, 10:13

wednesdays results

montreal with a convincing win for 88% profit for today. the not quite bet-worthy LA was also victorious giving us a ridiculously good overall success rate when you take into account all the matches i have mentioned in these posts

thursdays predictions

interestingly there are no prime targets for betting today, even though there are a lot of games played. here are the almost bet-worthy matches, i might put in some fun bets but its kind of hard to decide which of these would be worthy.
calgary against colorado, 2,18 odds
dallas against edmonton, 1,76 odds
boston against ottawa, 1,48 odds
carolina against pittsburgh, 2,18 odds (this one was actually bet-worthy according to the model but i wont include it because of carolinas goaltending)
islanders against toronto, 1,76 odds

edit:
i put a small fun bet for calgary against edmonton, win in regulation 2,95 odds. no other bets for me today

cremisi · Post by **cremisi** » 28 Feb 2013, 12:41

Tirri ~ LOL in Finland we are 12 hr's ahead gg USA.

Pogue · Post by **Pogue** » 28 Feb 2013, 18:45

Even though I've been a Bruins fan all my life I haven't watched a game all season. Those dumb fucks have had 2 lockouts in less than 10 years. The NHL can shampoo my crotch. Watching NBA. K thx bye.

tirri · Post by **tirri** » 28 Feb 2013, 19:46

extra update:
i bet on calgary when the odds were 2,95 and now the same bookie has calgary at 2,6!
needless to that even if calgary loses, the odds at 2,95 were definitely too high and it was a nice time to bet

switch · Post by **switch** » 01 Mar 2013, 03:58

http://www.thestar.com/sports/leafs/201 ... rtime.html

Toronto Maple Leafs defeat New York Islanders in overtime
Nazem Kadri’s hat trick the first by a Leaf at Nassau Coliseum since Paul Henderson did it Feb. 28, 1974 — 39 years ago to the day.

tirri · Post by **tirri** » 01 Mar 2013, 05:19

thursdays results
first losing day in a while when only 2/5 matches mentioned were right and calgary loses when leading 4-2 after the 2nd period. remember that you should never actually bet on 5 matches or something like that because its very rare that the odds would be so off and you will always end up losing money. the good thing was that the only match that the model deemed bet-worthy was a clear 4-1 victory for carolina. too bad i was too scared to follow the model

only 3 matches today and at a glance they dont seem too good but ill run the model later

Myrk · Post by **Myrk** » 01 Mar 2013, 06:04

Bet on Boston, unfortunately I bet on a 1.5 spread and boston only won by 1 goal so I'm down $5. I'm betting on tirri-recommended matches until I lose $50 or get rich, whichever comes first

tirri · Post by **tirri** » 01 Mar 2013, 12:05

i will give some insight into how i interpret the models. usually there are only one or two indicators that tell me to bet when i say that the match is bet-worthy, and its okey because all the indicators i use have proven to be profitable. however your bet size should always be relatively small compared to your stack. i will be sure to mention when almost all indicators tell me to bet, because that kind of matches happen too, although there have been none yet.

fridays predictions
this is a tough one because only 1 indicator gives thumbs up for chicago at 1,36 odds and st. louls at 1,53 odds. im itching to bet because that indicator has a win % of 88% and 124% profit so far this season, but since no one other indicators tell me to bet, i think i will pass today.
so those are really close to being bet-worthy, but not enough. if you do happen to bet, be sure to check if elliott is playing for st. louis because he has a horrible terrible 84,9 save-% in 9 games this season.
http://www.leftwinglock.com/starting-goalies/

tirri · Post by **tirri** » 01 Mar 2013, 20:17

i ended up betting chicago for win in regulation or overtime and 3 bets for st. louis, win in reg or ot, win in reg, and -1,5 spread, mostly because halak is starting. but these were only fun bets that have no real effect to my cashroll. fun bets should probably be less than 1% of your total and i only do them to keep me interested

browning · Post by **browning** » 01 Mar 2013, 23:26

what % of your bankroll would you bet on every match, assuming you bet once a day on average?

NewMutator · Post by **NewMutator** » 02 Mar 2013, 04:21

Very interesting thread, tirri.

I don't know much about sports betting. Are there any recommended venues for placing bets? It might be fun to make a small bet or two to test your model.

tirri · Post by **tirri** » 02 Mar 2013, 06:28

you can compare odds that bookies offer here: http://www.oddsportal.com/hockey/usa/nhl/
i havent had time to see which bookie seems to give the best odds on average, although marathon seems to have a lot of really good odds but its a russian betting company and im not sure if i want to put money there. im sure any respectable bookie will do at first.

browning:
i would use kellys criterion or rather less than kellys criterion http://en.wikipedia.org/wiki/Kelly_criterion, although anyone following my advices wouldnt be able to use those unless i always told the predicted probability.

fridays results
back on track with all 4 bets winning but in all honesty, i doubt any of those would bring profits in the long term, which is why i didnt put them here as bet-worthy

Myrk · Post by **Myrk** » 02 Mar 2013, 08:14

won $3 betting on st louis.

tirri · Post by **tirri** » 02 Mar 2013, 10:53

saturdays predictions

los angeles against vancouver at 2,15
new jersey against buffalo at 1,95
montreal against pittsburgh at 1,86
normally i'd be wary of betting on 3 matches but the model has served fine so far and all 3 matches are showing strong indicators, so i didnt really know which match i should exclude. those 3 should be bet worthy with those odds but dont go crazy cause vancouver and pittsburgh are no pushovers. i assume that Price is starting for montreal and will make my bet with those odds, but if budaj would start for some reason, i would hope i hadnt bet.

good but not bet-worthy:
san jose against nashville, 1,7 odds
boston against tampa, 1,47 odds

NewMutator · Post by **NewMutator** » 02 Mar 2013, 18:32

Other than driving to Vegas, I don't really know of any reputable bookies.

Cutard · Post by **Cutard** » 02 Mar 2013, 18:33

I don't know how open most teams are about saying who is starting but beat writers are usually pretty good at predicting which goalie it'll be after watching the morning warm ups.

tirri · Post by **tirri** » 02 Mar 2013, 19:24

yes you can find starters easily after morning warm ups from a number of pages but i often dont wait for them and just do educated guesses, mostly just to get the betting over with but also to catch some good odds.

here is an important update that affects a bet-worthy match:
vanek will be out for buffalo, making new jersey an even better option. i made my bet at 1,95 and the same bookie now offers 1,78 odds. i have no idea how much losing vanek changes the predicted probabilities so i cant say if new jersey at 1,78 is still good or not, but im very happy about my bet at 1,95

tirri · Post by **tirri** » 03 Mar 2013, 05:29

saturdays results

first real stinker of the season for me as it looks like all model recommended matches went wrong for -100% profit.
probably the biggest reason was quicks .818, prices .806 and hedbergs .870 save percentages and even then njs and montreals games went to ot/so. oh well, days like this happen and the model had been doing a little too good anyway

tirri · Post by **tirri** » 03 Mar 2013, 07:44

sundays predictions
rangers against buffalo at 1,57
columbus against colorado at 2,1
and small condifence pick for calgary against vancouver at 2,1

better but not bet-worthy:
carolina against florida at 1,9
chicago against detroit at 1,84
minnesota against edmonton at 1,64
st. louis against dallas at 1,9

i urge you to only follow these predictions for a month or two and not make too big bets and ill post some summaries of profits then so you can make educated decisions!

East Wind tmnt · Post by **East Wind tmnt** » 03 Mar 2013, 16:22

What bookies do you usually use - and are you betting on Home/Away?

tirri · Post by **tirri** » 04 Mar 2013, 06:45

i usually make bets where overtime and SO are included. in rare circumstances i bet on win in regulation, usually if i make combination bets and want to have some higher odds there too. this backfired just last night as i made a 5 bet combination where i put rangers win in regulation but it went to shootout. i have also experimented with spread bets and some people think that betting +1,5 spread on underdogs might be more profitable than regular bets on underdogs but i havent done any research on this. it might be more profitable in the long term if my last night's calgary and columbus bets would have been +1,5 bets for example, even though they both happened to win this time.

i have usually bet on unibet but i am now mostly using pinnacle and will probably put more money on a couple of other bookies.

sundays predictions

[youtube]http://www.youtube.com/watch?v=kIfOjkB17BA[/youtube]

after my worst day so far comes one of the best as all 3 model recommended matches won, including 2 underdogs in columbus and calgary, for a total profit of 92%. also 6 out of all 7 matches mentioned were correct.

tirri · Post by **tirri** » 04 Mar 2013, 11:30

mondays predictions
there are no bet-worthy matches today.

good but not really very close to being bet-worthy:
los angeles against nashville, 1,6 odds
new jersey against toronto, 2,04 odds (this would be close to bet-worthy but i will abstain as hedberg is the starting goalie)
tamba against pittsburth 2,6 odds

ill make a fun bet on los angeles and thats it

Myrk · Post by **Myrk** » 04 Mar 2013, 18:57

lost $20 on Friday's bets, risked $15 on Saturday and won $16.63.
So far I've found a couple sites I can use, bodog and sportsinteraction.com. I'll look at pinnacle and unibet too and put my bets where the best odds are.

The odds I got on sportsinteraction.com on Saturday were Buffalo v Rangers 1.48, Calgary vs Vancouver 2.1, Columbus vs Colorado 2.15.
Are you picking bet-worthy matches based on the betting odds being better than your model's predicted odds tirri?

tirri · Post by **tirri** » 04 Mar 2013, 21:07

i just predict the probability of winning, from which i can calculate expected values with odds, or just odd*predictedprob/100. there should be value if thats > 1. but deriving working betting rules is more complicated.

the odds are an inverse of the bookies estimated probability of winning, but they always sum up to more than 1 because of the in-built percentage that bookies take. the bookies dont of course necessarily set their odds to what they think are the real probabilities of winning, but rather how they want the bet distribution to be in order to always make money from each match. so in a sense even in fixed odds betting youre not really trying to beat the bookies but other betters and if someone can predicted the possible outcome well, there are chances for profitable betting even if the predictions are nowhere near as good as what the bookie could predict.
this was more of an explanation to creme who said that betting isnt worth it because the bookies always have better models

Cutard · Post by **Cutard** » 04 Mar 2013, 22:52

SHOCKING USELESS GUT FEELING PICK OF THE WEEK

Wild > Blackhawks in regulation

The Wild have been playing a bit better lately, they already have an SO win over the Blackhawks, and Backstrom is most certainly good enough to steal a game for his team and is having his best season.

tirri · Post by **tirri** » 05 Mar 2013, 10:59

mondays results

uneventful day as i did say that no match was close to being bet-worthy but los angeles easily won for my only bet of the day.

also notice devils vs toronto which is a good example on how no matter how good the model is, you often need to do some extra research. devils deserved to win that game and the model keeps telling to bet as their stats indicate that they are a better team than their current form shows. my model however doesnt fully take into account how badly hedberg is playing right now, with .848, .808, .852, .900, .870 and .826 sv% in the last 6 games. so as it is right now, i probably wont bet on devils matches until brodeur comes back or hedberg starts playing better. the same is also true for st. louis when elliott plays.

tuesdays predictions
columbus against edmonton, 1,99 odds, but dont bet too heavily

very nearly bet-worthy:
detroit against colorado, 1,57 odds

better but not bet-worthy:
boston against washington, 1,74
chicago against minnesota, 1,5
montreal against islander, 1,8
san jose against vancouver, 2,18
new jersey against tampa bay, 1,66 (this would again be bet-worthy if not for hedberg)

more analysis on chicago against minnesota because of cu's gut feeling pick:
i have chicago winning at about 64,3% and minnesota at 35,7%, which at 1,5 odds for chicago and 2,6 odds for minnesota makes chicago a slightly better option but in the long run you would lose money no matter who youd bet for

East Wind tmnt · Post by **East Wind tmnt** » 05 Mar 2013, 17:32

definitely impressive tirri, keep it up

Cutard · Post by **Cutard** » 06 Mar 2013, 02:21

First period isn't over and Backstrom has been tightly packed dog shit in goalie pads.

zagon667 · Post by **zagon667** » 06 Mar 2013, 02:28

you spelled 'custard' wrong.

tirri · Post by **tirri** » 06 Mar 2013, 06:32

tuesdays results

a very interesting day. i was scrolling down the results and thinking oh man, one of those days again as devils lost badly with hedbergs .727 effort, montreal with prices .783 and boston with rasks .846. 0/3 matches mentioned were correct so far.

but columbus managed to overcome masons awful .625 effort with a goalie switch and a comeback win, for a nice 99% profit for today. there were even better odds than that up for grabs.

the almost bet-worthy detroit also won, so the two strongest recommendations were victorious. i also got san jose and chicago right so 4 out of 7 mentioned wasnt that bad.

tirri · Post by **tirri** » 06 Mar 2013, 15:06

wednesdays predictions
chicago against colorado, 1,41 odds. might also be a good time to experiment with regulation win bets or possibly even -1,5 spread, although chicago has played a lot of one goal and overtime games.

better but not bet-worthy:
ottawa against toronto at 2,3
calgary against san jose at 2,04. model shows calgary to be bet-worthy, but i cant call it yet without knowing the goalies. niemi is playing lights out so if he starts for sj and if he starts calgary isnt such a good option. also calgarys goaltending is so iffy right now that it might be better to but calgary bets on hold.

absolutely no edge in odds: anaheim vs phoenix at 1,64 and 2,26 odds.

par73 · Post by **par73** » 06 Mar 2013, 17:31

boston has been on a beastly winning streak

honestly they put in charra and he knocks heads in to provoke a fight, then once it's 4 on 4 boston absolutely rapes the competition with four players.

Myrk · Post by **Myrk** » 06 Mar 2013, 22:45

Signed up with Pinnacle since they seem to give the best odds and am cashing out of bodog since their odds are relatively shitty.
Pinnacle has an e-sports section and I placed a for-fun bet of $5 on Stephano to beat Golden in the IEM season 7 world championship. E-sports being a fairly new thing I'm not sure how they calculate the odds but 1.87 seems like insanely good odds since Stephano is the winningest foreign player and I've never heard of Golden. We'll see how this goes

dac · Post by **dac** » 06 Mar 2013, 23:33

par73 wrote:boston has been on a beastly winning streak

honestly they put in charra and he knocks heads in to provoke a fight, then once it's 4 on 4 boston absolutely rapes the competition with four players.

fighting doesnt cause a 4x4. jesus.