Wednesday, August 27, 2008

Trivia Craps

This picture from my photo album is going to bring back a lot of memories for a chunk of the Oilogosphere. Pictured from left to right that's me, Pat, Matt, Dennis, Lain and Tyler. Judging by the street and the way that we're dressed, I'd peg it as North Edmonton in about 1978. We're playing trivia craps, like we did almost every warm day back then.

Those were the salad days, we'd buy a stack of sports trivia cards from the five and dime, and then go head-to-head in hour long games of trivia craps. Whoever answered the trivia question first got to roll the dice, and if they rolled an eleven they won a dollar from their opponent. Simple as that. After an hour the game was over and other guys played.

Now Lain was the bomb a sports trivia, still is. When I was squaring off against him I'd always try to drag it out, limit the number of rolls. I'd lollygag when it was my turn to shoot, and otherwise keep him engaged when it was his turn. He'd be about to throw the dice and I'd casually throw in a "Hey, that's a nice tie, Lain. Is it store-bought?", or something similar. And he'd usually bite. This was good because there was no way in hell that I was going to beat him at trivia in the long run, and keeping the number of total dice throws down, just dragging the game out, that gave Lady Luck the best chance of letting me win.

Along that vein, you can see a bit of someone's head behind Dennis there, does anyone else think that's Paul Maurice? You remember Little Paul from the neighbourhood, no?. If so, let me know, because I'm thinking of suing that effer for theft of intellectual property, based on his game planning vs Detroit in the 2002 Stanley Cup Finals. I mean that's pure 'Vic v. Lain at trivia craps' I think, no mistaking it. Maybe the lawyerin' types around here can tell me if I have a case. But I digress.

Now everyone was rolling the same dice here, so it seemed to me that the obvious long term winners were going to be the guys who were good at the trivia; Lain first, then Dennis, and the rest of us about the same, and the rest of the neighbourhood well back of us. The other neighbourhood kids aren't shown here for a reason; the pictured kids already owned all of their cool stuff, such as scooters and yo-yos.

It turns out that we may all have been wrong.

In a recently published article in the prestigious Journal of Embarrassing Hobbies, it's been proven that even over a massive sample, the overwhelming driver of results at trivia craps is 'thinking positively when you roll the dice'. This is peer reviewed research, and carries merit. Forget what you thought you knew about trivia craps, people. The fact is ... it's all in the psychokinetic control of the dice.

In 50 simulated half-seasons of trivia craps, featuring 30 players, each with 41 hour-long games, positive thinking while rolling the dice is the clear and obvious driver of results.

To be completely random, and to account for the "Vic vs Lain" effect, they chose the number of dice rolls to match the shots-directed-at-net for 30 NHL teams, and used an average dice weighting to match, i.e. odds of an 11 being rolled the same as NHL goals/shots-directed-at-net (4.5%).

This is compelling stuff, people. We're talking about dice roll numbers that range from 2,712 to 3,642. That's a whack of rolls, surely to God there is no luck left in that big of a number. And the truth is living in the Positive Thinking folks, to hell with skill at trivia, hell I'm starting to think that it doesn't even exist. Now that I think about it, I must have just imagined it. Damn! I've played, coached and watched a lot of craps, as well. I feel much shame. Bruce tells a great story about a craps player who'd twist his wrist just a little bit before he threw the dice, something the casual observer probably missed, and that's never made more sense to me than now. And now is neither the time nor the place to be talking about repeatability or predictive value, or any of that other mumbo jumbo. Just grab your dice and a bit of money out of your Mom's purse, and meet me at the corner.

21 Comments:

Blogger Lowetide said...

You're being kind. I was the kid who had a job as stock boy at the grocery. I was definitely a little older than this bunch, but knew all your sisters.

:-)

As an aside, does the power of positive thinking work in the season after you rolled 7 in shootouts from November to spring?

8/27/2008 6:39 pm  
Blogger PDO said...

Someone better tell GlenX, Rawbear and Cogliano all of this information ASAP.

8/27/2008 11:26 pm  
Blogger Vic Ferrari said...

Well, it's just done for teams here, PDO, but the point remains. Hot streaks are usually just that. Shit happens for no other reason than it must.

So I took the Edmonton Oilers, saw that of all the EV shots directed at net in their games, they had taken 46.1% of them. That becomes their expected winning percentage at trivia. The trivia cards, of course, roll off the way they will, there are bound to be stretches where you get questions you just don't know the answer to, and vice versa.

And the Oilers were on the ice for 3160 total EV shots directed at net in their first 41 games last season. That became the number of trivia questions they faced.

The dice, of course, are the same for everyone. A 4.511% chance of rolling an eleven, the same odds as an EV shot-directed-at-net finding twine.

Then do the same for the other 29 teams, random numbers determine if you win the trivia, and if you do another random number determines if you win the dice roll.

Then do the same for the back half of the season. Then see how it shakes out.

I ran it 50 times, and the average correlation to winnings in dollars (same thing as goal differential):

Dice rolling: 0.76
Trivia answering: 0.63

It varies year to year, depending partly on how the trivia cards fall, but mostly on how the dice roll. Still, in 44 of the 50 seasons, thinking positively while dice rolling had a stronger correlation to winning than being good at trivia did.

The problem is that there is no repeat in it. Nil.

There is terrific repeat in trivia answering scores (analagous to corsi +/-), averaging .93.

And while I was at it, what the hell, I simulated the back half of the seasons at the same time.

It turns out that trivia results from the first half of the season was a pretty good predictor of results in the second half. .60 on average. Not overwhelming, because the clatter of the dice will still be heard.

Dice rolling had almost no repeat on average, .01, though there were a few seasons in there where it was really significant.

The 50th season, by coincidence alone, mirrors the 2007/2008 NHL season.

First half NHL:
Corsi +/- to first half +/- : .38
Shooting% diff to +/-: .68
Corsi +/- as a predictor of the second half +/-: .49
Shooting% diff as a predictor of second half +/-: -.22

First half Trivia craps:
Trivia +/- to first half +/- : .39
Dice shooting% diff to +/-: .73
Trivia +/- as a predictor of the second half +/-: .62
Shooting% diff as a predictor of second half +/-: -.35

.

Of course Trivia +/- is more repeatable in this method, (between .88 and .98) and hockey teams have injuries, fluctuations in schedule difficulty, young players who get better, etc. So repeatabilty of corsi always seems to be in the .75 to .85 range, using sample correlation as a convenient yardstick here.

Makes sense, no?

Not saying that shooting % and save% aren't important, they obviously are. Just that so much of it is pure shithouse luck, that sensible people who are trying to analyze this stuff are well advised to treat them as pure luck to start, and add an element in for them after. Of course that's a Bayesian approach, we're taking what we know about hockey and trying to build a model. We learn more, we change the model.

If you don't do that, if you take a frequentist approach, then I think you'll lead yourself into the woods and you may never come out.

8/28/2008 1:08 am  
Blogger Lowetide said...

Vic: What about the first 10 games of a season? Is that a tell for how the rest will roll out?

8/28/2008 7:47 am  
Blogger Oilman said...

Depends whether you're using composite or wooden dice LT:o)


and before anyone says it...yes, I'm being an ass.

8/28/2008 9:47 am  
Blogger Vic Ferrari said...

Lowetide:

I just checked, and moreso than I would have thought, LT. A guy would have to remember injury situations and look at the early season schedule for these teams to be fair, but I think a guy has a decent indication for most teams at that point.

I mean Philly was +13 at evens after 10 games, best in the league. But the corsi was brutal (-108), worst in the league. as was the zone +/- (-34), that was a bubble about to burst.

Over the remaining 72 games they would have the 4th worst EV+/- in the league at -22. This even though, like most young teams, they saw the underlying numbers improve quite a bit as the season wore on.

.

Toronto was +5, had an impressive +71 corsi and a decent +17 zone+/-. Maurice teams are always a bit shot happy, still good underlying numbers though. But they had played 8 of the first 10 at home. They wouldn't maintain that pace, but were still a decent EV team over the rest of the year except for poor goaltending numbers.

.

Ottawa was 3rd in the league at EV+/- 10 after 10 games. But the corsi was not particularly good (-22), and the zone +/- only decent at +16. They would go on to be just +4 at evens over the rest of the season.

.

And WSH had really good underlying numbers in spite of being -1 at evens after 10 games. Using Corsi you would have projected them to be +25 over the rest of the season at evens, they were in fact +21 over the remainder.

Hrm. Maybe Boudreau isn't as much smarter than Hanlon than I thought. WSH always has that brutal November road trip as well, have bad luck with the bounces in October and it's almost inevitable that your coaching job will be in jeopardy by American Thanksgiving.

8/28/2008 11:29 am  
Blogger Vic Ferrari said...

OTOH, VAN and DAL stumbled out of the gate badly. Bad EV+/- numbers were earned, but they'd both bounce back a bunch over the rest of the season.

And PIT had poor underlying numbers and a very poor EV+/- after ten games. The underlying numbers would continue to be pretty bad right up to the trade deadline and then the return of Crosby. Stellar save%, and perhaps special teams (didn't check) kept them in the playoff picture though.

And PHX looked to be for real after 10 games at least at evens. EV +1, Corsi +1, zone +23. It wouldn't hold though, -12, -154, -98 respectively over the next 72 games. I don't know why they got out of the gate so well, might have been a soft sked, maybe just well prepared for the start of the season.

8/28/2008 11:40 am  
Blogger Jonathan said...

Okay, so for those readers of yours (like me!) who aren't quite so quick on the draw, let me know if I've basically gotten the thrust of your post (between the analogy, sarcasm and numbers I may be lost):

1) Underlying numbers are the key factor in prediction because they're a) repeatable and b) unaffected by luck

2) The Edmonton Oilers last season had ugly underlying numbers for good chunks of the season, and despite improvement as the season went on, the team's overall finish was very flattering to their actual level of ability.

3) All those bright-eyed optimists (me again) who are predicting big steps forward should try and remember that the team's final position in the standings wasn't indicative of ability and therefore be conservative in predicting the future (given that the improvements this summer were only modest in nature).

Is that about right?

8/29/2008 1:17 pm  
Blogger PDO said...

Hows this for ugly? Lets pretend there was a stat called "blind luck." Said stat was simply adding SH% and SV% together. I know there's a way to check what this number should generally be, but I hate math so lets just say 100% for shits and giggles.

Oiler players who had over 101%:

Nilsson (103.9), GlenX (103.8), Cogliano (103.4), Stortini (102.8), Horcoff (101.6), Rourke (101.4), Moreau (101.4), Gilbert (101.1), Greene (101.1).

And Oilers who had under 99%:

Smid (98.7), Brodziak (98.5), Roy (98.5), Tarnstrom (98.00), Stoll (97.6), Visnovsky (97.3), Sanderson (96.9), Reasoner (96.8), Pouliot (96.7), Thoresen (95.5), Jacques (87.1).

You'll notice the first group tended to get extensions while the second ground tended to get shipped out of town.

Is it worth much? Probably not.

But I found it interesting none-the-less.

8/29/2008 2:13 pm  
Blogger Vic Ferrari said...

Wow, that's very cool, PDO. Is that on-ice SH% and SV%?

There should be some sustain, from season to season, in the former and next to none in the latter.

They bought low on Visnovsky though, that's good. And sold high with Greene at least.

And Sanderson has always given back as much as he created, so it was convenient that he had a bad year with the bounces or they might have inked him again, so that worked out well.

Nilson has decent underlying numbers as well, did all year really, and you know I value that, but giveaways at the offensive blue line are my cowbell (as the five-hole-fanatics lads would say) so I can't find a place in my heart for the guy. In any case that's a very movable contract imo.

And though you don't have him listed, Cole drives possession like Ryan Smyth, or close to it, you win games because of players like that. He'll make a big difference this year.

They let GlenX walk over $ and didn't cave to Stortini's initial contract demands, that's good.

I think that Horcoff is a terrific player anyways, I mean it would have been better if the extension year had happened after a bad-bounce season, but these things happen. He's a terrific bet to outperform that deal in any case.

I would bet real money that if you used this and looked at windows of opportunity for rookies (not the obvious stars, but the rest) ... roll sevens early and you have a long Dubinsky/M.Sullivan/W.Primeau NHL career. Roll 11s early and it's off to Europe.

I've always thought that, generally anyways, but this metric would probably drive it home.

Brodziak will get better in terms of outchancing, and very likely outscoring as well. The net crashers deserve a better rate of finish anyways. Guys like Ciccarelli and Roberts probably sport the best career 5v5 shooting%'s around (Mario and linemates aside), if the NHL ever decides to publish those stats for prior years we'll no for sure.

And I liked Thoresen, that Thor/Stoll/Reasoner line played unspeakably tough ice time for a long stretch there, right up until Gagner was oddly subbed in for a bit iirc, I think against L.A, which was a good choice really (I won't bitch at the coaching staff there, the development of that guy is far more important than anything else). But Thor just had no finish. Saw him at the WHC's, and same thing, the Norwegians always got the puck to the right end of the rink with him out there, message board fans took notice of the wee Norseman, but you just knew they weren't ever going to even get a clean chance on goal, much less score. Still, he's probably a better player than a hundred guys who will take an NHL paycheck this year.

8/29/2008 3:35 pm  
Blogger Vic Ferrari said...

jonathon:

1) ... yes, absolutely. It also draws a line in the sand at the high tide level, the best point where we can ever consistently predict future success.

2, and 3.) On the Oilers ... well fuck, everything went in there for a stretch near the end. If you believe that will carry on you've been smoking the drapes. I don't think we need to bring out numbers on that. The important thing, at least to me, is that MacTavish knows better. Which you'd hope to God he would anyways, but he seems to know better in the right measure as well, based a CHED interview that I heard during the winning streak. That's encouraging, at least to me.

8/29/2008 4:18 pm  
Blogger Jonathan said...

Vic:

Well, I certainly don't expect the Kid Line to be able to duplicate their success (or for that matter, to still be a unit by game 20) because it was so far and beyond what could be reasonably expected. I also think the performance of that line was what really drove Oilers success late-season, and it can't be expected again. My chief reason for expecting improvement is two-fold: a) reliance on Garon early in the season (and hopefully, a dupliation of his efforts this past season, which seems within the bounds of probability) and b) the acquisition of Erik Cole, which you've already commented on.

As for MacTavish, I think he knows what's coming. The braintrust must realize how precarious the team's back-end is, given that it sounds (from Huddy esp.) that Souray/Staios is going to get thrown to the wolves this year.

The team can ice two lines who should be able to handle toug sledding; Pisani's line and Horcoff's. I don't know if it's enough, but it's better than what Vancouver and Calgary have and that's a good start.

8/29/2008 5:02 pm  
Blogger Lowetide said...

They need a nice veteran defenseman. It'll make a difference. Where have you gone Joe Watson?

8/29/2008 5:34 pm  
Blogger Vic Ferrari said...

Yeah jonathon, I always refer to the "kid line" as well, but calling them "Gagner and two other guys" would probably make more sense. I don't give two craps about set lines anyways (I don't play pools), so for me you do the right thing for Sam, and what happens to the others happens to the others.

I mean coaches have to set priorities, and that one seems too obvious to miss.

8/29/2008 5:35 pm  
Blogger Vic Ferrari said...

LT:

If it's the 78th game of the year, vs VAN, playoff position in the offing for both teams, Oilers leading by a goal with five minutes left in the third, Vigneault sends the Sedins over the boards ...

Who do the Oilers counter with?

IMO the Oilers don't need Joe Watson (though I'd take Behn Wilson off that team, even if he's 55 years old now, that cat was off the hook).

The Oilers need a couple of Colorado's defensemen (Foote, Salei, Clark, Leopold ... is Skrastins still there?) In any case Staios is on the fringe of that class, and Souray and Vis are behind, and it's a big drop after that, howver promising these kids might be.

Like BDHS says, no clear top end defensive guys, but the bottom D pair should be the bomb, relatively speaking.

8/29/2008 5:42 pm  
Blogger PDO said...

Vic:

Bang on. It just kind of occurred to me one day that we always talk about a guy having an unreasonably high SV% or an unreasonably high SH%... but it's rare that someone brings up both in the same line.

By this Metric last season Cole finished at an even 100.5%... so it's fair to say he worked for everything that he earned last year. What interests me, is most were saying last year he took a significant step up in competition, which very well may be true.. I haven't checked, but his defensive/offensive zone draw split left him at a -31. Positively though... he definitely drove possession, finishing with a 107 Corsi. Not sure what to think of him quite yet... love the style, but I'll have to watch him closely to see what I think of the substance. I'm really hoping he starts with Horc-Hemmer though... I think an egg line will be necessary on this team. We need to be able to run the other teams bench, because we don't have the bench to match theirs.

On on-ice save percent, have you found any "type" of player who has relatively had high and/or low regardless of anything? I'd expect the soft-minute killers to have very low for example... simply because most chances against would be odd man rushes with someone like Iginla jumping on the ice. There's a few defenseman (Pronger, Lidstrom) who I would also expect to traditionally have higher SV%'s behind them, simply because they're true difference makers back there. Finally, a guy like Pahlsson I'd expect to have a poor SV% as well... he's very good at preventing the shot and forcing it outside, but when he doesn't, it's Zetterberg in the slot, not Cleary.

I love the Horcoff contract as well.. not worried about it. I think Nilsson and Stortini are both movable contracts and likely good bets, but still found it interesting none the less. My big problem is guys who bail out on the physical stuff to make a play every single time.... Nilsson got over that, so I can find some love for him.

The two that really interested me though, were Gilbert and Grebeshkov. Gilbert saw more ice time, and more defensive draws (379-329), but Grebeshkov still saw more as well (260-243) AND drove the Corsi far more (-187 versus -67) while posting a much less lower SV% (.918 vs. .906). He was aided a bit by the shooting % (10.4 vs 9.3).... but I don't see much to suggest there's much to choose between the players outside of their birth certificates.

I've been a proponent of trading Stoll for a while. The SH% kinda drove it home... but the guy hasn't ever been able to produce at EV's, but always got paid like he could because he could produce on the PP as a one-trick pony. He did a lot of grunt work, but I think he became less valuable because of his clapper to be honest...

And I love the Visnovsky deal. Crawford even said that they track every scoring chance (I'd love to see those #'s....) and that Vis posted nearly identical numbers this season and just had a lot less finish. This backs that up.

8/29/2008 11:40 pm  
Blogger PDO said...

Oh, and I suggested trading Cogliano long before I ever saw these numbers... but looking at them, it drives it home for me. He had one of the best EV SH%'s and SV%'s on the team, with terrible underlying numbers and shiny offensive ability in a rookie season....

My guess is he could've brought back a lot in a package with Pitkanen....

8/29/2008 11:42 pm  
Blogger Black Dog said...

I remember winning that hat but I also remember when Dennis won my shoes. I caught hell from my Dad that night let me tell you. Off with the belt.

Great stuff Vic. It will be an interesting year for sure - three things to hang your hat on if you are an optimist - which I am.

Cole, like Smyth, is going to drive results in the right direction.

The additions of Visnovsky and a healthy Pisani will do the same.

Calgary and Vancouver have taken obvious steps back and I'm inclined to say Colorado and Minnesota have too.

Bonus - many of the kids will likely be better

Here's hoping. Colour me cautiously optimistic that the numbers will be better but for the right reasons

8/30/2008 12:04 pm  
Blogger Vic Ferrari said...

Pat:

Yeah, I'm optimistic about this team too. Though I'm less moved by the late season surge than most, I think that the Gagner line is going to spend less time in their own end of the rink, and end there less often. The consequences of that should ripple right through the lineup, I think.

And for the other reasons you mention too, though they did lose a good player in Pitkanen, and Reasoner's and Stoll's icetime (own zone draws and playing a bunch against good players) that didn't leave town with them, so Pouliot or Brodziak, in combination with Horcoff, are going to have to step up.

8/30/2008 2:17 pm  
Blogger Black Dog said...

Yeah Vic that issue is the big one for me as well - I think that they are fine on the wings but they need a kid to step up in the middle and someone on the blue line.

As for the last ruah they made I did take some positives from it. I do remember the SJ game and a Flames game where the Oilers were outplayed and still came away with the points but they also had some straight out wins where it was hard to argue that they were the better team.

I remember talking to Dennis halfway through the year and the number of regulation wins they had in the previous sixty games were in single digits. They had some poor luck in there early in the season - I remember a couple of games against the Canucks and Wild for example where they deserved better results for sure. But the truth was they were just awful.

But down the stretch thye manhandled some opponents pretty well - thumped Phoenix LA and the Ducks iirc and were also full value for wins over the Flames and Canucks.

Its those games that I'm hanging my hat on. A lot depends on a lot of kids not named Gagner continuing to trend upwards rather quickly.

9/01/2008 11:30 am  
Blogger Vic Ferrari said...

Pat,

Well Dennis always feels anger to the Gods whenever the Oilers outchance but lose. The fact of the matter is that the better team loses about 1/3rd of the time in hockey. Even moreso in soccer, but there is less parity in the Euro leagues so that tends to get lost.

Still, even though the Oilers were getting outplayed and outchanced in almost every game early, they deserved better I thought. I mean NHL teams aren't all that much better than each other nowadays. I didn't think that the Oilers were losing games that they had the better of the play in, I just thought that even when you're bad, you deserve better from the bounces than they were getting.

I remember an early Oiler game that they clearly had the better of the chances in, vs Nashville in late Oct or early Nov, the Preds on the back half of a B2B and hurt by injuries to boot. They eneded up losing by two or three, but they were clearly the better team I thought.

MacT was full into the madness then, running a curious combo of Penner/Stoll/Nilson vs Arnott's line. The H2H may or may not show that, the shift charts will, that was certainly MacT's intent on the night. Mad stuff.

I remember that late in the game the Oilers were pressing and Grebeshkov tried to keep it in on the players bench side of the rink by looking right and then trying to bank it of the boards the other way. It wasn't a bad gamble, but on this night it ended up in his own net, and Grebs the target of talk show "trade that effer!" callers.

Dude got way, way more than his share of breaks later in the season, at least by my eye, so I won't cry for him. But that wasn't a bad bet in the circumstance.

Long and the short of it, you can only control what you can control. I think that the Oilers will be well prepared, and I think that they have the horses to be the better team on a lot of nights this season. Maybe even more than half. Special teams will be huge, and the hockey gods get last say. That's it.

9/01/2008 2:48 pm  

Post a Comment

<< Home