EVERY GENERATION THROWS A HERO UP THE POP CHARTS

The Edmonton Oilers have been procuring specific talent this summer, adding size whenever possible. An example: Griffin Reinhart, brought in despite a premium to be a part of the young group of blue who will play for the team in the next several years. Another big player, forward John McCarron, will be a free agent as of today, as unsigned college players find freedom on 15 August after their senior season.

  • Kirk Luedeke on McCarron July 2012: “He’s another one of those raw guys, pretty hard nosed for college. Nothing flashy, meat and potatoes north-south type player his production was pretty good. He’s got a chance to be a bottom 6 player, at very best he would be a fringe third liner or solid fourth liner. He finishes his check and goes hard to the net. He’s a pretty hard nosed guy for NCAA, has a chance to play because he’s big and strong. Straight line player, drops his shoulder, finishes his check and fights for position in front of the net.”

RUSSO!

  • Red Line Report’s Max Giese before draft day: “He’s playing well and I’d expect him to be an early second-rounder. He reminds me a lot of Matt Carle– he’s not the biggest (5-11, 190), most explosive defenseman out there, but he’s very intelligent and calculated with his puck movement. I like him OK. But he doesn’t really light a fire under me. I would imagine that NHL teams like him, but if I were with a team, he wouldn’t be my guy. Not taking anything away from him as a player, but he just doesn’t have the kind of dynamic game that I would want to spend a second-round pick on.” Source

Russo is a substantial offensive prospect (NHLE: 82GP, 10-18-28) and is a RH shot. I imagine he’ll sign with Chicago or Toronto or Detroit, but he would be a nice addition.

PRONMAN’S WORDS ON REINHART

Corey Pronman posted his 2015 top 100 prospects this week, Edmonton had three men on the list. It would have been four, but Leon Draisaitl didn’t qualify because of his NHL games (he was No. 8 a year ago, Oscar Klefbom was No. 46). Here are the Oilers on Corey’s list this season:

  • Connor McDavid at No. 1
  • Darnell Nurse at No. 20
  • Griffin Reinhart at No. 56

Pronman noted (in regard to Reinhart) He’s high-end defensively, and average offensively, which is still a pretty good player. He performed well versus men this season in the AHL and could make the Oilers very soon.”

I think that’s about right.

laleggia capture1Back to Russo. The Oilers already have a fantastic college grad coming into the system and it’s Joey Laleggia. Quality offensive defenseman at the NCAA level, the downside is size (5.09, 182) and the upside is a splendid puck-moving defenseman. His equivalency is eye popping. NHLE: 82GP, 14-23-37. If Edmonton does pursue Russo, one hopes he qualifies in the ‘two-way category’. In terms of offense, the numbers suggest Laleggia brings more to the table than Russo.

OILERS PROSPECTS, ALL IN A ROW

With McCarron no longer in the organization, and Marco Roy back (kind of), I thought it might be an idea to tweak my list from the last time it was published. I haven’t changed the ranking (that happens next in December) but did move out McCarron and placed Roy back in (not at previous level).

  1. C Connor McDavid NHLE: 82GP, 23-40-63. Destination: NHL from opening night.
  2. C Leon Draisaitl NHLE: 82GP, 15-26-41. Destination: NHL before Christmas.
  3. D Darnell Nurse NHLE: 82GP, 7-16-23. Destination: NHL this season.
  4. D Griffin Reinhart. NHLE: 82GP, 6-11-17. Destination: NHL by the trade deadline.
  5. C Bogdan Yakimov. NHLE: 82GP, 10-13-23. Destination: AHL for 2015-16, with a cup of coffee in the NHL.
  6. W Iiro Pakarinen. NHLE: 82GP, 16-11-27. Destination: AHL to start, callups to the NHL early and often.
  7. C Kyle Platzer. NHLE: 82GP, 12-17-29. Destination: He’ll spend his rookie pro season in the AHL.
  8. C-R Greg Chase. NHLE: 82GP, 8-16-24. Destination: AHL all year, with an NHL look-see possible.
  9. L Anton Slepyshev. NHLE: 82GP, 17-11-28. Destination: AHL marksman.
  10. D Ethan Bear. NHLE: 82GP, 5-9-14. Destination: WHL all year, hopefully plenty of PP time.
  11. G Laurent Brossoit. Destination: A full season in Bakersfield as starter.
  12. D Joey Laleggia. NHLE: 82GP, 14-23-37. Destination: Full AHL season, possible recall if he flourishes.
  13. D Dillon Simpson. NHLE: 82GP, 2-9-11. Destination: Full AHL season.
  14. G Eetu Laurikainen. Destination: Minor leagues for sure.
  15. R Andrew Miller. NHLE: 82GP, 16-19-35. Destination: AHL to start, significant NHL time possible.
  16. D Jordan Oesterle. NHLE: 82GP, 6-12-18. Destination: AHL all year, with a callup possible.
  17. D Caleb Jones. NHLE: 82GP, 2-5-7. Destination: Feature role in WHL with Portland.
  18. C Jujhar Khaira. NHLE: 82GP, 4-5-9. Destination: Full AHL season.
  19. R Tyler Pitlick. NHLE: 82GP, 8-16-24. Destination: NHL all season or the waiver wire/AHL.
  20. D William Lagesson. NHLE: 82GP, 1-6-7. Destination: UMass and college hockey.
  21. F Marco Roy. NHLE: 82GP, 11-10-21. Destination: Somewhere in pro hockey, climbing.
  22. D Ziyat Paigin. NHLE: 82GP, 2-2-4. Destination: KHL with more playing time.
  23. G Anders Nilsson. Destination: Backup goalie in the NHL.
  24. D Ben Betker. NHLE: 82GP, 2-10-12. Destination: Full AHL season.
  25. D Brandon Davidson. NHLE: 82GP, 3-4-7. Destination: AHL to start but could see significant NHL time.
  26. D Brad Hunt. NHLE: 82GP, 11-19-30. Destination: Impact AHL defender, with a chance for recall.
  27. L Kale Kessy. NHLE: 82GP, 8-8-16. Destination: A full AHL season.
  28. D David Musil. NHLE: 82GP, 1-7-8. Destination: Full AHL season, with a chance to make an NHL appearance.
  29. D John Marino. Destination: Another season with the South Shore Kings.
  30. G Zach Nagelvoort. Destination: Another seasons in the NCAA.
  31. C Alexis Loiseau. NHLE: 82GP, 12-20-32. Destination: AHL/ECHL and a big pro chance. S
  32. R Tyler Vesel. NHLE: 82GP, 7-13-20. Destination: Another full season of NCAA hockey in Omaha.
  33. L Mitch Moroz. NHLE: 82GP, 3-3-6. Destination: Full AHL season.
  34. L Evan Campbell. NHLE: 82GP, 7-8-15. Destination: UMass-Lowell and another NCAA season.
  35. L Josh Winquist. NHLE: 82GP, 8-11-19. Destination: AHL all year, possibly ECHL if numbers dictate.
  36. G Miroslav Svoboda. Destination: Somewhere in Czech Republic.
  37. L Connor Rankin. NHLE: 82GP, 11-16-27. Destination: AHL/ECHL contract.
  38. L Braden Christoffer. NHLE: 82GP, 9-11-20. Destination: AHL/ECHL contract.
  39. L Aidan Muir. NHLE: 82GP, 4-6-10. Destination: Another NCAA season in Western Michigan.
  40. G Ty Rimmer. Destination: Whatever’s left.
  41. G KevEN Bouchard. Destination: Another QMJHL season with a SP in the .880s.
  42. D Martin Gernat NHLE: 82GP, 1-7-8. Destination: ECHL, AHL or Europe.

written by

The author didn‘t add any Information to his profile yet.
Related Posts

62 Responses to "EVERY GENERATION THROWS A HERO UP THE POP CHARTS"

  1. J-Bo says:

    Is it just me or is anybody else a bit disappointed Nurse isn’t higher on Pronman’s list? My hope is that Nurse becomes not only our number one D but a top 10 D in the league – that Weber, Doughty, Pronger type anchor for the blue-line. I didn’t see the whole list, how many defensemen were listed in front of him?

  2. Samson Loveblast says:

    J-Bo:
    Is it just me or is anybody else a bit disappointed Nurse isn’t higher on Pronman’s list? My hope is that Nurse becomes not only our number one D but a top 10 D in the league – that Weber, Doughty, Pronger type anchor for the blue-line. I didn’t see the whole list, how many defensemen were listed in front of him?

    Not many, Pronman favors forwards and boxcars seemingly. It’s Werenski and Hanifin ahead of Nurse, I wouldn’t get too worried about this list. Nurse also has qualities which can’t be measured in a prospect list.

    Anywho, this is becoming a nice looking list of prospects for the Oilers.

  3. Jordan says:

    J-Bo:
    Is it just me or is anybody else a bit disappointed Nurse isn’t higher on Pronman’s list? My hope is that Nurse becomes not only our number one D but a top 10 D in the league – that Weber, Doughty, Pronger type anchor for the blue-line. I didn’t see the whole list, how many defensemen were listed in front of him?

    It took pronger more than a few seasons in the NHL learning from McInnis before he became the pronger that put a hard-nosed oilers team on his back in the playoffs. And then did it again for the ducks a year later.

    He wasn’t always seen as we remember him in his prime.

    Nurse will take time to be recognized as well.

  4. "Steve Smith" says:

    I have nightmares that “These are the days of miracle and wonder, so don’t cry, baby, don’t cry” could become applicable to the McDavid Oilers.

  5. frjohnk says:

    I do wonder where Pronman would have ranked Draisaitl.

    I believe the top 6 were

    1. McDavid
    2. Eichel
    3. S Reinhart
    4. Marner
    5. Bennett
    6. Nylander

    Draisaitl could be anywhere from 3rd to 7th I would guess.

    Last year Draisaitl was 8th on Pronmans list.
    1. Drouin
    2. Ekblad
    3. Sam Bennett
    4. Nylander
    5. Reinhart
    6. Ehlers
    7. Fiala
    8. Draisaitl
    9. Dal Colle
    10. Kuzneztov

  6. RexLibris says:

    frjohnk,

    I’d wager Draisaitl would have come in right behind Bennett (due to SBs late-season NHL success) but ahead of Nylander on account of his WHL playoff success.

    For me, Bennett and Draisaitl are still neck and neck even though they are not very comparable players stylistically.

    And while Hanifin is a tremendous prospect, I’d argue that Nurse has earned a higher ranking than Hanifin based on what he has accomplished thus far. It isn’t fair to Hanifin because Nurse is further along in his development, but it is fair to recognize what Nurse has accomplished and the steps he has taken in his development over the last two years.

    However, once again I don’t believe the two players are not comparable in style, only by the position they play.

  7. Lowetide says:

    Tough to see Leon’s season as a positive because oilers he was fine they were not

  8. commonfan14 says:

    Pronman said on Twitter last night that he hadn’t hashed out exactly where Drai would be if he were included, but that it would be high – somewhere in the top 10 or 12.

  9. slopitch says:

    Ya I was the one who asked Pronman about Drai’s ranking. He said 10-12 which I feel is a bit low compared to Bennett and Nylander. But ok compared to some guys around 20.

    https://twitter.com/coreypronman/status/632209755952578560

  10. Hammers says:

    What I like is our top ten have 4 centres 3 D and 3 wingers with the goalie at 11 .

  11. frjohnk says:

    I think Draisaitl’s value is lower because of his 37 games with the Oilers. I know Pronman really liked Reinharts jr performance, that is probably 1 of the main reasons Reinhart rose to the top of the 2nd tier of prospects.

    Reinhart
    WHL regular season and playoffs
    54 games 26 goals 49 assists 75 points ( 1.39 points/game)

    NHL
    1 assist in 9 games

    WJC
    7 games 5 goals 6 assists 11 points

    Draisaitl
    WHL regular season and playoffs
    51 games 29 goals 52 assists 81 points ( 1.59 points/game)

    NHL
    37 games 2 goals 7 assists 9 points

    Mem Cup
    5 games 4 goals 3 assists 7 points

    Reinhart was top 3 player of team and named to WJC all star team, 2nd team allstart WHL (East)

    Draisaitl was the WHL playoffs MVP and Mem Cup MVP

    I know there will be comparisions of Reinhart and Draisaitl ( and of course Bennett) for years to come. But I think right now, the gap ( if there is any) is extremely small between these players.

  12. flyfish1168 says:

    RexLibris:
    frjohnk,

    I’d wager Draisaitl would have come in right behind Bennett (due to SBs late-season NHL success) but ahead of Nylander on account of his WHL playoff success.

    For me, Bennett and Draisaitl are still neck and neck even though they are not very comparable players stylistically.

    And while Hanifin is a tremendous prospect, I’d argue that Nurse has earned a higher ranking than Hanifin based on what he has accomplished thus far. It isn’t fair to Hanifin because Nurse is further along in his development, but it is fair to recognize what Nurse has accomplished and the steps he has taken in his development over the last two years.

    However, once again I don’t believe the two players are not comparable in style, only by the position they play.

    I totally agree with you. If we would have drafted bennett he would have failed miserably along with requiring shoulder surgery. We are a horrible at developing players playing at the NHL level and eakins was one of the major causes of this. Leon if he was drafted by the phlegm would be looking like a gem. I believe it is very dependent on who drafted you.

    Both players are going to be great but bennett playing similar to Hall with a slender frame will have a short career.JMHO

  13. RexLibris says:

    flyfish1168: I totally agree with you. If we would have drafted bennetthe would have failed miserably along with requiring shoulder surgery. We are a horrible at developing players playing at the NHL level and eakins was one of the major causes of this. Leon if he was drafted by the phlegm would be looking like a gem. I believe it is very dependent on who drafted you.

    Both players are going to be great but bennett playing similar to Hall with a slender frame will have a short career.JMHO

    I go back to an argument I’ve made here in the past, that is drafting players who complement your core.

    Draisaitl and Bennett had very little to differentiate themselves in terms of pure numbers.

    The Oilers were lacking in size and centers and Draisaitl checked both those boxes as a player-type they lacked who had a very similar skill set to Bennett (in terms of scoring impact relative to their team).

    He was the right choice then and circumstances have conspired to make it an even more appropriate choice today in that the Oilers can now afford to bring him along more slowly and eventually place him in a depth scoring role where he has a greater chance of success.

    The Flames did well by Bennett last year in keeping him on the NHL IR list so he could rehab professionally and they could keep him close to their trainers. He’ll pick up the offensive side of the NHL quickly and I suspect he’ll be closer to Patrick Kane than Gaudreau (a common comparison amongst some Flames fans). However, on the Oilers roster his skill set would be duplicated by a combination of Hall and Eberle.

    He’s a great prospect for the Flames. But Draisaitl is a better fit for the Oilers.

  14. G Money says:

    frjohnk: I know there will be comparisions of Reinhart and Draisaitl ( and of course Bennett) for years to come. But I think right now, the gap ( if there is any) is extremely small between these players.

    In some ‘conventional’ sense of the word, Bennett is more talented than is Drai. Most observers had ranked SB above LD at the draft, some (including that closet Flames fan LT) even having him ranked #1.

    Personally, I don’t think the talent differentiation between the two is arguable – Bennett and Draisaitl had similar numbers their draft years, but Bennett achieved his with a wonky zero-chinups shoulder. I suspect if he were healthy, he would have done better.

    However, it is for exactly that reason that I wanted us to draft Drai, and would rather have him over Bennett.

    Bennett plays Hall’s swashbuckling style (and is a talent probably on par with Hall in fact) … but he has a physique even slighter than Nuge’s. I do not think those two issues are compatible for the long-term, as has already been demonstrated by Bennett’s injury history in the CHL and early in the NHL.

    The biggest risk with Taylor Hall has always been injury, and with Bennett, I think that concern is multiplied at least two fold.

    Draisaitl conversely is an absolute horse, and his playing style would be a lower injury risk even if he weren’t already a massive 215 lbs.

    When both are healthy and on the ice, I don’t doubt that Bennett will be more impactful than Draisaitl will most games. But the whole ‘healthy and on the ice’ thing is nowhere near a sure thing with Bennett.

    I’ll take a long healthy career at .8 to .9 ppg over what I suspect will be a shortened intermittent 1.0 ppg career any day.

  15. G Money says:

    RexLibris: Draisaitl and Bennett had very little to differentiate themselves in terms of pure numbers.

    As mentioned, on an absolute basis this was true.

    Digging in a bit deeper though, Bennett was a. younger, and b. not healthy when he put up those numbers. I think if you age and health adjust them, you would find that Bennett’s numbers are better than Draisaitl’s.

    It’s not all a negative adjustment mind you – you also have to adjust the other way, in that Bennett’s league was higher scoring than the more defensive WHL, and his team provided considerably more support than Dr. Drai’s, so the gap between the two isn’t all that wide.

    But I do think there is a gap.

  16. Bar_Qu says:

    To weigh in on the Sam Bennett vs Leon thing, there was no chance the Oil were ever going to roll the dice on a small skilled forward that high (maybe not ever again) so the fact the Flames took him first meant they got a good player and the Oil got the large skilled center they wanted. I think in the long run Bennett might outscore Drai, but Drai will be the perfect complement to the top 2 centers the Oil have now, even able to step up when/if one of them gets hurt. Which is more important to the Oilers than having another skilled winger.

    My 2 cents

  17. RexLibris says:

    OT but some more information about the explosion in Tianjian here: http://www.theguardian.com/world/2015/aug/15/tianjin-blasts-police-order-mass-evacuations-amid-further-explosions

    Scroll down to the video of the initial explosion. Absolutely incredible. Can’t imagine how terrifying that would be.

  18. Ca$h-McMoney! says:

    The constant discussion of Bennet vs. Drai is interesting, but what I don’t understand about a list like this is why Reinhart is so ahead of either of them.

    In my book, he was the least impressive of the three last year.

  19. RexLibris says:

    G Money: As mentioned, on an absolute basis this was true.

    Digging in a bit deeper though, Bennett was a. younger, and b. not healthy when he put up those numbers.I think if you age and health adjust them, you would find that Bennett’s numbers are better than Draisaitl’s.

    It’s not all a negative adjustment mind you – you also have to adjust the other way, in that Bennett’s league was higher scoring than the more defensive WHL, and his team provided considerably more support than Dr. Drai’s, so the gap between the two isn’t all that wide.

    But I do think there is a gap.

    Going back to your article about elite NHLE, you mentioned that Assists were a better predictor of NHL production than Goals and tailored a formula to represent that.

    Using that formula I came up with extremely close NHLEs for both Bennett and Draisait (I believe it was 39 for Bennett and 41 for Draisaitl, iirc). The league scoring could factor into it, but then it seems we’re splitting very fine differences.

    I think we’re probably eventually looking at comparing a shooting scorer like Iginla vs a playmaker like Thornton. Can’t really go wrong with either one, but my own team building bias takes the player who makes the other four guys on the ice better along with him. Draisailt may be that player more than Bennett. He may not. We’ll have to give it another four years to really determine that.

    Even then, things can go sideways.

    Sam Gagner was the 2nd highest scoring player in his draft year for quite a while until he took a stick in the mouth and his career was undermined by a horrible season.

    I’m analyzing Bennett right now in my Flames RE and it will be intriguing to see the background numbers on him – Hartley is going to give him a very, very soft landing, in my opinion, which will increase the appearance of his skill set.

  20. JD¡™ David O'Connor's Reel says:

    RexLibris: video of the initial explosion

    Another good video of the explosion, but it includes a fair amount of profanity from the videographer:

    http://www.liveleak.com/view?i=0e5_1439474009

    You’ve got to wonder what that chemical storage company was doing, because they have obviously overlooked a few safety principles like segregation and ventilation, not to mention adequate fire suppression. They’ve said today that sodium cyanide has been detected within the blast zone, which just adds to the wtf nature of this incident.

    And then a fracking chemical facility caught fire in Texas yesterday, although not nearly as severe.

    Edit: And this is quite chilling: http://i.imgur.com/qc7FPxa.jpg

  21. Lowetide says:

    Love Leon had Bennett no1 on merit

  22. G Money says:

    RexLibris,

    Using my own work against me! Well played, sir, well played!

    That is a good point – assists *did* carry over more to the NHL than did goal-scoring.

    To my mind, that probably closes the gap a bit more between SB and LD.

    But bear in mind that one thing I was not effectively able to do in my regression was incorporate age. I took a casual swipe at it, but abandoned it – it takes either more mathematical skill or more time (or both) than I’ve got.

    But when I did incorporate it into the machine learning experiment, age (unsurprisingly) made quite a significant difference. I think the SB injury situation has to be recognized as well. That is going to affect both the Jr numbers and the consequent NHL RE.

    (Man, objectivity aside, I hate arguing in favour of anything Flames-related. Is this how you feel when you write articles at FN?)

  23. spoiler says:

    G Money,

    **SPAM ALERT**

    Message for you, sir. Thought you might like this:

    http://ctrlp.press

    If anyone’s into Brit politics and knows Cameron’s propensity for photo-ops feeding baby lambs and holding orphaned kittens, then that piece is an absolute hoot.

    Been a busy little mofo this summer.

  24. G Money says:

    Lowetide:
    Love Leon had Bennett no1 on merit

    I know, but it’s *way* more fun to accuse you of being a closet Flames fan.

    Why do you hate the Oilers?

  25. G Money says:

    spoiler,

    Ha, too funny!

    I’m a little offended by the Koala holding a lightsaber. Seems a bit of a Jedi/Yoda/sizeist microaggression.

    spoiler: Message for you, sir.

    For many years, I had the ’email received’ sound on my Windows machine be the Monty Python clip that has the arrow sound (embedding itself in Patsy’s chest) followed by “Message for you, sir.”

    I had to turn it off when I started approaching a hundred+ emails a day, it wasn’t funny anymore! (Man, hundred+ was over a decade ago. Seems like such a quaintly small number now).

  26. theres oil in virginia says:

    G Money: Why do you hate the Oilers?

    What, you want a rundown of the last two decades?

  27. LadiesloveSmid says:

    Pouliot a comparable for that Hagelin contract?

  28. JD¡™ David O'Connor's Reel says:

    theres oil in virginia: What, you want a rundown of the last two decades?

    Just start at September, 1988, and work forwards from there.

  29. G Money says:

    theres oil in virginia: What, you want a rundown of the last two decades?

    Ah. Right.

    I think I must have blocked that part out of my memory. For some reason, I can’t remember anything about the Oilers prior to April 18, 2015.

    Did they even exist?

  30. G Money says:

    Anyone out there in LT land familiar with neural networks? I have a question for you!!

  31. JD¡™ David O'Connor's Reel says:

    G Money,

    Here’s a place that might have some knowledgeable types: https://www.reddit.com/r/neuralnetworks

  32. G Money says:

    JD¡™ David O’Connor’s Reel:
    G Money,

    Here’s a place that might have some knowledgeable types: https://www.reddit.com/r/neuralnetworks

    I’m fiddling around with a NN just for “fun”, in case I ever want to experiment with using them for, what else, hockey! Was hoping I could get an answer to a perplexing question without going too far afield.

    To wit:

    Using python. I loaded up neuralpy and first set out to implement the ever standard first “hello world” NN example: the exlusive or function. Used a standard 2-3-1 feedforward (sigmoid) trained with backpropagation.

    No problem, got it to converge right away, it behaved exactly as expected. Cool.

    My next test was to see if I could teach “plus 1 module 10”. That is to say, if I give it a number between 0 and 9, it gives me back that number plus 1 remainder10. e.g. 0->1, 5->6, 9->0.

    I configured the network as 1-10-10 (assuming I need one hidden neuron to encode each answer). I had to scale the training data, so my data actually is from 0 to 0.9 in increments of 0.1. Also, since a NN only provides ‘answers’ in a 0 to 1 output, I have 10 output neurons, each corresponding to the digits 0 to 9.

    So if I feed in “0.1”, the output neurons should read: 0.0 0.0 1.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 or as close to 1.0 on the third neuron as reasonable, indicating an answer of 0.2. That’s how my training data is structured.

    But I just cannot get the damn thing to converge!!!

    I’ve experimented with the topology (more and fewer layers, more and fewer neurons) and the training time, but it doesn’t converge no matter what. Doesn’t seem like a problem too complicated for a NN to solve, so I’m a bit mystified as to why this is happening.

    Since this is just for experimenting on a cold and rainy day, I was hoping someone here (I know there’s a handful of techies) might have some familiarity with them and could point me to something obvious e.g. maybe the problem expression is somehow amateurish and dumb or something.

  33. RexLibris says:

    G Money: I’m fiddling around with a NN just for “fun”, in case I ever want to experiment with using them for, what else, hockey!Was hoping I could get an answer to a perplexing question without going too far afield.

    To wit:

    Using python.I loaded up neuralpy and first set out to implement the ever standard first “hello world” NN example: the exlusive or function.Used a standard 2-3-1 feedforward (sigmoid) trained with backpropagation.

    No problem, got it to converge right away, it behaved exactly as expected.Cool.

    My next test was to see if I could teach “plus 1 module 10″.That is to say, if I give it a number between 0 and 9, it gives me back that number plus 1 remainder10.e.g. 0->1, 5->6, 9->0.

    I configured the network as 1-10-10 (assuming I need one hidden neuron to encode each answer).I had to scale the training data, so my data actually is from 0 to 0.9 in increments of 0.1.Also, since a NN only provides ‘answers’ in a 0 to 1 output, I have 10 output neurons, each corresponding to the digits 0 to 9.

    So if I feed in “0.1”, the output neurons should read:0.0 0.0 1.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 or as close to 1.0 on the third neuron as reasonable, indicating an answer of 0.2.That’s how my training data is structured.

    But I just cannot get the damn thing to converge!!!

    I’ve experimented with the topology (more and fewer layers, more and fewer neurons) and the training time, but it doesn’t converge no matter what.Doesn’t seem like a problem too complicated for a NN to solve, so I’m a bit mystified as to why this is happening.

    Since this is just for experimenting on a cold and rainy day, I was hoping someone here (I know there’s a handful of techies) might have some familiarity with them and could point me to something obvious e.g. maybe the problem expression is somehow amateurish and dumb or something.

    42.

  34. RexLibris says:

    Lions 1.

    TiCats 31.

    When I think about Hamilton fans I recall that Tom Cochrane line “after all the shit you know we’ve been through it, there ain’t a shovel big enough in the world that could move it”.

  35. G Money says:

    RexLibris: 42.

    I tried that. The thing just got all mopey and paranoid on me!

  36. G Money says:

    RexLibris,

    The Cats are definitely looking dangerous this year, no question.

  37. Pajamah says:

    G Money,

    No team can beat the Eskimos defense.

    Now if Coach Jones can figure out a way to keep his offense off the field.

    ….funny, my phone autocorrected offense to “offends”. How apt.

  38. RexLibris says:

    G Money,

    I admire your Saturday afternoon craft.

    I’m pursuing personal interests of a slightly different manner.

    Just got back from Sherbrooke liquor and I’m deciding between a porter and an orange wheat ale for beer bread. Once I’ve finished making the homemade pizza I’ll mix up the bread, checking the quality of the “ingredients” as I go of course, and have it ready for breakfast tomorrow.

    Days like these are made for this kind of stuff.

  39. RexLibris says:

    G Money: (Man, objectivity aside, I hate arguing in favour of anything Flames-related. Is this how you feel when you write articles at FN?)

    I think the age thing between Bennett and Draisaitl gets minimized due to development time lost to injury, as you mention.

    But, no, I actually don’t mind writing about the Flames.

    Here’s the thing.

    I used to get so pissed off everytime the Flames or Stamps beat the Oilers or Esks, and let’s be honest, it happened a lot, that it would ruin my whole day.

    I got sick of giving them that power and was angry at myself for being that kind of fan.

    When a window to write for the Flames opened up, by circumstance and my own sadistic interest as I recognized they were closing in on a collapse, I took it because I wanted to use my outsider’s perspective to bring a different voice to the blog and I relished the opportunity to write for what I knew would be a highly critical audience.

    I can’t say that I’ve never struggled to write objective but positive things about the team or organization, but I’ve always tried to be a better kind of fan and a better kind of person if only to spite those damned YYCers.

    I’m expanding the RE series I did last year to include this year’s version of the Canucks and Oilers and it is proving to be an interesting exercise to maintain a balanced approach and to challenge one’s biases and expectations. It should prove a very intriguing experience for fan response.

  40. RexLibris says:

    Pajamah:
    G Money,

    No team can beat the Eskimos defense.

    Now if Coach Jones can figure out a way to keep his offense off the field.

    ….funny, my phone autocorrected offense to “offends”. How apt.

    Imagine what we could accomplish if Reilly were healthy right now.

    Bell and Chipman have done well to replace White.

    They need to find a fill-in for Bowman. Stafford is doing relatively well but they need more. Chambers’ return will help a lot.

    But if Reilly and his dynamic skill set were available we’d really be having some fun.

  41. G Money says:

    Pajamah,

    Yes, the Esks D is looking awesome, no question.

    Ham will be a really good test next week.

    The flatlanders looked good statistically, and we already know they couldn’t handle the Esks. Then its Ham and Tor, and we have a loss to those guys in blue. They look to be having the kind of season the Flames had last year!

    If the Esks can do what they’ve done on D to Ham and Cgy (who always seem to play their best against us), I will be a true believer. Right now I’m still a wee bit gun shy.

  42. RexLibris says:

    JD¡™ David O'Connor's Reel: Another good video of the explosion, but it includes a fair amount of profanity from the videographer:

    http://www.liveleak.com/view?i=0e5_1439474009

    You’ve got to wonder what that chemical storage company was doing, because they have obviously overlooked a few safety principles like segregation and ventilation, not to mention adequate fire suppression. They’ve said today that sodium cyanide has been detected within the blast zone, which just adds to the wtf nature of this incident.

    And then a fracking chemical facility caught fire in Texas yesterday, although not nearly as severe.

    Edit: And this is quite chilling: http://i.imgur.com/qc7FPxa.jpg

    I saw some drone footage of the blast area from a few days ago.

    The Chinese are learning a number of lessons that many of the Western industrialized nations learned the same (hard) way many, many years ago, albeit likely on a less dramatic and reported scale.

    From a purely academic perspective, I’m intrigued by the exercise the Chinese government faces now in regards to message control and the balance it chooses between response and reaction.

    I expect the response to be very technical, trying to return the area to industrial productivity as quickly as possible.

    The reaction, however, will likely have more political connotations. There will be scapegoats, of course. The most recent inspector to that site prior to the explosion is a likely candidate to be tried and found guilty. Provided they aren’t too close to the Communist Party’s inner circle, the factory owners are another potential target.

    The government has shown in the past a willingness to engage in populist rhetoric, drawing on anachronistic People’s Republic language to divert criticism and cynically present individuals as the sole responsible parties.

    We’ll wait and see what happens.

  43. G Money says:

    RexLibris,

    Your craft sounds much tastier than mine!

    Good news though, I figured it out!

    After digging through the literature a bit, it seems that Neural Networks do not like to discern fine gradations in a single input neuron. They like things in black and white as much as possible. It doesn’t actually *say* this anywhere, but all the training data examples being used very much tend to break the input down so that most input neurons are either as close to 1 or as close to 0 as possible across all the inputs, without too much discernment about values in between.

    So I recast my problem so that the input digit was passed in to four neurons as a binary number. That is to say, 0 -> [ 0, 0, 0, 0 ], 1 -> [ 0, 0, 0, 1 ], 5 -> [ 0, 1, 0, 1 ], 9 -> [ 1, 0, 0, 1 ], etc.

    That made the different digits very distinctive, and a 4-12-10 network (that is to say, 4 input neurons, 12 in the middle layer, and 10 output neurons) was able to train up to effectively learn the answer to the problem in just a few seconds, about 300 iterations.

    Sweet!

    It seems you *can* teach the rickety old neural network inside the head of an Old G Money new tricks!

  44. RexLibris says:

    G Money:
    Pajamah,

    Yes, the Esks D is looking awesome, no question.

    Ham will be a really good test next week.

    The flatlanders looked good statistically, and we already know they couldn’t handle the Esks.Then its Ham and Tor, and we have a loss to those guys in blue.They look to be having the kind of season the Flames had last year!

    If the Esks can do what they’ve done on D to Ham and Cgy (who always seem to play their best against us), I will be a true believer.Right now I’m still a wee bit gun shy.

    I’m not worried at all about the D.

    I’m more interested to see what kind of play calling they use for Franklin and Lynch. If they get creative they could really play those two off each other, using their difference skill sets, to keep the opposing D on their heels and guessing all night long.

    If they go with a more conservative approach we’ll get killed because Lynch will get hammered on the run and Franklin is still learning to throw the ball away rather than eat it.

    Ray spent his first season learning the same lesson. He was so patient and calm he refused to let a play die so he was always getting taken down or forced out of bounds. After his Jets camp he came back and his willingness to throw it away and fight on another down was far more mature.

  45. RexLibris says:

    G Money,

    So without giving away anything you may be working on, to what end are you conducting this experiment?

    And from someone who is only very tangentially connected to programming language, from where does the term Neural Network derive? Is this intended as a software exercise designed to best mimic some decision-making or discerning properties of the mind?

    From here it sounds like it could be adapted to be applied to making strictly objective decisions on a series of comparison situations.

    Or have I been making too much beer bread?

  46. RexLibris says:

    G Money: Your craft sounds much tastier than mine!

    Wait until I make the Dr. Pepper brownies.

    Always a favourite.

  47. G Money says:

    RexLibris,

    Very true. I do give the O a bit of slack due to the minor issue of missing QB1, WR1, and RB1!

    That said, the offense still looked OK previously because Bell/Simpson were doing OK filling in for White, and with a running game behind him, Nichols looked OK at times.

    But the fact that the O was able to do almost nothing when Bowman went down is very disturbing (but not surprising given e.g. his numbers against BC). He’s clearly the go to guy.

    Despite the kudos being tossed at Montreal’s D, the reality is that they aren’t very good. 6th in total yards, 7th in passing yards, 5th in running yards against.

    They are a mediocre to below average defense, and the Esks were stymied against them most of the game. Against a truly dangerous defense, that game might have been a big loss.

    I recognize that a Reilly-led offense with a healthy Bell and a healthy Bowman may very well have shredded them.

    But I also think that Reilly is like Taylor Hall – game changing player, but health is always a concern.

    I’d say at this point, Juan Blanco and Shake n Bake Bell are showing the same concerns.

    So it might be problematic to talk about how well the O would do with them in the game – because we might not see a lot of them.

  48. Pajamah says:

    RexLibris,

    I have an almost irrational hate for Nichols. A hatred of Kerwin Bell levels.

    I liked Walker in the 2nd half against the Als. Coehorn and Bowman are missed, Bell has done great for White, and no QB can hold a stick to Reilly.

    That BC loss was infuriating, and I was cursing the offense hoping for first down punts against Montreal.

  49. G Money says:

    RexLibris,

    Right now it’s just one of those things about learning some tools that might be fun (and eventually useful) to apply to the hockey world.

    I see a possible use of NN’s in the always-interesting world of draft assessment and projection.

    I also can’t help but find myself mulling on the problem of assessing defensemen. We know the best and the worst defensemen, but really, when it comes to the vast majority of them, we have no objective measure at all. CA is questionable, EVGA is terrible, etc. WG’s Fenwick idea the other day has promise, but still has some of the challenges we see in using CA.

    A neural network is called that because it’s based on biological neural architectures. Basically, you have a bunch of “neurons”, each of which is like a simple computational machine that sums up its inputs, applies some weights to those inputs, then produces an output.

    When you layer these things, where one layer of neurons is fed into a second (and third and fourth – you can get as complicated as your computational power allows you to), the behaviour of this network seems chaotic – and it is at first.

    But there is a technique where you train the network, where you give it input, compare the actual output with what you want as output for that particular input, and then feed the result backward through all the neurons, where each of them adjust a tiny bit to get closer to the desired answer.

    If you structure the thing right, the network can eventually learn *very* sophisticated patterns this way. Chances are, the software that is filtering your email spam, or assessing whether your credit card transactions are fraudulent, could very well be neural networks.

    I’m pretty sure the part of the “self driving car” technology that maps camera inputs to an actionable internal 3d world map are almost always neural networks.

    It’s cool stuff, yet at its core, built on an amazingly simple concept.

  50. Oddspell says:

    Question for lowetidians:
    I was talking with a friend about the goaltending improvements this year and how it should lead to a marked improvement in the team. Now obviously the Oilers team defence data is a tough sell to someone who hasn’t any familiarity with advanced stats. It got me thinking as to why the Frye test and nnumbers seem to suggest such different situations and I came to the following thought. Would score effects affect the team defence scores from last season?

    I haven’t done any work on the topic myself however I’m curious as to whether the topic has been discussed here and taken into account.

  51. G Money says:

    Interesting!!!

    I typed up a long reply to Oddspell. Then I edited it for typos and added another link to a goalie article on BlueShirtBanter, and BAM! I’m told it’s now spam, and the post was deleted.

    I then typed up this reply (which included the link to BSB, and it got deleted outright).

    Seems the link is now completely toast. Or maybe I can no longer post links because I’m a spammer??

  52. G Money says:

    It’s just the blueshirtbanter link that seems to be treated as spam. Weird. It’s a good article!!

  53. wheatnoil says:

    G Money,

    Oddspell,

    I also typed up a long response that got deleted because I’m an idiot and closed my browser. To summarize, the Oilers trailing so much should lead to a slightly lower save percentage, but that should be at least partially due to allowing more high danger shots against, which we know they did not disproportionately do.

    The following link is the first part in a series that looks at how score effects impact shooting and save percentage.

    http://puckplusplus.com/2015/08/10/examining-the-drivers-of-score-effects-i-shooting-and-save-percentage/

  54. wheatnoil says:

    Huh… weird. I also just tried to put up a Blue Shirt Banter link, assuming it was the one you wanted to put up G Money, and the post just got swallowed up.

    LT, why do you hate Nick Mercadante?

  55. RexLibris says:

    LT hates the Rangers so much he’s marked the words Blue Shirt as spam.

    Whoda thunk it!

  56. RexLibris says:

    G Money:
    Test.

    Failed.

    Study harder and try again.

  57. Lowetide says:

    G Money:
    It’s just the blueshirtbanter link that seems to be treated as spam.Weird.It’s a good article!!

    I don’t have anyone in block so this appears to be a weird thingy.

  58. stush18 says:

    No post this evening? Not that you don’t enough work LT, just used to my normal reading habits.

  59. Lowetide says:

    stush18:
    No post this evening? Not that you don’t enough work LT, just used to my normal reading habits.

    Yes, it’s up now.

  60. RexLibris says:

    Lowetide: I don’t have anyone in block so this appears to be a weird thingy.

    Could you use non-technical terms for the rest of us. 😉

Leave a Reply

Want to join the discussion?
Feel free to contribute!
© Copyright - Lowetide.ca