# THE FIFTH ROUND

By the fifth round, we’re looking at about a 12% chance of seeing an NHL player. You’d have to pile 8.3 picks one on top of another to ensure one player arrives on your doorstep, so we’re in the ‘once a decade’ territory of the draft.

## A PROPOSAL

Scott Cullen of TSN estimated the success rate (100 NHL games or more) to be 12% back in 2009. Let’s also use 150 games as the line in the sand, as we’ve done through each round.

## OILERS FIFTH ROUND 2008-2014

That set, let’s move on to the Oilers during the MacGregor era (2008-2013). We’re looking at one in eight being a success (average) and anything more than that above average. The Oilers have selected seven players in Round 5 during the MacGregor era, including one last month:

- 2008: Phil Cornet
- 2009: Olivier Roy
- 2010: Tyler Bunz
- 2011: Martin Gernat
- 2012: Joey Laleggia
- 2013: Evan Campbell
- 2014: Liam Coughlin

Seven players have been drafted by Edmonton during these years. 12% of twelve players is one NHL player (84% of a player) of 150 games or more from this group, that’s the expectation for an “average” team in the fifth round.

**Tyler Bunz-Rob Ferguson (all rights reserved)**

## NHL GAMES PLAYED

- Phil Cornet 2

Like the fourth round, the Oilers have (so far) struck out in the fifth round. Unlike the fourth round, where Dillon Simpson and Tobias Rieder are trending well, the fifth round offers only Martin Gernat as a promising NHL prospect.

## 2014 SUMMER

- No longer a prospect:
- Very bad arrows: Phil Cornet, Olivier Roy, Tyler Bunz
- Bad arrows: Evan Campbell
- Lukewarm arrows: Joey Laleggia
- Good arrows: Martin Gernat
- Very good arrows:
- New arrows: Liam Coughlin

Players selected this late don’t often have long careers anywhere, so a guy like Cornet is probably going to slide out of hockey (or go to Europe) in the next year or two. Goalies are like knuckleball pitchers, they do nothing for awhile and then perk up their careers. Defensemen Gernat and Laleggia are the best hopes here, Gernat being the only one in this group to put in an AHL season of note.

- Phil Cornet: Looked like a tweener on draft day, and that’s what he is; passed 200 AHL games this season.
- Olivier Roy: Famous junior goalie has been around .900 as an AHL goalie.
- Tyler Bunz: Final year of entry level deal on the way, young man has played 6 AHL games.
- Martin Gernat: Was an AHL regular for most of his rookie pro season (57 games).
- Joey Laleggia: NCAA offensive defenseman, numbers faded in year three.
- Evan Campbell: Older prospect started very slowly in first NCAA season, scored 9 goals after Christmas.
- Liam Coughlin: Power forward from BCJHL.

(all OKC photos by Rob Ferguson, all rights reserved)

http://www.hockeydb.com/ihdb/stats/leagues/seasons/teams/0005712014.html

Sometimes wonder how the Chris McCarthy’s and Mario Puskarich’s of the world don’t get rookie camp invites.

I’ve never seen them play, but my guess would be skating is an issue for one and size is the issue for the other.

Reider stings because he was one of my favourite prospects and he was traded for an absolutely crappy prospect. I also fully expect Tobias Reider to play 500+ games in the NHL, because it just wouldn’t be Oilers like if the move didn’t bite them in the ass for the next 10 years.

Boy Tambellini sure hated good hockey players.

The Oilers and their prospect goalies

http://oilersnation.com/2014/8/16/the-goalie-of-the-future-where-in-hell-is-he

I think you might be making an incorrect conclusion about these statistics. You seem to be saying that if you make ten picks, one of them should turn out to be an NHLer. That’s not the way it works. Each player has a 12% chance of making it (league average for this round). A team could make 100 years of picks in this round and get no one, but each of them would have the same 12% chance.

dcsj,Oh spare me.

You know, this blog and its posters take some measure of pride in this being a ‘numbers friendly’ place. Part of that means making an effort to understand how ‘the numbers’ work. Go ahead and spare yourself that effort, but I suggest keeping the wiful ignorance to yourself.

Technically, this is true, in the sense that you can also flip a coin and hit heads on the first one, or not hit heads in a 100 straight flips. But it’s not a very useful statement either. If you go 100 years of such picks without a hit, chances are, something’s wrong and you should fix it!

Statistically speaking, the way you might ask this question is: “Given ‘n’ fifth round picks, what is the likelihood that I will get at least one NHL player?”

When asking a question like this (“at least”), it’s usually easier to ask the reverse question and then invert to find your answer. So what you’d ask is: “Given ‘n’ fifth round picks, what is the likelihood that I get no NHL players at all?” You can then flip the answer to find your likelihood of getting at least one.

The chance of NOT getting an NHL player in the fifth round is 88%, so the numbers shake out like this:

1 pick = 88% none = 12% at least one

2 picks = .88*.88 = 77.24% none = 22.56% at least one

3 picks = .88^3 = 68.1% none = 32.9% at least one

By the time you get to LT’s standard of 8 years:

8 picks = 36% none = 64% at least one

9 picks = 31.6% none = 68.4% at least one

and by the time a decade has rolled by you’re at a 27.9% chance of not getting even a single NHL player, which means a 72.1% that you’ll get at least one (or more) NHL players.

So the odds are in your favour at that point, but at 27.9%, there’s still almost a 1 in 3 chance you’ll come up empty.

The math isn’t obvious because the pick sequence generates another probability distribution, but simulation is usually convincing. Imagine picking 12 times each with uniform 1/12 chance of success. You might hit 0, 1, 2, or even 12 home runs in any sequence of 12 years of picking. I just simulated 1000 teams picking for 12 seasons and got (numbers rounded):

34% had 0 picks turned out

41% had 1 pick turn out

18% had 2 picks turn out

6% had 3 picks turn out

1% the rest

Say we follow the last 12 Oilers 5th round picks and say they have 0 picks turn out. Is that bad? Just by random luck 34% (just over 1/3) of the teams will be just like this. Are you willing to bet they’re just bad or just unlucky? There’s just too little evidence to say with much confidence that they would be worse than a random picker. I mean, it’s fun to look at it, but I daresay one would not want to make strong statements about the record.

In the case of the 7 picks LT notes, there’s about a 54% chance that a random picking team comes up empty in 7 picks. And about 90% that the random picking team gets either 0 or 1 players. The 8.3 number LT cites basically is the 50% barrier: once you look at 8 pick sequences the probability of having 0 successes falls under 50%.

Now, if they had 4 or 5 picks turn out, maybe we say they either got stupendously lucky or they can pick out players better than random in that round.

oilswell,Hey, nice job running that Monte Carlo simulation, but for cases like this, I find it’s easier (and more accurate) to come up with the closed form solution!

I have used Monte Carlo’s to good effect in trying to determine how to compare the track record of different scouting departments, specifically how long it would take to conclusively determine that one scouting department (that was 10% better than the other) was actually demonstrably better in track record. Answer: given the enormous random variation involved (as we’ve shown above) and small sample sizes, the correct answer is “over 100 years”.

You know where else I find Monte Carlo’s most useful?

Goalies. Always those f*ckin’ goalies. Closest thing in hockey to a random number generator.

Yeah, the simulation closes in really quickly though. At 10k sequences there’s very little variation. But I may have wronged it because I get 8 picks generating just about 50% chance of 0 picks turning out compared to your 36%. I used 12% chance for each pick, uniform random distribution.

OK. But can you guys tell me definitively how many angels can dance on the head of a pin? Or why Maggie the Monkey was as smart as Yogi?

Better yet, read what LT wrote more more carefully. There are assumptions and presumptions. He says “an average team”. But you guys speak in terms of “statistically” and “technically”. You run Monte Carlo simulations on your Univacs.

Me, I like to read about hockey.

Oh look. It’s raining pendants.

Well since McCarthy is already attending Blackhawks rookie camp as an invite, I’m just going to have to set my sights on Chase Balisy and Isaac MacLeod!

I like Gernat. Like him a lot. Has size, and brought tons of offense as an Oil King which is translating to the pro’s. Would love to see what he could do with a lot of pp time on the Barons this year.

This defenseman is like a flank steak – will need to marinate for a long long time, but so long as Todd Nelson can continue to help players like him hone their defensive craft, you’ll be glad you waited.

This is a fun summertime exercise. That said, analysis based on games played shouldn’t be looked at as proof of draft success or failure. For me, I like picks in rounds 5 and up from less scouted leagues like BC or Slovakia. I like lottery tickets that have a comparatively high chance of acquiring an impact player. No surprise I liked the Gernat pick when it was made and I still do.

Nice exchange on the statistical chance.

This is the round I see the scouting staff taking there walkabouts.

I prefer this here rather than round 1, or round 3.

Self-pitching.

http://www.theoilersrig.com/2014/05/speakin-piece-evaluating-scouting-isolation/

Some time back I wrote about how we should consider expanding our thinking when it comes to draft picks, esp. the late rounders.

The NHL success rate doesn’t give us much information. We should try to distinguish between the players that bust out of hockey altogether and those that end up with success pro careers elsewhere, for example.

I love coming here and reading a statistical analysis of Oilers-related topics. The mathletes are doing the heavy lifting for me, and I appreciate the effort.

Nothing wrong with knowing some stats and probabilities when watching the games. For instance, knowing that Sam Gagner had a 79% chance* of missing his defensive assignment and being the goat on the ice for a goal against certainly helped keep my blood pressure down last season.

“Statistics found on the Internet may not necessarily be true.” – Abraham Lincoln, Gettysburg Address

Sorry to make a comment and run, but I’ve been away all day. Thanks for the more detailed explanations, I understand some of it.

I think that drafting is not purely a numbers game since there are thinking humans making choices based on information. It’s not just random chances, someone who is good at it probably is slightly better than someone who is not so good at it. So if the average is 8% good players in a particular round, a good drafter is probably say 9 or 10% in that round (over time). Would that make sense?