They did have Lehner as their #1 for a couple years in there. Risto shouldn't be a #1 (at least right now. In 3 or 4 years, who knows?), but he is a very good #2 or #3. He would improve Toronto's D immensely. His loss in Buffalo would be significant. Just as Buffalo's main need right now is a 2C (having traded away ROR) their main need if they trade Risto will be a D like Risto.
Too many people have put their faith in the authority of algorithms. I love stats. I am not a statistician, but I have a very strong math background and while it not my specialty I have taught stats off and on at a post-secondary level for years. There is the old saying: "There are lies, damn lies and statistics" which is extremely true. Bad stats are used to bolster weak arguments all the time. We all do this. Some more than others. When I came back to watching hockey after many years away I was thrilled that there were now advanced stats, although they were still in their infancy. But at the same time I knew that the advanced stats folks were saying that they really had no way of adequately assessing D. So then they started making more complicated algorithms to look at things like relative to teammates to get at those questions, but their results are extremely flawed.
So the first question should be - how accurately are advanced stats recorded. Rachel Dorrie has said that when she was working for the Devils their own numbers through their own tracking did not accurately align with publicly available statistics. Having tracked several games myself, I find this to be the case. Even things like simple faceoff wins/losses I have seen just baffling examples of players being awarded wins when, to me, they clearly lost. I have watched games when recorded hits seemed to be completely out to lunch. When it comes to things like CF, SCF, HDCFs that all gets harder to record accurately.
But even if all those stats were recorded completely accurately, these numbers only lead to questions, not answers. Dom L. is a smart guy. I remember reading an article by him a couple years ago in the hockey news about Risto. He looked at the top 3o or 35 D in terms of ice time using a couple metrics and determined that among them Risto was near the bottom in terms of ability, especially defensive ability. He said that he was probably a decent bottom pairing D. Now first of all, being near the bottom in terms of ability of the top 30 or 35 TOI D in the NHL does not make you a 3rd pairing D. He also admitted that Ekholm finished below Risto, but stated that Ekholm was a very good D for reasons that he did not give or explain.
He said that people could make the case that Risto faces harder competition, but his algorithm takes that into account. What algorithm? What does it take into account? How does it take it into account? Two D can face identical QoC with one jumping on the ice every time every time his own team gains possession and jumping off each time his own team loses possession and the other D jumping on the ice every time his own team loses possession and jumping off each time his own gains possession. The OZ starts wouldn't be different either as these are changes on the fly. Completely different situations. The algorithms view them the exact same. One should have great advanced stats and the other horrible advanced stats and they are both completely meaningless.
Dom also said that Risto's Relative to teammates stats were bad compared to the other D on his team meaning that the other D were carrying him and not the other way around. Other advanced stats proponents make this argument all the time. It is a completely dubious argument. If you are the #1D, especially on a bad team, and matched up against the opposing teams top players (which is not the case for #1D on all teams - as many teams simply play their #1D with their own top lines, or even try to match their second pair against the opposing teams' top players - again the algorithms can not account for any of this at all) then your teams' other pairings know that when your opponents' top players hop over the boards either you or your partner (usually which ever is closer) head off so your #1 can go on. In that case whenever you are on with your teams' #1D you are facing tougher competition than when you are not on with him. Of course your relative numbers should show that you are "better" when you are on the ice without him. This is especially the case for weak teams that lack depth on both forward and defense. And we have seen this over and over again with great veteran Ds who play with young inexperienced D where the rookie only plays the easy minutes and the vet is paired with other D for the tougher minutes. Algorithms dramatically over-value the young sheltered D and claim that the very good vet is now bad. This is first year cause and effect stuff when evaluating the validity of statistics that the advanced stats people don't consider. In part that is because while they claim that advanced stats gets rid of bias it doesn't and they are often extremely biased. When the Leafs' acquired Ceci the advanced stats people said, as they had been saying for years, that Ceci was the worst defensive D in the NHL. They were, of course, wrong. Ceci is not a great D (Risto is far better), but he is not a bad D, and he was heavily relied on by Toronto in a defensive role and did very well. The advanced stats people were too biased and closed-minded to change their view. But the sad thing is that Ceci joined a new team with a large chunk of the fan base already having their mind made up about him because of flawed numbers.
Last edited:
