Another playoff conundrum

[Midnight]

It's be interesting to compare the 'mismatches' and the more evenly matched. For example, in 3-6 and 4-5 series' compared to 1-8 and 2-7 matchups. I could see the even matches depending more on the first goal, or the mismatches showing a strong trend because the good teams will score first.

While I'll agree with this, this is somewhat accommodated by my analysis. If teams are 1 vs. 8, they are going to generally have a big discrepancy in their goals for to goals against ratios, which is how I obtain the "probability of the next goal" shown on my x-axis. I know there are people here who are against accumulating goals (you know - the one or two big games against lousy opponents "skew" the results). There are good arguments to counter that kind of thinking; modeling that kind of thinking gets complicated; and finally, results from analysis of MLB scoring (I've never seen it applied to the NHL) show that if you want to predict future W/L records, you will do better using past runs scored/allowed than using past wins/losses.

 

Another example: I'd much rather know a pitcher's ERA than his W/L, even if on the bottom line all I care about is W/L. Past ERA is a better predictor of future W/L records. W/L has a lot of fluke variability in it (like: your own team's offense varies a lot).

 

As for the even matches: That's exactly what my analysis shows! The discrepancy between the Green and Red curves dwindles as you get further from x = 0.5.

[Midnight]

Whether or not you like (or even get) what I've proposed . . . Here's what the pre puck drop chances are for each of the series, as per the analysis above. I don't think its that unreasonable for such a simplistic method (based on goals scored/against). Yeah - Detroit looks better than 50.5% to me too - but I don't claim to have a method to incorporate late season rallies (nor do I think such a method would add much if anything). Some of these - I'll predict 2.5 of them - will be wrong. (So will some experts.)

 

[Eleven]

You complete me.

 

We need a Venn diagram for the realms of carp and midnight.