Winning Streak Stats

[SnowblindNYR]

I was just having a debate on the Rangers board with a poster that said the Rangers won 3 in a row and teams in the "recent NHL" don't usually win 4 in a row, so they'll likely lose. I told him that "in a vacuum", if every game had the same probability of winning, you're as likely to win the 4th game as any other game that season. However, this is not a coin flip and it's not roulette. The NHL has many different factors. So is there any stat that shows that there is a point in a winning streak that a team tends to be less likely to win the next game than it won on average that season? For example, a .500 team after winning 4 in a row would win the 5th game only 40% of the time. The factors that can factor into this in my head are as follows:

1) Quality of opponent, home game/road game, schedule density (i.e. back to backs). Hopefully with a large enough sample size this factor gets diminished as the quality of opponents and type of game will average out.

2) Psychology: A team on a slight winning streak might get a little more full of themselves and not pay as much to detail. This can also help as maybe they can have momentum or a positive amount of confidence (kind of like a team on a losing streak is missing confidence).

3) Exhaustion: A team on a winning streak may be exerting themselves more than usual to play as well as they have, and be emotionally or physically more spent.

I'm sure there are other factors that might skew the results. Anyone know if there is a point in a winning streak where likelihood of winning the next game diminishes? Thanks!

Click to read more...

 

[Garfinkel1]

I was just having a debate on the Rangers board with a poster that said the Rangers won 3 in a row and teams in the "recent NHL" don't usually win 4 in a row, so they'll likely lose. I told him that "in a vacuum", if every game had the same probability of winning, you're as likely to win the 4th game as any other game that season. However, this is not a coin flip and it's not roulette. The NHL has many different factors. So is there any stat that shows that there is a point in a winning streak that a team tends to be less likely to win the next game than it won on average that season? For example, a .500 team after winning 4 in a row would win the 5th game only 40% of the time. The factors that can factor into this in my head are as follows:

1) Quality of opponent, home game/road game, schedule density (i.e. back to backs). Hopefully with a large enough sample size this factor gets diminished as the quality of opponents and type of game will average out.

2) Psychology: A team on a slight winning streak might get a little more full of themselves and not pay as much to detail. This can also help as maybe they can have momentum or a positive amount of confidence (kind of like a team on a losing streak is missing confidence).

3) Exhaustion: A team on a winning streak may be exerting themselves more than usual to play as well as they have, and be emotionally or physically more spent.

I'm sure there are other factors that might skew the results. Anyone know if there is a point in a winning streak where likelihood of winning the next game diminishes? Thanks!

What do you think the odds are of flipping a coin and getting heads 4 times in a row? The probability of that single event is 50/50 but the probability of the event sequence is much much lower... (.5)^4 I believe. I'm not really sure there is a right answer. Data will be your only friend but I would imagine the sample size is so small those numbers won't tell you much.

 

[Tawnos]

As I posted in that discussion we were having.

Last season, there were 85 winning streaks of 3 games or more. 41 of them made it to 4 games. I don't know how that compares to the way things were when there was less parity in the league than there is today, but my sense is that there would be more 4 game streaks in the past. Actually, my sense is that there would be more 3 game streaks as well.

By the way, my original point in all of this was that I wasn't going to get my hopes up for the Rangers to win last night, after having won the previous three. The data is interesting, but I wouldn't change my mindset on that even if it were proven that 4 game winning streaks are relatively easy to attain. Especially being a Rangers fan.

 

[Bear of Bad News]

It'd be interesting to test the options that you've described (and I think that the sample size would be there).

The question would be what other obfuscating variables are there that are being ignored?

 

[SnowblindNYR]

What do you think the odds are of flipping a coin and getting heads 4 times in a row? The probability of that single event is 50/50 but the probability of the event sequence is much much lower... (.5)^4 I believe. I'm not really sure there is a right answer. Data will be your only friend but I would imagine the sample size is so small those numbers won't tell you much.

Yes the probability of flipping a coin 4 times in a row is in fact 0.5^4. Only problem is that sports is not a coin. There are too many factors, psychological and physical. I also forgot, just plain old luck. Say in an obviously unrealistic scenario a team was somehow judged to have the skill to win every game 75% of the time. Say that team wasn't affected by psychology (over and under confidence for example) or fatigue. Also didn't matter what the strength of the team they were playing. There's still luck involved. Bounces even up talent disparity regularly in the NHL. However, the hope is that luck, like quality of opponents gets evened up in a large enough sample size.