Quick study on streaks (baseball)

[DougOzzzz]

I got into an argument with a friend of mine at work over the state of the Empire (aka the Yankees). He seemed to believe that since the Yankees have played better recently, that they are likely to continue their success and outperform their projected W-L based solely on record.

So, I did a quick study. This was mostly for fun - and also to get my feet wet working with Retrosheet files.

Anyways, what I did was obtain a list of all teams since 1960 that went on 10+ game winning streaks. I then compared the record after the 10th consecutive win to their record for the rest of the games in the season.

In total, I came up with 137 teams that won 10 consecutive games. These teams averaged a record of 48.83-34.82 after the 10th win (some of the teams went on to extend their streak beyond 10 games) for a winning percentage of .584. The rest of the season, the average record was 40.387-36.109 - a winng percentage of .528. The average of just over 160 games/season is mainly because of 5 teams in the strike shortened seasons of 81, 94, and 95.

Of course it is not surprising that teams coming off 10 game winning streaks tend to do worse the rest of the season - this is simple regression to the mean.

To do a comparison, I tested the 43 teams that started the season 48-34 since 1960 and summed their Wins and losses over the rest of the season. They totaled 1621 wins and 1464 losses for a .525 winning percentage - or about 2 points below the rest of the season percentage for the "hot" teams. In other words, the "hot" teams performed basically the same as all teams with identical records, regardless of their current streak.

So, don't fool yourself into thinking the Yankees are out of the woods yet. They are what they are - a team 1 game over .500 through a quarter of the season. And though their past season's success DOES mean they are likely to perform better in the future, the winning streak has very little relevance to projecting their future W/L record.

[shemp]

[ QUOTE ]
To do a comparison, I tested the 43 teams that started the season 48-34 since 1960 and summed their Wins and losses over the rest of the season.

[/ QUOTE ]

.585 is damn good. And looked at over the 92 games from the start it's .630, which is damn damn good. A lot of it has to do with scheduling. Also we've seen teams like this year's white sox jump out of the box and struggle to play .500 the rest of the year -- and not been surprised by that.

Which is a long way to say, I'm not sure that the hard analysis of data gets you much. I look at the roster. The Yankees' roster has several problems in my opinion which I've alluded to elsewhere. I think they'll be competetive, and while they could win that division, I don't think they will. In any case, they've been on a good run, and hats off to them.

[DougOzzzz]

[ QUOTE ]
[ QUOTE ]
To do a comparison, I tested the 43 teams that started the season 48-34 since 1960 and summed their Wins and losses over the rest of the season.

[/ QUOTE ]

.585 is damn good. And looked at over the 92 games from the start it's .630, which is damn damn good. A lot of it has to do with scheduling. Also we've seen teams like this year's white sox jump out of the box and struggle to play .500 the rest of the year -- and not been surprised by that.

Which is a long way to say, I'm not sure that the hard analysis of data gets you much. I look at the roster. The Yankees' roster has several problems in my opinion which I've alluded to elsewhere. I think they'll be competetive, and while they could win that division, I don't think they will. In any case, they've been on a good run, and hats off to them.

[/ QUOTE ]

I picked 48-34 because it was closest to the average record of each team after they won 10 games in a row (actually, 49-35 would have been slightly more accurate). The whole point of this analysis was to test the theory that teams that are currently on hot streaks tend to outperform other teams with equivalent records the rest of the season.

But yeah, I agree with everything else you said. Certainly there are alot of factors involved when projecting a team's W-L record at the end of the season. I just don't believe that a current streak is much of a factor, if any, and the data supports that.