As I mentioned in the Series Thread, I was looking at the standings at the close of business yesterday and something struck me upside the head. Because there sit the Texas Rangers, with the best W-L record in the entire American League. They've done this by scoring 554 runs and allowing 555 runs.

Which seems a strange formula for success to be sure.


But we saw something very similar just a few years ago, when the 2012 Baltimore Orioles won 24 more games than they lost (93-69) while scoring just 7 more runs than they allowed. Those Orioles went 29-9 in one-run games; this year's Rangers have gone 26-8 in one-run games.

Not so long ago, in the latest of my many, many, many efforts to untangle the phenomena of close games (obviously I'm still haunted and traumatized by the 2005 Blue Jays), I began stumbling towards a way of establishing a team's expected record in one-run games. The principle that began to arise from the murk suggested that in a team's one-run games, the results of roughly half would reflect the team's quality. The results of the other half would reflect the flip of a coin. After a bit of mucking about, I found a formula that works pretty well for large samples.

In 60% of the team's one-run games, use their expected Winning Percentage, using the Pythagorean formula, based on their runs scored and allowed in non-one run games.
In 40% of the team's one-run games, flip a coin. Which should come out at .500, and if we flip the coin often enough I'm pretty sure it will come out at .500....

So - if a team has a Pythagorean W-L expectation of .600 and they play 40 one-run games, we expect them to go something like 22-18 in their one-run games (14-10 plus 8-8). Which is .550 ball, and it's pretty much what you should expect from a .600 team in their one-run games.

This formula actually works for large groups of teams and seasons.

But a single season, with just a few dozen one-run games, is a tiny, tiny sample size and anything can happen. Like with the 2012 Orioles and the 2016 Rangers, teams which appear to have flipped the coin and seen it come up heads every time. In fact, they've come up heads so often, they warp the friggin' formula. The quality of the Texas team suggests that they should play .around 500 ball in 60% of their one-run games. So even if the coin came up heads every single time, it wouldn't be enough to get them to 26-8. The best you could expect, with impossibly great luck, would be about 24-10. This is an extremely weird team, folks. )

Anyway, what I was really wondering about this time is this: exactly how did Texas win those 26 games by a single run? Besides being ridiculously, and quite undeservedly lucky - because it's a fluke, folks.

What were the mechanics of this particular fluke? Were they good at holding close leads? Were they bad at holding big leads? Were they good at late inning come backs? How did they do it?

Let's have a look:

WINS
1 - April 4: Held 3-2 lead from 5th
2 - April 20: Held 2-1 lead from 6th
3 - April 27: Held 3-2 lead from 6th
4 - May 2: Scored late. Tied after 7.
5 - May 11: Held 6-5 from 6th
6 - May 14: Walk off at home in extras. (Blew 5-2 lead in 9th)
7 - May 15: Late rally, down 6-3 in 7th.
8 - May 20: Allowed late run (ahead 2-0 after 3)
9 - May 21: Held 2-1 lead from 3rd
10 - June 5: Held 3-2 lead from 5th
11 - June 6: Walk off at home. Tied after 7.
12 - June 7: Allowed late run (ahead 4-2 after 8)
13 - June 11: Won in extras.
14 - June 17: Held 1-0 lead from 5th
15 - June 18: Late rally, down 3-0 after 7, 3-2 after 8
16 - June 19: Late rally, down 4-3 after 6
17 - June 20: Held 4-3 lead from 4th.
18 - July 1: Scored in extras on the road. Tied after 7.
19 - July 8: Held 6-5 lead from 6th
20 - July 24: Held 2-1 lead from 7th
21 - July 25: Late rally for walkoff at home, down 6-5 after 7.
22 - July 28: Scored late. Tied after 5.
23 - July 30: Walkoff at home. Tied after 3.
24 - August 6: Allowed late run (ahead 3-1 after 8)
25 - August 8: Late rally on road, down 3-1 after 8.
26 - August 10: Blew 3-1 lead after 7, rallied from 4-3 in 8th.

LOSSES
1 - April 7: Walk off loss on the road. Tied after 8.
2 - April 23: Walk off loss on the road. Tied after 10.
3 - May 4: Walk off loss on the road. Tied after 6.
4 - June 1: Walk off loss on the road in extras.
5 - June 24: Blown lead. Ahead 7-4 after 678
6 - June 30: Walk off loss on the road. Tied after 5
7 - July 3: Couldn't catch up. Down 5-2 after 5, 5-4 after 7.
8 - August 3: Down 3-2 after 2.

So the next thing I want to do - and I welcome input here, faithful readers - is to define the types of one-run games. This Texas log provides some possibilities. Maybe something like this will cover the various possibilities

1. Walkoff wins  - Texas has 4 of these)
2. Late run(s) scored (7th inning and later) - Texas has 9 of these
3. Late run(s) allowed (creating a one-run game) - Texas has 3 of these
4. One run lead held through final 3+ innings - Texas has 10 of these

And something matching for one-run losses

1. Walkoff losses - Texas has 5 of these
2. Late runs allowed (7th inning and later) - Texas has 1 of these
3. Late runs scored, creating the one-run game - Texas has 1 of these
4. One run deficit through final 3+ innings - Texas has 1 of these

And having devised a formula, and having created categories of one-run results... I think I'm ready to get to grips with the 2016 Blue Jays. And the 2015 Jays. And, heaven help us, the 2005 Jays. And other anomalies that hover into view...