Why is Indiana able to win without Katie Douglas in the lineup?

The Indiana Fever have now won two consecutive games on the road against the regular season champions from each conference with star guard Katie Douglas out due to injury.

It seems like a remarkable feat that has caught most people off-guard. But is there an explanation for it? The following are five possible explanations for their ability to win without Douglas, from simplest to more complex.

1. It's a fluke/small sample size

We have to entertain the idea that there's one extremely simple explanation for this: random chance. Play out this sequence of games 1000 times and the Fever are going to win both games at some rate above zero.

Luck is often an underrated aspect of basketball outcomes and there's a strong possibility that the Fever have ironically had a short run of it despite losing Douglas to injury: they had really good games and their opponents had objectively terrible games - or quarters - in one significant way or another. Losing a star changes the way opponents have to scout you - perhaps there are different points of emphasis in the Fever's offense. Maybe the Lynx really were thrown off by their time off between winning the Western Conference Finals and Game One of the WNBA Finals.

But of course, successful people make their own luck: the Fever wouldn't be to this point had they not figured out how to rebound. Erlana Larkins had a record-setting rebounding effort in Game One. Guard Erin Phillips has been playing close to the best basketball of her career offensively, entering the Finals as the Fever's most efficient scorer and taking some of the shots Douglas would normally get. Tamika Catchings has been playing basketball worthy of finishing higher than third place in MVP voting.

Saying it's a fluke or just dumb luck might seem to minimize the work the players put in to winning games and actually I don't think this is simply a matter of luck.

2. The Ewing Theory: Losing a star player helps a team

In short, Bill Simmons' Ewing Theory posits that a team that never wins anything substantial with a star player who gets an inordinate amount of attention will suddenly get better in that stars absence.

thewiz06 brought up this idea during the game on Sunday night and on the surface there seems to be some validity to it: star player goes down, everyone wrote the Fever off, the team actually exceeds fan/media expectations. To the points above about people making their own luck, multiple other players stepped up in Douglas' absence to "replace" Douglas' scoring.

But reading through Simmons' original explanation, you probably see the problem with applying the Ewing Theory to the 2012 Fever: losing Douglas and being left with a surprisingly dominant rebounder and former MVP is not like losing Dan Marino and replacing him with Jay Fiedler (Jay Fiedler!). Certainly they lost a star, but I don't think anyone will dispute that the heart and soul of the team is Catchings, who has been a top candidate for the MVP award for the past three seasons.

Nevertheless, the Ewing Theory does lay the groundwork for another plausible explanation.

3. Braess’s Paradox: eliminating a (primary) scoring option can improve efficiency

Brian Skinner of the now-defunct Gravity and Levity blog once laid out a plausibility argument for the Ewing Theory by framing a basketball possession as a network problem. It's a thought-provoking piece that doesn't prove the Ewing Theory but does lay out some principles that are useful.

Without diving deeply into explaining that piece, the core principle - and the way in which he demonstrates it - is probably more significant than we give credit for: removing a star (offensive player) fundamentally changes the flow or traffic of an offense. His follow-up piece provides some illustrations relevant to this discussion of Douglas' absence and actually helps us to arrive at a conclusion that is more easily put in terms of basketball statistics that we're more familiar with: losing a high usage star player requires an offense to change its flow by redistributing shots to different players. With that, the problem of losing a star player becomes a matter of balancing usage rates (the percentage of possessions a player uses while on the floor) and efficiency (the rate at which a player actually makes successful scoring attempts).

There are some clear flaws that emerge when imposing that theory on reality - especially when you switch to discussing a star wing vs. a star post player - but this is still probably the direction we want to look at when considering what's going on with Douglas.

First, Douglas has been the Fever's second-highest usage player this season (with a usage rate of about 25% in both the regular season and playoffs) and had a true shooting percentage of 54.28%, which is solid though not spectacular. So independent of the tempo at which the Fever plays, we might assume that those shots will either be redistributed or that the Fever would turn the ball over more often because Douglas' remaining teammates struggled to even get a shot off. Looking at the Fever's numbers over the last two games, the play of Erlana Larkins and Erin Phillips really stands out.

Name	Season USG%	Season TS%	Game 3 Usg%	Game 3 TS%	Game 1 Usg%	Game 1 TS%
Briann January	18.80	56.48	25.38	48.70	21.63	39.28
Erlana Larkins	13.80	56.92	12.02	40	17.48	66.66
Erin Phillips	15.84	51.52	17.27	71.83	18	54.16
Shavonte Zellous	19.24	47.45	13.28	68.75	16	35.42

For reference, Ken Pomeroy considers usage rates below 12% as "nearly invisible" and you'll note that Larkins has been hovering around that range this season. The other significant cut offs in Pomeroy's framework are "limited roles" (12-16% of possessions used) to "role player" (16-20% of possessions used) to "significant contributors" (20-24% of possessions used. Acknowledging that basketball games cannot be broken down entirely by numbers, there is in fact a redistribution of shots for the Fever that is significant if we buy Pomeroy's framework. (Just to add - and to reinforce an earlier point - Catchings' team-high usage rate was relatively similar in each of those two wins without Douglas and center Jessica Davenport did have a much higher usage rate in Game One of the Finals but only played 8 minutes.)

In plain terms, January and Phillips have increased their roles in the offense by significant margins while beating the Lynx in Game One included Larkins increasing her role quite dramatically as well. But the key here might be what happened on the wing: normally we'd expect players to get less efficient with more shots. Both Larkins and Phillips got more efficient with more shots in Game One. And Phillips was much more efficient in Game 3 against the Sun.

In other words, the Fever haven't just redistributed shots but they've distributed shots around to a wider range of players without Douglas consuming shots - they have, in effect, added points of attack to their offense which you could argue makes them harder to guard, even if not all of them have been efficient. That leads to a logical extension of Skinner's theory that we could explain with a separate point.

4. Resolving the network problem with ball movement

If we accept that more players are getting involved in the Fever's offense with Douglas out, then we might assume that they're moving the ball more to find shots rather than looking to one or two players to generate offense as they would normally with Catchings and Douglas on the floor. A brief look at their synergy numbers just complicates that a bit.

Stat	Season	Playoffs	Game 3	Game 1
Ast/FGM %	61	45	50	48.3
eFG%	47.84	50.66	62.90	46.53
Adj. Synergy	38	29	58	50

Synergy is a metric that I used to approximate not how much a team is moving the ball but how efficiently they're doing so - in other words, is a team effectively moving the ball to spots where they're actually making shots. Two things emerge from the chart above: first, the Fever actually relied more heavily on assists in the regular season. Second, playing without Douglas and assisting on less field goal attempts, the Fever were on fire in Game Three and shot about the same as they did all season with Douglas. Yet they have a higher synergy rating in both games.

Acknowledging the small sample size and that synergy is simply a descriptive statistic with no proven explanatory power, this would seem to present a conundrum: is this all really just a fluke of sample size after all? There's still one more thing that might explain this.

5. Katie Douglas' negative regular season plus/minus

@kpeltonwbb Couldn't help recalling her significantly negative +/- while this game was playing out...

— Richard Cohen (@RichardCohen1) October 15, 2012

Here's a statistic that I alluded to only indirectly in my WNBA Finals preview that's worth looking at more directly now: in the two regular season games with the Lynx, Fever stars Catchings and Douglas combined to shoot 25-for-67 (37.13%), which is not disastrous but a low percentage for the two players who shot by far the most of anyone during the season. Just for kicks, if we throw in Catchings' 6-for-20 performance in Game One of the WNBA Finals, in three games against the Lynx this season the Fever's stars have shot 31-for-87 (35.63%) from the field.

A lot of that might be explained by Minnesota's defense, but part of that is that neither Catchings nor Douglas shot much better this season than they did in the second game (in fact, Douglas' 6-for-14 shooting in the second game against the Lynx is actually marginally better than the percentage she shot for the season). And this leads to an interesting point about winning without Douglas, aside from the point that guard Erin Phillips has played well or maybe an extension of that point: Katie Douglas had a negative plus/minus during the regular season. In fact, here -9.5 plus/minus was the worst on the Fever.

It's almost counter-intuitive to think that Douglas could have a negative impact on the game - it's one of those observations that one might first question as a quirk in the metric, which are plentiful, rather than a true reflection of the player's value. But consider what we know about Indiana: Indiana was not a good rebounding team during the season. Douglas, like Catchings, put up shots at a high rate and (especially against Minnesota) didn't make a whole lot. That hurts a typically poor offensive rebounding team against a very good defensive rebounding team (more missed shots just lead to giving up the ball to the opponent more). In turn, Catchings and Douglas contributed to a number of potential empty possessions that the Fever weren't adept at extending with offensive rebounds during the season.

The big difference between Catchings and Douglas is that, as we all know, Catchings fills the stat sheet. In contrast, in the event that Douglas has a poor shooting night and goes 0-for-6 from 3-point range as she did the first time against the Lynx, her efficiency on her shots only drops - she doesn't get to the free throw line that often and doesn't pick up a lot of assists. So in essence, Douglas produced a lot of empty possessions this season (the best way to sort this out would be to look at Douglas' adjusted four factors plus/minus numbers, which would help tell us specifically what aspects of the game Douglas' presence/absence affects. Alas, nobody has done that for the WNBA.).

But the other part of this is that there's a defensive and offensive plus/minus (which has been explained at length here). Comparing Douglas' numbers to last year we see this:

Season	Offensive	Defensive	Net
2011	+8.2	-1.7	+6.5
2012	+0.9	-10.5	-9.5

That's quite a significant drop on both ends of the floor. Let's start with the offensive side though: her 41.3% field goal percentage this season is below her career average; in 2011, she shot career-highs of 46.5% from the field and 44% from the 3-point line, her most efficient scoring year in Indiana (earlier in her career she got to the line more often which bumped up her scoring efficiency). So the explanation above could apply offensively in addition to the fact that she got to the line less often than normal and fouled a bit more often herself. But there's still a lingering hang up: Douglas had nearly identical numbers this season as she did in 2009 across the board and ended up with a plus/minus of +13.6 (and + 8 defensively).

So what gives here? Defensively, she moved to guarding small forwards this season as a starter after playing the two most of her career. Some combination of that with sub-par shooting has hurt Douglas' plus/minus. It really could be that simple.

And there's more to that plus/minus story: Phillips led the team in plus/minus this season (+9.3) and Larkins was positive (+1.6). Both have seen more minutes and an increased role in the offense (for at least one game) with Douglas being out. That means at least two players who had a positive effect on the season all season, including Larkins who is playing better than she did this season, are playing more and more involved when on the floor.

There are interaction effects that have to be taken into account with plus/minus that we cannot really look at here, but there's probably some measure of each of the last three options working together to help the Fever get through their time without Douglas. Does that means they shouldn't want Douglas back or should trade her? Probably not - I'm sure they'll welcome her presence. But it is still interesting to think about what might be going on without her.