d4 Na6 e4 only 29% Success Rate

I was investigating various rebuttals to 1. d4 when I saw the analysis engine say 1. d4, 2. Na6, 3. e4 only has a 29% chance of success for white, which seemed low. I continued to play 3. e4, the numbers drastically change to about 44% chance of success for white. Why are these numbers so off when they should be identical? Thanks.

Wait....what? You want to know why an algorithm thinks an opening position after 3 moves is (somewhat) better for black than white?

With respect, why would you care? Your question would only be of any relevancy if chess games where 3 moves long. 

The other thing is using algorithms to study openings is notoriously unreliable. If you don't thinks so plug in 1.e4 e5 2.f4 exf4 to any engine worth its value will tell you black is better, yet practical examples by strong players like Short show otherwise.

Have a nice day!

Transposition

I don't see why White wouldn't win every game, since Black apparently doesn't get to move! 

It is likely showing the percentage of master games won with this opening. Out of the 24 games played with this opening, 29% of the games White Wins, 38% were drawn, and 33% were won by Black. Since their are not many games played with this opening, we cannot use the database to measure this system.
Another example of this phenomenon is 1. Na3, which White wins 70%, Black wins 10%, and draws are 20%. There are only 10 games to base this estimate off of, so it still remains unmeasurable using the database.

The reason the percentage changes after you play 2. e4 is because the database only shows games that are based off of positions. In other words, it is showing games that started with 1. e4 Na6 2. d4, alonside the games that started with 1. d4 Na6 1. e4. Most of the games that are being added into the equation show 2...c6 from this line that is being merged in.

All these lead to the same position (known as De Bruycker’s Defense)

1. e4 c6 2. d4 Na6 (Some weird Caro-Kan line.)
1. e4 Na6 2. d4 c6 (Lemming Defense transposition.)
1. d4 Na6 2. e4 c6 (Australian Defense transposition.)
1. d4 c6 2. e4 Na6 (Caro Kan Defensive System transposition.)

The 24 games it had shown you earlier were just the Australian-Lemming transpositions, and were few and unreliable. But as the game goes on, additional lines (such as the more popular Caro-Kan transpositions) start connecting at De Bruycker’s Defense. The number of games displayed suddenly increases above a hundred, and we begin to see that most games from this position actually end badly for Black (probably due to his dubious startup). The few De Bruycker Defense games that were reached by Australian-Lemming transpositions were likely won via chance or rating disparities, an element that starts dwindling as more games are added to the database.

Hope this helps! =D

-Lukay

Ahh I see now, I didn't know it was based off positions (I assumed it was linear and followed the game move-by-move). Makes much more sense now. Thanks Lukay.

Only full newbies played this by white?

do you guys not know how statistics work?  win percentages tell you nothing of the sample size or the rating of both players.

with extremely rare openings with suspiciously high win percentages, you will notice almost universally a sample size of below 100 (often below 10) and the player playing the really exotic opening is significantly stronger and is just having fun.