How chess engines can mislead you (and how you should stop falling for it)

I remember playing a tournament game a couple of years ago in which I felt I had been dominating the entire time. My pieces were active, my rooks and bishops robust, and my Stonewall plan against the Slav was fully in motion.

Feeling proud of myself after winning the game, the first thing I did was analyze my game with Stockfish (more so for an ego boost than for actual analysis). I was expecting to see a sizeable advantage throughout the middlegame. Even my higher-rated opponent had told me so.

I clicked on the engine.
I was apparently only +0.2 in a position I thought was almost completely won.

How engines evaluate

To evaluate most positions, many of us have been taught to assess king safety, initiatives and attacks, sequences, piece activity, and pawn structure, among various other concept-based and non-quantifiable factors.

Engines take evaluation completely differently.
They don't grade positions based on these factors; they grade them based on raw calculation.
They ignore all the pressure of an attack or a complete strategic stranglehold if the position is tactically holdable for the receiving player and slap on a +0.00 if it is.

This +0.00 indicates a draw to many beginners who don't understand strategy and pressure, thereby completely misleading them into thinking the position is dead. In fact, many 0.00 positions can be hard to hold for the defensing side or just incredibly complex. Here's a "0.00" like this from one of my OTB tournament games.

The most important thing to keep in mind when looking at these kinds of positions is that two positions with a similar engine evaluation are usually not the same. For example, an opposite-colored-bishop endgame may have the same "0.00" evaluation as an active position full of play, such as the example above. Because of this, you should definitely NOT exclusively rely on the eval bar to accurately evaluate a position.

How you should (generally) go about evaluating

Now, you may be asking yourself, "If I can't see that number, how on Earth am I supposed to know if I'm better or not?" The answer lies mostly in those aforementioned non-quantifiable factors. Here are a few basic factors you should typically use when evaluating your position, whether you're in a game or looking over it.

1. Sequences — does either side have a forced sequence of moves that gains material or leads to a winning position? / if not, analyze the other factors
2. Material — who has more points of material? / any notable material imbalances?
3. King safety — how safe is each of our kings?
4. Pawn structure — who has an overall weaker pawn structure? / any doubled, isolated, backward, etc. pawns? / who has more pawn islands?
5. Pieces — whose pieces are more active and well-positioned? / what are each player's poor pieces?
6. Space — who has more room in general for their pieces, and to what degree?
7. Weak squares — does either player have weak squares or holes for outposts, and can they be exploited?
8. Initiative/pressure — is either side leading a significant attack that could overrule many of the other characteristics? / is one side positionally dominant over the other?

Here's a sample position for you to analyze and evaluate based on these metrics. Answers for each metric in the order above and the final evaluation are under the insert, so don't peek until you're done (unless you aren't doing this or you're really struggling lol).

1. Nothing here is rapidly winning, so we look at the other factors.
2. Both sides have equal material, with no major differences or imbalances.
3. Both kings are safe.
4. Black has a backward pawn on d6, which can definitely become a target for White's pieces in the future. White's e4 pawn is also somewhat weak.
5. White's pieces are generally better than Black's pieces, with more squares covered and space to operate on. Black's knight on c7 is somewhat awkward, although it could be useful for future ...Nb5 moves.
6. White has much more space for their pieces, which invites thematic moves like Nc4 and f4-g4 advances in the near future.
7. Black has a weak b6 square, but White can't effectively use this weakness, as a knight on b6 doesn't do much. White has a weak d4 square that might pose issues in the future, considering Black's dark-squared bishop.
8. Although there is no current overwhelming initiative for either side, White can begin a kingside march with the aforementioned thematic f4-g4 in many scenarios.

Overall, White enjoys a better (but not winning) position, mainly due to their higher piece quality and more space on the board. In case you guys are wondering, White is roughly +0.8, but the whole point of this is kinda not using Stockfish's numbers to help us evaluate a position :P

As always, don't hesitate to ask for a more extensive blog post on evaluation if you want one.

Thanks for reading, and hope you found this post helpful!