If you've ever run a Stockfish analysis, you've seen a number like "Depth 25" or "Depth 40" next to the evaluation. Most players nod along like they understand what that means. Most don't.

This guide explains depth in plain English — what it measures, why more depth usually means stronger play, where it stops mattering, and what depth you actually need for your own game analysis.

What "depth" actually means

Depth is the number of half-moves (called ply) the engine looks ahead from the current position.

1 ply = one side's move (e.g., just white's move)
2 ply = a full move (white + black's response)
Depth 20 = 10 full moves ahead for both sides

So "Depth 25" doesn't mean the engine considers 25 future moves. It means it searches 25 half-moves — roughly 12-13 full moves for each side.

Why deeper usually means stronger

Chess has a branching factor of about 35 — that's the average number of legal moves in a typical position. To look one move ahead, an engine needs to consider ~35 positions. Two moves ahead: ~1,225. Three moves: ~43,000.

Each additional ply multiplies the search space by ~35. Naive search becomes impossible very quickly:

Depth	Raw positions (no pruning)
5	52 million
10	2.7 quadrillion
20	7 × 10³⁰

Modern engines like Stockfish use alpha-beta pruning, transposition tables, and neural network evaluation (NNUE) to cut this down by orders of magnitude — but the point stands: every extra ply of depth is genuinely hard-won.

The reason depth matters is tactics. A two-move combination that wins material is invisible at depth 1. A five-move sacrifice is invisible at depth 4. The deeper you search, the more long-term tactics the engine can see coming.

Selective depth: the sneaky number

Look carefully at Stockfish output and you'll see two numbers:

info depth 25 seldepth 42 score cp +0.38 ...

depth 25 — the nominal main-line depth
seldepth 42 — the deepest line actually searched for tactical sequences

Modern engines search selectively — promising moves (captures, checks, threats) get extended deeper while passive moves get cut off earlier. So a depth-25 search might actually look 40+ ply deep down the most interesting lines. This is why top engines crush classical depth analyzers even at lower stated depths.

When does depth stop mattering?

At some point, more depth returns diminishing evaluations. Here's roughly where that happens:

Depth 15 — spots 90% of tactics a club player would see. Good enough for most casual analysis.
Depth 20 — catches most typical master-game tactics. This is where Chess.com's free analysis usually caps out.
Depth 25-30 — where Stockfish starts finding quiet positional moves that humans struggle to calculate. Useful for serious post-game study.
Depth 35+ — where the engine evaluates endgames with long forced sequences. Relevant for opening theory research and top-level correspondence chess.
Depth 50+ — mostly wasted on "practical" chess. Two evaluations at depth 50 vs depth 35 rarely disagree by more than 0.1 pawns on a typical middlegame position.

For almost every player, depth 25-30 is the sweet spot. More than that is academic unless you're analyzing endgames or specific tactical shots.

Why Chess.com's analysis is often "enough" — until it's not

Chess.com's free Game Review runs at modest depth to save server load. Paid tiers go deeper, but still cap well below what a local engine can reach.

Where this matters:

Endgames — rook endings, pawn races, and fortresses often need depth 35+ to evaluate correctly. Chess.com will call a winning position drawn and vice versa.
Deep tactics — a 6-7 move combination where the refutation is itself a 3-4 move sequence. Cloud analysis sees the surface; local depth sees the truth.
Quiet positions — when the best move isn't forcing, cloud engines sometimes miss subtle improvements that deeper local analysis finds.

This is exactly the use case ChessPilot's unlimited depth mode solves. Premium users can set depth as high as their CPU can sustain — or just let it run to a time budget — for real analysis of your most important games.

Time vs depth

Depth and time are two sides of the same coin. Stockfish reaches deeper as it gets more time:

Hardware	Rough depth in 10 seconds
Phone / old laptop	Depth 20-22
Modern laptop (4 cores)	Depth 25-28
Desktop (8+ cores)	Depth 28-32
Workstation (16+ cores)	Depth 30-35

Multi-core matters a lot. Stockfish parallelizes well — doubling cores gives roughly +2-3 depth at the same time budget. This is why running Stockfish on your own machine (instead of Chess.com's shared cloud) gives noticeably better analysis.

Multi-PV: the other dial worth knowing

Beyond depth, there's Multi-PV — the number of candidate lines the engine reports. Default is 1 (just the best move). Setting it to 3 or 5 shows alternative moves with their evaluations.

Why this matters:

At Multi-PV 1, the engine only fully analyzes what it thinks is best. Alternative moves get shallow analysis.
At Multi-PV 3, all three top moves get evaluated thoroughly — great for studying "what's the 2nd-best option?"

Multi-PV costs time. Getting depth-25 Multi-PV-3 analysis takes roughly 2-3x longer than Multi-PV 1 at the same depth. Tradeoff accordingly.

Practical depth recommendations

If you're studying your own games, here's what I'd suggest:

Quick review during a game — Don't. This is how people get banned.
Post-game review (casual) — Depth 20-22 is plenty. Skim for blunders.
Post-game review (serious study) — Depth 28-32 with Multi-PV 3. Spend 30 seconds per critical position.
Opening repertoire research — Depth 30+ with Multi-PV 5 on key positions. You want alternatives, not just the absolute best.
Endgame study — Depth 35+ or use tablebases instead for 7-piece positions.

What ChessPilot does with depth

ChessPilot runs Stockfish locally on your machine with these advantages over cloud analysis:

No depth cap (Premium) — set depth to 40 if you want, constrained only by your CPU
Your CPU, your time — no rate limits, no queue, no sharing with other users
Runs in the background — analyze finished games while you play new ones
Multi-PV support (Premium) — see the top 3 or 5 lines simultaneously

For most players, depth 25-30 with Multi-PV 3 is a genuinely better analysis experience than Chess.com's built-in review — and it's unlimited.

The one-sentence summary

Depth is how many half-moves Stockfish looks ahead; more is usually stronger, but returns diminish after depth 30 for typical chess, and time/hardware matter as much as the number itself.

Now when you see "Depth 27" in your analysis, you'll actually know what it means.

— ChessPilot