Most engine panels show one move. MultiPV shows two, three or more candidate moves with separate evaluations. That sounds like a small interface option. In practice, it changes the kind of question you can ask.
What MultiPV does
MultiPV stands for multiple principal variations. Instead of searching only the line the engine currently considers best, it keeps several root moves alive and reports a line for each. The evaluation numbers are still engine opinions, but now you can see the gap between the top choices.
When MultiPV is useful
- Choosing a practical repertoire: if the top two moves are within a few centipawns, the second move may be easier to understand and remember.
- Comparing plans: one line may win a pawn, another may improve the worst piece. The evaluation alone will not explain the difference.
- Testing your intuition: write down your candidate move before opening MultiPV. If your move appears at position three, the engine is telling you something about your evaluation.
- Avoiding tunnel vision: in quiet positions, the top move can hide a completely different strategic idea.
When MultiPV is misleading
More lines do not mean more truth. A deep search with one line is often more reliable than a shallow search with five. MultiPV divides the engine's effort, so each line gets less calculation. On a node budget, MultiPV can make every line weaker.
Also, a small evaluation gap does not mean the moves are equally practical. A forced line that requires twelve precise moves may be objectively equal and practically losing for a human.
A practical routine
- Analyse the position with one line first. Form your own plan.
- Turn on three lines and compare the ideas, not only the numbers.
- Click through the first five moves of each line and say the plan out loud.
- Choose the move you understand, then test it in a game.
Use MultiPV on the ChessUrge analysis board and compare it with a single-line search at the same depth. The difference in clarity is usually more instructive than the evaluation itself.
Frequently asked
How many MultiPV lines should I use?
Two or three is enough for most study. More lines reduce the depth of each search and make the output harder to compare.
Is the top MultiPV move always best?
It is the engine's current preference at that depth and node budget, not a guarantee. Practical difficulty and your own understanding also matter.
See the whole game, not just the blunder.
Choose a review depth, see the maximum cost, and get move-by-move accuracy with blunder, mistake and inaccuracy markers.