A double dummy solver shows the best possible result when every player can see all 52 cards and nobody makes a mistake. It is a great learning tool. But it is not real bridge. At the table, you see only your own 13 cards, plus dummy after the opening lead.
This article explains the main limits of double dummy analysis. It also looks at a well-known study by Richard Pavlicek, who compared real expert results with double dummy results on almost 80,000 deals. If you are new to the topic, read What is Double Dummy Analysis? first.
1. The Defense Sees Your Hand: The Opening Lead Problem
This is the biggest single difference. A double dummy defender always finds the best opening lead, even a very strange one. For example: leading a low card away from an ace against a suit game, because it happens to work on this exact deal. No human would choose that lead without seeing all four hands.
Real defenders lead with much less information. They often give declarer an extra trick, or even the contract, with a completely normal lead. So double dummy says some contracts fail, when in real life most declarers would make them.
2. Declarer Never Guesses Wrong
The same problem works in the other direction. A double dummy declarer never loses a two-way guess for a queen, always drops a singleton king offside, and always picks the winning line. Real declarers have to guess.
Bridge writer Henry Bethe estimated that on 10 to 15 percent of all hands, a double dummy solver takes more tricks than a human declarer could fairly expect to take. The two effects (perfect defense, perfect declarer) push in opposite directions, but they do not cancel out exactly on any single deal.
3. The Solver Plays the Cards, Not the Odds
A double dummy solver picks the line that works on this exact deal. It does not care about probability. A line with a 5 percent chance "beats" a line with a 95 percent chance, if the 5 percent happens to work this time.
So on a single deal, the double dummy result cannot tell you what the correct play was. The correct play is the one with the best odds, and sometimes the best play fails. This is why BridgeDummy grades your choices over thousands of simulated deals, not over one deal.
4. No Deception, No Signals, No Human Mistakes
Double dummy bridge has no psychology. Falsecards do not exist, because nobody can be fooled. A swindle line, where you make an impossible contract by tempting a defender into an error, is worth nothing at double dummy. In real bridge it wins matches.
Real defense is also a partnership game. Defenders need signals and discards to tell each other what to do, and they often get it wrong in the middle of the hand. Double dummy defenders cooperate perfectly without exchanging any information at all.
5. No Safety Plays, No Scoring Strategy
A double dummy solver always plays for the maximum number of tricks on the given layout. But experts do not always play for the maximum. At IMPs, a good declarer may give up an overtrick to make the contract safe. At matchpoints, a good declarer may risk the contract for one extra trick.
See Matchpoints vs IMPs for why the scoring form changes the right play. Double dummy analysis knows nothing about any of this.
The Pavlicek Study: Real Play vs Double Dummy
How big are these effects in practice? Richard Pavlicek answered this in his study Actual Play vs. Double-Dummy. He took real results from top expert events (the Vanderbilt, Spingold, U.S. Championship and World Championship) and compared them with double dummy results on the same deals.
"The database from vugraph archives 1996-2014 consists of 80,138 results from 72 events, however 196 passouts are removed to leave 79,942 contracts."
The headline numbers:
| Metric | Real play | Double dummy |
|---|---|---|
| All contracts made | 64.36% | 62.03% |
| Slams made (6 and 7 level) | 66.34% | 68.30% |
| Average tricks won by declarer | 9.28 | 9.21 |
Real experts made more contracts than double dummy predicted. The main reason is the opening lead. In Pavlicek's words:
"The opening lead gives the defense a considerable advantage at double-dummy; e.g., the perfect lead will sometimes bury a contract before declarer has a chance."
He then repeated the comparison starting after the real opening lead was chosen. With the lead fixed, the picture flipped: in his words, "double-dummy play surges to the front in each category". Once declarer no longer profits from a friendly lead, perfect play wins.
Slams are the exception in the first table. Real experts made slightly fewer slams than double dummy predicted. At the six and seven level, declarer's perfect guessing helps double dummy more than the perfect lead helps the defense. So double dummy slightly overrates thin slams.
TipPavlicek's own conclusion: "The overriding impact of the tables is that actual play versus double-dummy is pretty close, so analyses based on double-dummy play should generally be on track."
What This Means for Your BridgeDummy Scores
The gap between real play and double dummy is small on average: about 2 percentage points on contracts made, and less than a tenth of a trick per deal. On a single deal the gap can be large, but BridgeDummy never grades you on a single deal. Every puzzle is scored across thousands of simulated deals, so the deal-by-deal noise mostly cancels out.
Keep the known biases in mind when you read results:
- Real contracts, especially notrump games, make a little more often than double dummy says, because real defenders miss the killing lead.
- Thin slams make a little less often than double dummy says.
- A "double dummy best" lead is not always the percentage lead. That is why lead puzzles are scored over many deals, not one.
Double dummy is not perfect. It is simply the best objective benchmark that exists, and its errors are small and well understood.