MEZARPOLY

Which play style wins a property-trading game?

26 October 2026 · 6 min read

Everyone has a style in property-trading games: some buy every city, some keep their money aside, some never trade. Which one actually wins? Using Mezarpoly's own rules engine we ran 21 experiments; in each, we changed a single habit of one player and played 6,000 games. Some results are what you would expect, some are not.

1. How the experiment worked
2. 1. Offer trades: the biggest difference
3. 2. Don't put off building once you have a set
4. 3. Don't sit on too much cash
5. What made no difference: how picky you buy, auctions, Jail
6. Bot personalities: who are you playing against?
7. How different are the difficulty levels?
8. How many turns does a game take?
9. Summary
10. Method and limits

How the experiment worked

In each experiment, one player at a four-player table changed a single habit; the other three were Mezarpoly's normal-difficulty "balanced" bots. The seat of the changed player rotated from game to game, which evens out the effect of rolling first. Each experiment ran for 6,000 games on the Crimson Classic map. If nothing changed, every player would win 25% of games; our standard player won 24.4%.

95% interval: if the experiment were repeated, the result would land this close 95% of the time. Differences of less than 1.5 points between two rates can be explained by luck.

1. Offer trades: the biggest difference

A player who never offers a trade wins only 5.2% of games. That is the worst result in the experiment (apart from the easy bot). A player who offers trades whenever possible goes up to 29.2%.

The trade offer in the experiment is simple: ask the owner of the missing city of your set to sell it for cash, starting at 1.6 times its price, and raise the offer after each refusal. The player who never offers still considers the offers that come in, but rarely completes a set of their own. In a four-player game the cities get spread out, and most sets are completed by trading, not by the dice. A player who doesn't bargain is left unable to build while their opponents sell cities to each other.

Paying several times the price for the city that completes your set looks expensive. But a set brings double rent and the right to build, and a built-up group earns far more than its cities are worth one by one.

2. Don't put off building once you have a set

A player who often puts off building when they could build drops to 18.5%, and one who puts it off a lot drops to 6.8%. Set rent may be double, but rent really grows with the second and third house. We worked out how many turns each building takes to pay for itself in Houses or hotels? Rent also grows with inflation after turn 21, and a building put up early benefits from that for longer.

3. Don't sit on too much cash

Spare cash is the money you want to keep aside before you buy anything. The standard player keeps about ₺120 at the start of the game, and more as their buildings grow. A player who keeps about twice that drops to 22.6%, and one who keeps three times that drops to 19.1%. Keeping less cash costs nothing (25.2% and 24.8%).

Cash sitting aside collects no rent. On Mezarpoly, when you run short you can mortgage cities or sell buildings, so a large reserve is unnecessary. Put your money into cities and buildings.

What made no difference: how picky you buy, auctions, Jail

• How picky you buy. Buying every city you can afford (24.4%) and mostly buying only set pieces (24.2%) gave the same result. Both players always buy a city that completes a set or breaks an opponent's set, and that is where the real difference lies.
• Auctions. Bidding low (23.7%) or going above the price (25.2%) did not differ from the standard player by more than luck can explain.
• Jail. There is no meaningful difference between never paying bail (23.6%), paying early and staying in late (24.6%), and paying bail straight away in the first 25 turns (the standard player, 24.4%). The Jail decision matters less than people think.

Bot personalities: who are you playing against?

Mezarpoly bots play with one of four personalities: balanced, careful, aggressive and trader. The trader offers lots of trades and wins 28.8% of games. The aggressive personality keeps little cash and bids above the price in auctions: 22.3%. The careful personality keeps plenty of cash, offers few trades and builds little: only 13.7%. The results point the same way as the lessons above: trading wins, idle cash and delayed building lose.

How different are the difficulty levels?

Against three normal bots, an easy bot wins 1.7% of games and a hard bot 34.2%. The easy bot sometimes misses chances, builds little and never pays to leave Jail. The hard bot offers trades more often, asks a higher price for trades that would complete an opponent's set, and late in the game, when opponents' buildings pile up, chooses to stay in Jail. Start your first games against easy bots; once you beat hard bots, it's time to invite your friends.

How many turns does a game take?

Turn: the time it takes for everyone to play once. 12 standard maps, including Crimson Classic, Turkey, Istanbul, Europe and the themed maps, use the same 40-square price layout; World Tour and Big Turkey have 48 squares. On a 40-square map the median game ends after 45–48 turns whatever the number of players. At a crowded table everyone plays less often, but more rent changes hands each turn; the two effects balance out. Games on the 48-square maps are about 20% longer. If you want a shorter game, use the time limit in the room settings: when time runs out, the richest player wins.

Summary

• Offer trades. A player who never offers wins only 5.2% of games.
• Build once you have a set. Putting off building drops the rate as low as 6.8%.
• Don't sit on too much cash. You can always mortgage if you need to.
• Don't agonise over Jail and auctions. They make no measurable difference.

Method and limits

• The games were played with the rules engine that runs on the Mezarpoly server, with default rules: ₺1,500 starting cash, auctions, even building, the Vacation pool, and rent collected while in Jail. 126,000 games for the strategy experiments and 24,000 for the length measurements.
• The opponents were bots. People judge trade offers differently from bots, so the rates may differ at real tables. But the direction (trading wins, idle cash loses) should hold in real games too.
• A few games that had not finished by turn 600 were left out of the length measurement. In real games even the longest game ends within 2 hours: when time runs out, the richest player wins.

More from the blog