Saturday 29 June 2013

Europa Universalis III Simulation: The Sick Man of Europe and Friends (the Ottomans, Austria, and Bohemia)

Continuing my 50 game simulation of Europa Universalis III from last week, I've decided to process videos for some of the big players in Central Europe and the Balkans: the Ottomans, Austria, and Bohemia.

The Ottomans:

The Ottoman Empire has largely replaced the Byzantine hegemony in the Balkans and Near East by the 1399 start date, and looks poised to capture the last Greek holdings, namely Constantinople (later Istanbul).


While the Ottomans are riding on the momentum of many successful campaigns against the Byzantines, historically they are actually on the eve of Timurid invasion in 1402 which would lead to the capture of the Sultan and a decade long civil war.

Very rarely does Europa Universalis take that course of action however. In the first years of the game, the Ottomans normally consolidate their position in Asia Minor by incorporating the Turkish minors into their realm.

By the late 1400's, the Ottomans begin expanding northeast in most games, into the Caucasus and around the Black Sea, around the same time as the historical Ottoman Empire was preparing to go to war with Persia. Constantinople is rarely conquered by 1453, but rather closer to the turn of the century, if conquered at all. In fact, close to 50% of the time, another Turkish minor or Austria strikes first.

By 1600, the Ottomans have typically reached their peak, looking a little ridiculous, resembling the western half of the Mongol Empire. This I think highlights a flaw in the Europa Universalis AI; typically they conquer what is easiest, not what is most beneficial to them.

From 1600 onwards the Ottomans normally begin to collapse into themselves, having greatly overextended themselves. This collapse is most often precipitated by Austria, who themselves often collapse in a similar manner. Still, in the minority of games, the Ottomans manage to fulfill their weird dream of setting foot in the Pacific Ocean. Maybe they envisioned a Turkish Alaska?

Austria:

The famous House of Habsburg takes the helm of Austria in 1399, which historically was a house divided at the time, with much bickering and dispute between family members. As you'll find out, this isn't the case in Europa Universalis III; Austria quickly becomes a great power, if not a superpower.


Most games begin with the absorption of Hungary, which historically doesn't happen until after the Battle of Mohács in 1526, when it was inherited along with Bohemia by Ferdinand I. 

You will find that the Europa Universalis Habsburgs are much more bloodthirsty and impatient than the historical Habsburgs, as by the late 1400's they often begin to push into the Ottoman Empire. The is completely reversed from history, with the Ottomans very nearly conquering Austria during the Siege of Vienna in 1529.

By the mid-1500's, one can forget entirely about history as the Austrians begin to form a nice blob in the Balkans. Their holdings, depending on the game, will stretch anywhere from France, to Persia, to Siberia. It is not uncommon for the outer reaches of the Austrian Empire to fall into rebellion, then be reconquered, and cycle through these two states many times during the game.

Significant colonisation in most games begins by the mid-1600's along the Brazilian shore. By the early 1700's the Austrians begin moving into North America, normally focusing on New England and later the American South and Mexico. It is important to to note that all historical Austrian colonial efforts were insignificant and limited to just a few islands, all of which eventually failed.

By the game's end date, 1820, much of the globe is apparently speaking German. I would argue that, in Europa Universalis III, Austria most often becomes the world's strongest nation. That being said, they are typically also the most unwieldy and unpredictable nation, making for many interesting games.

Bohemia:

The King of Bohemia, Václav "the Idle" (who is historically deposed in 1400), begins the game as the Holy Roman Emperor. Things are looking up for Bohemia, a land of Czechs and Germans.


Bohemia's first moves are typically expansion into Poland, after which they continue slowly making their way east. Bohemia, compared to their eastern neighbours, is much more technologically advanced in 1399, making expansion an easy prospect.

Historically, Bohemia would become embroiled in many regional conflicts due to their position in the Holy Roman Empire, but their borders were relatively stagnant. Despite this, in many games, Bohemia continues their expansion into northern Hungary in 1500.

Historically, as mentioned previously, Ferdinand I of Austria would inherit Bohemia in 1526 after the Battle of Mohács, and Bohemia would remain Habsburg up until the conclusion of the First World War. In by far the majority of games Bohemia will defy this course of history, and would expand even further, often reaching Lithuania and the Golden Horde by the year 1550.

From the 17th century onwards, Bohemia's borders would remain relatively unchanged. They appear to make frequent eastern excursions, most of which eventually fail. Bohemia very rarely undertakes any colonial endeavours, and effectively stays put until the game's end, content with their little empire spanning the plains of Eastern Europe.

What's next?

With many of the major European powers already covered, I think I will focus on a particular region, or perhaps the East Asian or Indian nations next week. Please leave a comment and let me know what you would like to see!

Saturday 22 June 2013

Europa Universalis III Simulation: The Big Three (France, England, and Castille)

One of my favourite video games is Europa Universalis III, a strategy game spanning the Renaissance to the Napoleonic wars. Its creator, Paradox Interactive, intends to simulate world politics, economy, warfare, religion, and culture, among other things, from the year 1399 to 1820.
Europe and the Near East at the game's start date: 1399.
The course of world events is gently guided through events in the game such as the appearance of the Christian reformation or the discovery of foreign lands, but the game largely takes a hands-off approach and allows for countless alternate histories to develop. I've always wondered, if left to itself, how often would Europa Universalis III reproduce history? What trends would develop? What countries would colonise what?

To satisfy my curiosity, I decided to simulate 50 games of Europa Universalis III. The game does not have an observer mode per se, but by creating a nation in the middle of a "Wasteland", an impassable land (the Northwest Territories to be specific), I created a permanently isolated nation. This allowed me to sit back and watch history unfold.
Spain's probability map in the year 1661.
Processing this data, I then isolated and overlayed the results for each nation, giving a map depicting the probability that that nation will occupy a province at a given time. This yields one video per country. Darker regions on the map indicate that the nation of interest occupied that province in a large number of games, and in contrast, light regions mean that this province was occupied only in a small number of games.

I've decided to start with the "Big Three" in any game of Europa Universalis, and in history in general: France, England (later Great Britain), and Castille (later Spain).

France:

You will notice a legend to the left of the map indicating what colour corresponds to what probability, and a year in the bottom left corner of the video. This shows how the map changes over time.



France starts the game as a patchwork of vassals in a peaceful period of the Hundred Years War with England. In most games the English are forced out early on, and France is united by around 1500.

After this, it is not uncommon for France to focus their attention on Aragon, pushing France down an alternate history and putting them at odds with Castille.

Large scale colonisation doesn't get going until around 1600, agreeing well with history as France historically founded Port Royal, their first colony, in 1605. French colonies tend to dot the east coast of the Americas and West Africa.

France is interesting in that around 25% of the time it is reduced to nothing but Paris by the end of the game. Observing the course of events in each game, I have found that this is mostly the fault of Burgundy or Austria. France will often dominate all of Europe, and then a coalition of nations led by one of these two will bring France to a halt. It is a bit reminiscent of Napoleon's bid for Europe!

England/Great Britain:

England has the opportunity to form Great Britain if it fulfills certain requirements (own London, Edinburgh, be of British culture, etc...). So, both England and Great Britain (if it was formed by England) are tracked in the following video. If Great Britain was formed by a country other than England however (such as Scotland), it is not tracked. The same concept applies to Castille, which may form Spain.


England begins the game with the coronation of Henry IV. It is on the brink of war with France over its provinces across the channel, and will frequently lose this war very early on in the game.

Ireland is conquered in most games by the mid-1400's, about a century before the Tudors historically completed their conquest. Likewise, Scotland is typically brought under English rule by the late 1400's, over two centuries before the Treaty of Union. This allows Great Britain to form very early in most games.

England normally begins colonisation in the late 1500's which is right on par with Roanoke colony, historically founded in 1585. English colonisation is focused along the east coast of what would become the United States and Canada, with a few areas of intense colonisation dotting Africa and South America. Australia and New Zealand are often settled by the late 1700's, which paints a very historical British Empire on the map with one major exception: India, which is largely untouched.

Castille/Spain:

Because Spain is yellow in game, I decided to change the colour scheme a bit to produce better contrast. This is the reason for blue water and darker grey land.


Castille moves quickly in most games to establish a presence in North Africa; no one nation along the southern Mediterranean is entirely safe from their conquest.

They begin early colonisation, most prominently during the early 16th century, as they did historically (Christopher Columbus "discovering" America in 1492). Castille has a tendency to colonise everything, but the focus of their colonies tends to be the southern United States, Mexico, Central America, and South America.

Castille, unlike England, often deviates from history however. It is not uncommon for the Levant and the Ottomans to be conquered by the 1550's, with Castille's borders often stretching into the Caucasus and surrounding the Black Sea.

Castille does not always form Spain either. This is likely because France often ends up taking Aragon for themselves, barring Castille from becoming Spain.

What's next?

I'll probably continue with more nations next week. Please leave a comment and let me know which ones you are interested in seeing!

Wednesday 12 June 2013

NHL Stanley Cup Final (Blackhawks vs. Bruins) - Goal Probability Map

With the Stanley Cup Final beginning tomorrow, I thought it might be topical to make a quick post about it.

Using data provided by the NHL website, I have collected shot data from the regular season up until the conference finals for both play-off finalists: the Boston Bruins and the Chicago Blackhawks. From this, I constructed a map of each teams percentage of scoring from a given spot in the offensive zone:

The differences are interesting. Boston is quite clearly a more defensive team than Chicago, with a lower probability of scoring all around, and many of their chances coming from behind the hash marks (likely from defencemen). This fits in with their style of play though; the Bruins frequently win games not from scoring a large quantity of goals, but rather from a solid defence.

Chicago on the other hand resides on the other side of the spectrum with very high probabilities of scoring near the crease. An interesting feature of the Chicago map is its asymmetry: when looking at shots taken behind the hash-marks, the Blackhawks have a higher probability of scoring from shots taken on the right side of the ice. This skew is caused by their first and third best scorers in the regular season, Patrick Kane and Marian Hossa, both right wingers. It's amazing that these two players can make such a huge difference when this sort of data is viewed at a high level.

All that said, we will see if this data holds true come tomorrow.

Saturday 8 June 2013

NHL - A History of Team Performance

The NHL has been running for 96 seasons since the inaugural 1917-18 season. That's a long time.

With such a long history, I thought it might be interesting to plot each team's regular season record against time, to see if any patterns emerged. Here is the result:

Probably the most obvious features is the dominance of the Montreal Canadiens for much of the league's history. Starting in the mid-1940's and not fading until the mid-1990's, Montreal was a force to be reckoned with, and this is reflected by their 24 Stanley Cups.

Periods of Bruins dominance can be seen in the 1930's and, more prominently, in the 1970's during the Bobby Orr years. Similar trends can be seen with other dynasties: Gretzky's Oilers, Howe's and Yzerman's Red Wings, etc.

Another interesting feature that can be pulled from this graph is the stability of the league over time. This can be inferred from large spikes in the graph. Most of these spikes are caused by teams either entering or leaving the league, disturbing the balance of power between the rest of the teams; the team entering or leaving the league's share of points are not evenly distributed between the remaining teams.

The NHL's early days were extremely turbulent; the First and Second World Wars and the Great Depression forcing many teams to fold. The Original Six era brought about a relative calm, with teams growing better or worse much more gradually. This peace would be upset by the NHL's expansion following the 1966-67 season, and wouldn't reach another equilibrium until the late 1980's, which was cut short by further expansion. The league hasn't really reached an equilibrium ever since.

The introduction of the salary cap following the 2004-05 lockout probably further destabilized the league (at least in the short term), but it is quite clear that teams are becoming much more equal in terms of their share of points. Barring further expansion, it is unlikely that a team will ever again dominate the league like the Montreal Canadiens once did.

There are probably more trends in this data, but I will leave it to the reader to find them.