Rail Baron Strategy
Jeff Goldsmith
I think that we've played RB to extinction and come up with some
statistics that imply the best strategies. There were some articles
in the Dragon way back when that display the numbers involved. I
can dig them up if anyone is too lazy to compute them themselves,
but here are the results:
I'm assuming a four player game, by the way.
The most common area of the country to roll a destination is the
Northeast. Thus, it is fairly critical that you have a rail into
there. There are exactly four roads that service this area. Far
and away the best is the PA. In fact, it is crucial to gang up on
the player who buys the PA or he will win almost all the time. If
he is allowed to get the NYNH&H, he will have a stranglehold on
the Northeast and will be hard to beat. The next best road there
is the B&O, not the NYC, especially since these should be your
first purchases and because the NYC is so expensive. Even if I can
afford the NYC, I buy the B&O if it is available. This statistically,
and mathematically turns out to make sense. The NYC is ok, though,
and is better than the fourth road in, the C&O.
Recapping: NE: PA, B&O, NYC, C&O
Note about the Northeast: the B&M and NYNH&H are awesome purchases,
far better than expresses. Buy them if it is safe and you don't
have anything more urgent to do. Boston is a big city (6th most
common?) and those are the only ways in.
The next most common destination is the Southwest, but three of
the four ways in are very expensive, so buying in there tends to
occur later in the game. Instead, the southeast tends to be
bought next, primarily because of the offensive value of the SAL.
Again, in the Southeast, there are four options. The L&N, the SOU,
the ACL, and the SAL. The L&N and the SAL are the best. The SAL is
the best buy. If you have the PA, buy the L&N; they link together
in a wonderful way. If you get one of the others, and have the B&O,
you may need to buy the RF&P to link them. No one else will want
this, so buy it when you are saving for a major, but don't delay too
long or someone will buy it to screw you. The combination of the
SAL and ACL looks dominant, but really isn't because the southeast
is rolled pretty infrequently. If you don't have to give up alot or
there are five or more players in the game, it might be well advised
to get this combination, but it is an all-or-nothing strategy, that
rarely works.
Recapping: SE: L&N, SAL, ACL, SOU
There are also four ways to get into the Northwest. They are all
relatively inexpensive, so they tend to go next. The CMSTP&P is
the best in that it reaches Chicago and is fast. The GN and NP
are about equal. Which of these you buy depends on which major
you are going to get. The GN works great with the "Sleazy major"
and the NP works best with the SP. The fourth way in is the UP,
which is a major and is not the best of the majors without the WP,
so generally, unless you are planning on the UP and already have
the WP, it is incumbant upon you to buy one of the others.
Recapping: NW: CMSTP&P, GN or NP, UP
About this point in the game, one or more players will be saving to
buy a major. The majors are the three $40,000 roads: SP, UP, and
AT&SF. The majors are also the primary ways west, especially into
the Southwest. Far and away the best major is the SP. It is fast,
covers alot of cities, and links very well with alot of other good
roads. This is not obvious initially, but it is true. For an
inexperienced player, the AT&SF covers the whole SW pretty well and
is easy to play (a "No-Brainer") so is a good choice, but if you
know what you are doing, it is best to get the SP unless all the
good connections to it are gone. The extra $2K is pretty crucial,
though, so get the AT&SF if the SP will break you. The fourth way
in (note that recurring number; Rail Baron is a four-player game
almost exclusively) we called the "Sleazy Major." If you get
the WP, D&RGW, and NP, you can cover alot of the west cheaply and
effectively. The T&P reduces the damage when you roll LA, Phoenix
or San Diego. The "Sleazy Major" is a last resort; if you are a
good player it is often worthwhile to buy the WP early to prevent
the weakest player from having a chance. The UP is also very weak
without the WP; SF/Oakland is the most common destination in the
game and the UP doesn't get close. In fact, the WP is great with
the SP for the locks, crucial to the sleazy major, crucial to the
UP, and ok for the AT&SF. The purchase of that road will often
swing the whole game.
Recapping: SW: SP, AT&SF, UP (+WP), (D&RGW, WP, NP)
The rest of the board (basically the middle) is most important
in that it links the rest of your roads well, rather than that
it reaches all the cities. The major cities (Chicago, St. Louis,
Kansas City) will be reached by any good network, so speed becomes
the primary concern here. Having loops and multiple ways to get
places is nice since you can perform complicated routes to avoid
paying the $10K twice. This technique is known as a "Death Spiral,"
and is best achieved via the CRIP. In the center of the board, there
are three (aha!) major connecting lines. The CRIP is the best and
allows you to buy either the NP or GN, since it gets you to Minneapolis.
The next best is the "inviso" or the light yellow rail whose name I
don't remember. The SLSF is a reasonable choice, but it won't get
you North. It works well with the UP, I think. Fortunately, there
are a number of smaller roads that will make these connections for
you in specific areas alleviating a need for one of the big three.
These include the MP (does the south central great), the C&NW (does
the northern Plains well), the GM&O (excellent North/South connector)
and the IC (crummy, but maybe crucial.) The N&W may help connect you
to your southeastern road too. It's not too uncommon to see people
go from Washington to Tampa via Chicago.
Recap: Plains/Sort of: CRIP, Inviso, SLSF, GM&O, IC, ...
Given appropriate tactics in each region, it is worthwhile to try
to generate a consistant, well-connected rail system from the very
beginning. Some roads connect better than others. Looking at the
board can show you that pretty clearly. My absolute favorite
minimal rail system is:
PA, NYNH&H, L&N, CRIP, SP, NP.
The GN substitutes fine for the NP, and the B&M is a terrific addition.
Note that with only six lines, you can cover almost the whole board
very well. The best part about this setup is that all these roads
(Except the NP) are pretty fast, so you can make alot of money without
other people paying you.
The only other real decision is whether or not to invest in trains.
And when. Mostly, expresses are useless. Any cheap rail is much
better and if you can just barely afford it, skip it and buy nothing.
When you are screwed and stuck in Miami, it might be worthwhile, but
you will probably lose the game anyway, so who cares. Superchiefs
are a different matter entirely. KNOWING that you have an extra
die can often save you $10K by allowing you to "Roll out" of a
city that your rails don't reach. This is a second reason why it
is worthwhile getting close to cities if you can't reach them. For
most games, it is clear to buy a superchief as soon as you can afford
it and don't want any more rails. This tends to happen shortly after
the last rail is purchased. The only times that you might not want
to buy a superchief are in five or more person games with no one knocked
out (very rare.) In these games you rate to have a substandard rail
system and don't really want to move very fast. This usually implies
that you have a stranglehold on one part of the country (hopefully
the northeast) and can't get close to another. In that case, you
might earn more by just putzing around on your lines and only paying
$1K per turn and hoping to win by having everyone else pay you. It's
not a great strategy because the other players try very hard to avoid
paying you if this is your strategy once you've announced (mandatory)
$150,000.
There is a bit of tactical consideration in the Rover part of the game
and actually winning the game, but if you know all these things and
your opponents don't, then no one will be close to you near the end
and you can just wait until you roll a destination inches from your
home city. You might have to do that anyway, but if you don't have
to worry about other players reaching $200,000, then it's easy.
Finally, a note about the worst roads. Nine out of ten games, the
last road bought is either the IC or the N&W. (We play 4-players,
so they are not that useful. In 5-or-more, I don't know.)
The MP is another candidate, but it's easy to get locked out the
South Central (not a big deal, by the way) so someone always buys
it. I can't think of another bad road; some get slighted if one
player is hopelessly lost and the other three have good systems
and don't care about what's left.
This is probably more than you wanted, so I'll summarize.
Buy the PA. You will probably win against inexperienced players.
--Jeff
are f
Scott Turner
>
>Northeast... Far
>and away the best is the PA... The next best road there
>is the B&O, not the NYC, especially since these should be your
>first purchases and because the NYC is so expensive.
That's very close to the analysis my computer RB comes up with.
Here are the actual ratings for these railroads when no other
railroads are owned (only the comparative values are interesting):
PA worth 3.017115
B&O worth 1.605306
NYC worth 1.242658
C&O worth .9946136
As you can see, the PA is rated *much* more highly. The surprise
here is that the B&O outweighs the NYC. As Jeff says, this is
largely because the NYC is so expensive. It's a better railroad than
the B&O, just too costly.
>Note about the Northeast: the B&M and NYNH&H are awesome purchases,
>far better than expresses. Buy them if it is safe and you don't
>have anything more urgent to do. Boston is a big city (6th most
>common?) and those are the only ways in.
Here are the ten most common destinations, and the probabilities to
travel there. This ignores the rules about re-rolling a destination
in your own region. Since the NE is the most common region, destinations
in that region should be a little less probable.
Oak-San .0447
New-York .0412
Los Angeles .0400
Chicago .0345
Boston .0294
Philly .0294
Seattle .0282
Kansas-City .0282
Portland .0267
Baltimore .0265
>The next most common destination is the Southwest
Odds for Regions
Northeast .210
Southwest .168
North Central .154
Southeast .126
South Central .126
Northwest .112
Plains .112
[Southeast]
> Recapping: SE: L&N, SAL, ACL, SOU
SOU worth 1.228332
L&N worth 1.016201
SAL worth .8333221
ACL worth .1137519
One of the surprises in the computer's ratings of the railroads is
the SOU. Most people dislike it because it doesn't seem to have
much coverage of the SE. But the coast cities it doesn't reach
(Miami, Tampa, Jacksonville, Charleston, Richmond, Norfolk and
Mobile) are much less probable than the cities it does cover (Atlanta,
Chattanooga, Knoxville, Charlotte and Richmond). What's more, it
also extends into the South Central (Memphis, Birmingham, New Orleans)
and to Washington. It's a badly under-rated railroad.
Another surprise is how much poorer the ACL is than the SAL. This
shows the value of the Miami (the monopoly) in this rating system.
[Northern Route]
> Recapping: NW: CMSTP&P, GN or NP, UP
CMSTP&P worth 1.123985
NP worth .8133436
GN worth .2957732
The Comstop is the best, agreeing with most player's intuition.
Interesting how comparatively poorly rated the GN is. I don't
consider the UP in the same class, but it rates somewhere between
the NP and GN.
>Far and away the best major is the SP.
This is my first major disagreement with Jeff. The SP is at best
about the same as the AT&SF.
> Recapping: SW: SP, AT&SF, UP (+WP), (D&RGW, WP, NP)
Here's the computer ratings:
SP worth 2.740767
AT&SF worth 1.715125
UP worth .6070125
The UP is lowly rated because it doesn't reach SF (the most popular
destination on the board). Notice that the computer agrees with
Jeff in rating the SP more highly than the AT&SF. I've punished
it severely for this :-).
The reason I rate the AT&SF more highly than the SP is that it
also gives coverage of the middle area of the board. To be the
first to buy a $40K railroad, you may have to run your cash reserves
low and you may not be able to purchase for a while. At this point
you are typically weakest in the Midwest (and often the Northern Route).
Rather than get (effectively) shut out in the Midwest, I'll buy the
AT&SF and get coverage of both the Far West and the Midwest. Also, I
can now pass on buying a Midwest railroad (which are generally expensive)
to get coverage somewhere else (i.e., the North). For these reasons
I rate the AT&SF somewhat better than the SP.
>The next best is the "inviso" or the light yellow rail whose name I
>don't remember.
The CB&Q.
>The SLSF is a reasonable choice, but it won't get
>you North.
One of the worst railroads on the board.
> Recap: Plains/Sort of: CRIP, Inviso, SLSF, GM&O, IC, ...
I won't post the computer ratings for railroads in this area, because
at this point you are most interested in linking up your networks, and
the computer rating changes to reflect this, and so will be different
for each set of railroads.
[Stuff about Superchiefs deleted...]
The Rick Gillespie strategy is to purchase a Superchief first if you
cannot get the PA. It sounds insane (like most things Rick does) but
it actually works out quite well.
>Finally, a note about the worst roads. Nine out of ten games, the
>last road bought is either the IC or the N&W.
The computer's bottom five:
N&W worth .1711633
GM&O worth .1504740
ACL worth .1137519
D&RGW worth .06191739
RF&P worth .04897735
This ignores tactical considerations. Depending on what you own,
railroads can become more valuable. (As soon as you own the SAL, the
ACL becomes a lot more interesting.) In our games, the last rrs purchased
are usually in the center of the board: IC, SL&SF, GM&O.
>This is probably more than you wanted, so I'll summarize.
>Buy the PA. You will probably win against inexperienced players.
Yep.
Scott R. Turner
UCLA Computer Science "Worn Down, Beaten Up, Spat Out"
Domain: s...@cs.ucla.edu
Jeff Goldsmith
>
> PA worth 3.017115
> B&O worth 1.605306
> NYC worth 1.242658
> C&O worth .9946136
>
>As you can see, the PA is rated *much* more highly. The surprise
>here is that the B&O outweighs the NYC. As Jeff says, this is
>largely because the NYC is so expensive. It's a better railroad than
>the B&O, just too costly.
Actually, the only reason that I'm concerned about the cost of the
NYC is that you need to buy it early. If that were not true, then
I would buy the B&O over the NYC even if they were the same price.
My rationale is that the B&O gets very close to the places (New York)
that it doesn't reach, which is not at all true for the NYC. Also,
the B&O allows for tactical sleazery due to its looping, which is
really worthwhile. Does your computer program factor in:
1) Closeness to unreached destinations
2) Extra (~useless) track that can save $$ on some rolls
3) Speed (more on this later)
4) Potential linkages
>
>One of the surprises in the computer's ratings of the railroads is
>the SOU. Most people dislike it because it doesn't seem to have
>much coverage of the SE. But the coast cities it doesn't reach
>(Miami, Tampa, Jacksonville, Charleston, Richmond, Norfolk and
>Mobile) are much less probable than the cities it does cover (Atlanta,
>Chattanooga, Knoxville, Charlotte and Richmond). What's more, it
>also extends into the South Central (Memphis, Birmingham, New Orleans)
>and to Washington. It's a badly under-rated railroad.
>
I don't think so. Firstly, it's really slow. Secondly, it doesn't
get anywhere near the death zone (Florida.) Thirdly, it doesn't
like up with anything very well. The point about the south central
is pretty much moot, because so many roads will get you close enough
to the one or two destinations that you might have there during the
game. The south central just never seems to be a problem, unlike
the four corners of the board. Besides, I'm an SP fan, and the SP
has the fast route right through the South Central.
>
>[Northern Route]
>> Recapping: NW: CMSTP&P, GN or NP, UP
>
>CMSTP&P worth 1.123985
>NP worth .8133436
>GN worth .2957732
>
I'm very surprised about the poor score of the GN. I think that
the other two are reasonable (at least relative to each other.)
The big bonus of the GN is the connection to the WP. If you have
both of them, you get a nice loop to play with.
>>Far and away the best major is the SP.
>
>This is my first major disagreement with Jeff. The SP is at best
>about the same as the AT&SF.
>
>> Recapping: SW: SP, AT&SF, UP (+WP), (D&RGW, WP, NP)
>
>Here's the computer ratings:
>
> SP worth 2.740767
> AT&SF worth 1.715125
> UP worth .6070125
>
>The UP is lowly rated because it doesn't reach SF (the most popular
>destination on the board). Notice that the computer agrees with
>Jeff in rating the SP more highly than the AT&SF. I've punished
>it severely for this :-).
It's almost impossible to play the UP without the WP for just
this reason.
The reasons that I like the SP are not obvious from the cost
or cities covered. I really like the link in Tucumcari with
the CRIP; the two of them are wonderful in the west, and with
a northern route are hard to beat. The SP is (I think, never
measured all of them) one of the fastest routes in the game.
You can get from the west coast to the east coast in quite a
lot shorter time that you can via the AT&SF. If you need to
go north, of course, the AT&SF is better, but with a central
road (the CRIP if possible) no one can beat your speed.
And, of course, the big bucks are in the coast to coast transits.
In a 4-player game, I almost never get shut out of the center of the
board. You can make some pretty nice networks with random crummy
roads and the SP. I've also never failed to get a good connection
to the SP, either. One trick is to buy the CRIP early and then you
are the primary customer for the SP.
>
>The computer's bottom five:
>
> N&W worth .1711633
> GM&O worth .1504740
> ACL worth .1137519
> D&RGW worth .06191739
> RF&P worth .04897735
>
Hmmm. For bad roads, I don't think that your value system
represents their value well at all. I like the GM&O just
fine; it's fast, connects to anything you want pretty well
and gets close to some tough to reach cities. Of course,
it is most valuable as a linkage route.
The ACL has the problem of getting to it; I don't like it
either.
The D&RGW is a pretty good 1st choice in some of our games.
It is almost always useful, mostly as a connector, but also
for that loop. I occassionally buy the WP first, at which
point the D&RGW is very valuable.
The RF&P is worthless unless you need it. Tautological, but
ever so true. That one goes last sometimes. On the other
hand, in our games, people would buy it to screw other players
who needed it. If no one did, it could be the last remaining
road.
The Rick Gillespie strategy isn't all that silly sounding. I've
won games with a similar "funny" strategy. Buy two majors. Lose
in the east. (Without the PA or B&O you will anyway.) If you
have both of the SP and AT&SF, someone else will die very soon.
Also, you might get $10K just from players established on the
things. It's tough to get the money to do this, but you can lock
two players out the the West, perhaps freeing up one of the nice
eastern rails later in the game. Besides, it is great fun going
for 25 turns on your roads nowhere near your destination so that
someone will have to sell you a road before you pay them.
Thanks for the comments; I'd like to see this computer Rail Baron
player. How does it deal with strategic considerations? Does it
ever beat a human opponent? At least a good one? Impressive program
if it does.
--Jeff
Rick Gillespie
>cannot get the PA. It sounds insane (like most things Rick does) but
>it actually works out quite well.
As much as I would like to take complete credit for this amazing
strategy, I can't. It is a small modification of the "Terry Crocker"
strategy which was to buy a Superchief right away. I simply added
the "if I can't get the PA" clause.
The rationale behind this "insane" strategy is to give up some early
purchases (by spending $40K so early) and pick up lots of purchases
later by making some many more runs than anyone else. I just figure
I will give up the early purchases 'cause the PA is the only thing
so much more valuable than anything else.
"Insane" Rick (don't backstab me in Diplomacy :-)
Ken Fishkin
>[ a whole bunch of solid-gold observations about Rail Baron]
Great post! My only quibble would be that in our experience, the
L&N is a "tastes great, less filling" railroad: looks like a great
buy, but doesn't really wind up doing any good.
>Again, in the Southeast, there are four options. The L&N, the SOU,
>the ACL, and the SAL. The L&N and the SAL are the best. The SAL is
>the best buy.
In an _n_ player game (N > 4), the SOU has another advantage:
it can serve as a "poor man's" railroad for the NE, since it will at
least get you up to Washington. Better than nothin'.
>Nine out of ten games, the
>last road bought is either the IC or the N&W.
yup. The GM&O also usually goes pretty late.
--
Ken Fishkin ...ucbvax!pixar!fishkin
Scott Turner
> 1) Closeness to unreached destinations
> 2) Extra (~useless) track that can save $$ on some rolls
> 3) Speed (more on this later)
> 4) Potential linkages
(1) No (2) No (3) No (4) Yes, in a sense.
(1) is an interesting point, though difficult to quantize in any
useful sense. I'll have to think about it a bit more, to see if
adding it to the rr evaluation routines would be worthwhile.
(2) and (3) I consider of marginal importance. Making loops to save
money is an infrequent (but not rare) situation, and I'm not convinced
that in general it makes sense. The marginal cost of waiting a turn
to make a delivery is difficult to calculate and (my guess) varies
wildly throughout the course of the game. I also discount speed
because (a) it is difficult to measure, (b) the Rail Baron board is
surprisingly equitable in the sense that most reasonable routes
between two destinations fall within about 10% of each other in length
(but not so for "unreasonable" routes), and (c) the program doesn't do
that good a job about using track anyway.
(4) The program does a good job of calculating potentional linkages,
although it cannot "plan ahead", ie., buying the DR&GW early intending
to use it as a linkage later. That's a weakness, but not as big as
you'd expect.
>I don't think so. Firstly, it's really slow. Secondly, it doesn't
>get anywhere near the death zone (Florida.) Thirdly, it doesn't
>like up with anything very well.
The point is that the real death zone in the SE is Atlanta, not Miami.
Your odds of going to Atlanta are much greater than your odds of going
to Miami. I can't argue with your other points, except to say that I
think that coverage so heavily outweighs speed that speed becomes a
very secondary consideration. (I could be wrong about this.)
[About the computer's bottom five...]
>The D&RGW is a pretty good 1st choice in some of our games.
>It is almost always useful, mostly as a connector, but also
>for that loop. I occassionally buy the WP first, at which
>point the D&RGW is very valuable.
The ratings I posted were for the situation where the computer
owns nothing else. In that situation, the D&RGW isn't a very
good railroad. When it starts to become useful as a connector,
its value rises.
>Thanks for the comments; I'd like to see this computer Rail Baron
>player. How does it deal with strategic considerations? Does it
>ever beat a human opponent? At least a good one? Impressive program
>if it does.
In a multi-player game, the Apollo version can sometimes beat a good
human opponent. Nearly never in a one-on-one or three player game. I'm
expecting somewhat better performance from the PC version.
Scott R. Turner
UCLA Computer Science "Where Science Gets the Business"
Domain: s...@cs.ucla.edu
Jeff Wu
1. Two player games
Some people mentioned this in passing. I never played one, but
it sound pretty dull. The only strategy that makes sense is to
buy the superchief and not buy any railroads (unless you can
catch your opponent going to Portland Me.). There are approximately
$550,000 worth of railroads, you can get to $200,000 before half
of the railroads are bought.
2. Buying the superchief as the first purchase (aka Rick Gillespie strategy)
This is the preferred way to go for a three player game. For a four
player game, its marginal (you'll end up with about the same amount
of railroads as everyone, except you won't get the first picks). I
would employ this strategy if I can buy it after the first destination
and that I reach that destination fairly quickly. I would not employ
this strategy for 5 or 6 player games.
3. Heartless strategy
One strategy which I've seldom employ is to NOT pay the weakest
player in order to force him/her to auction (or to go broke). You
should employ this strategy when he/she has a key railroad that
will likely be auctioned off. Of course rail baron is a social
game. If you get a reputation for employing this strategy, it
will backfire on you in future games.
4. NE <-> SW connection
To have high coverage in the NE and SW has an added advantage of
having some high paying destination to go when you get to choose
your destination. If you choose the NE, you'll have the highest
probability that the next time you "destinate" that you'll get
to choose again. In fact, there is almost a 40% chance of
going to the SW if you are in the NE and vice versa (assuming
that you'd choose NE or SW).
Jeff Wu