How to find cheap gas using game theory | Hacker News Reader

How to find cheap gas using game theory

mindyourdecisions.com

41 pointsstrategy24 comments

This is of course the well known secretary problem. Of course to equate this with the secretary problem, one must assume a random distribution of gas prices.

But I think you can do better in real life. You can find a gas station that is more likely than average to have lower prices.

Just look for the station that would seem to be in the worst competitive position. I.e., the one that is inconveniently placed, that is away from the highway, that has a small sign, or that has a less famous logo. It will probably sell you cheaper gas.

Such station should probably have higher prices to survive.

The out-of-the-way station should have cheaper real estate costs, and should have cheaper prices to lure repeat customers to travel further (more customers x cheaper price can be more than fewer x more expensive).

Help my understanding: is this really an example of game theory?

I thought that game theory was about making decisions in which your success depends on the choices of others.

In the original secretary/dowry problem the values of your choices are random. In this case the gas prices aren't random but the strategy doesn't depend on considering the choices made by the gas station owners in setting those prices.

So while it's a nice solution (and one I've used myself when looking for an apartment) it doesn't seem to me to be an example of game theory.

Not to mention that gas stations in a small region (especially a small town) are playing their own game which will probably tend to equalize prices between the stations.

I haven't seem gas stations with a 60c/gal(!) difference that are a reasonable distance apart.

Between Boston and Lexington? Sure it might be 60c cheaper, but that's almost 20 miles which would wipe out your savings and it has more to with city vs country which is predictable and doesn't require a game theory solution if you happen to be going that way.

Not sure if this is generally true, but in one freeway town I often pass through, you come off the freeway onto a main drag with lots of gas stations. Since we drive on the right side of the road, the gas stations on the left side of the road have marginally lower prices than the gas stations on the right side--presumably a premium to cover the inconvenience of making a left turn on a usually busy street!

Neat trick! Unfortunately, in the U.S. it's less interesting -- by the time you've gotten off the highway you are usually committed to the 2 or 3 signs within line of sight of the offramp.

lol, so... It's not generally a good idea to cut it this close in the US. When I drove out to cali, my buddy and decided to drive it at night to avoid traffic - this is KY to CA. So, we're clear into indiana somewhere at around 11pm and we get to about 10-15% of a tank of gas, decide to stop... pull off at the next gas sign, oh hey - it's closed??? That's interested, let's go to the next exit... 2 exits later the needle is on E and we finally get to a city with an open station. From that point on, we always filled it up at 25-33% - a fact which was crucially important in the desert and mountains.

That's definitely a risk when you get out of the city areas. Unfortunately, when minimizing stops, the greedy algorithm is the ideal solution -- it's always better to risk it.

(In graph theory, nodes don't close after midnight.)

In Western Australia, we have a state government run website, http://www.fuelwatch.wa.gov.au/ , that lists almost all WA servo prices across all fuel types for both today and tomorrow. Prices are cents per litre.

So, assuming all 5 retailers act rationally, and that they assume all their customers and competitors act rationally.

Station #1 knows most of its customers won't be comparing (don't care or running low). So they can jack prices up a little more.

Station #2 knows that the only comparison they can be sure ALL their customers make is with Station #1, which will be jacked up, so they can increase their margins a little (above avg, below #1).

#3, #4 and #5 could follow the same logic, except for this. Station #5 knows many of its customers won't compare either. They're the last option without turning around, and many comparison shoppers won't turn around for a few lousy cents / gallon. So they can increase their margin a little.

Which leaves Station #3 wanting to look reasonable compared to #1 & #2, and able to survive a comparison with at least #4. And Station #4 knows they will be a good comparison with #5 but need to compete with #3 and so will probably be at or about the same level as #3.

In other words, #1 will be most expensive. #2 & #5 will have good cause for being a little extra. And #3 & #4 will both be subject to the largest competitive forces and will likely have similar, lowest prices.

Shame the initial assumption I made was completely crap.

Given that people can drive down the road in two directions, station #1 is also #5 and #2 is also #4. This changes your analysis slightly, but the result remains that you should go to station #3.

And that the initial assumptions were crap.

If I can plan it...I usually fill up a gas station a few miles away...why? Because there are 3 gas stations on that one intersection..and their prices are usually 20-30 cents cheaper than other gas stations in the area.

And if I can't plan it...I just fill up whenever, since the difference between the super duper cheap gas station and arm and a leg gas station...is pretty small.

If you have a 10 gallon gas tank, and the difference is 50 cents....you'll only be overpaying $5 for a tank of gas.

And since the difference is usually much smaller...you'll usually just pay 1-2 bucks more to fill up now...instead of trying to find a cheaper station.

This is an amusing post. For me, buying gas is just as similar to stocks. The trick is to decide whether to fill 'er up full or not. Almost always, a full tank is a bad bet unless if you know from experience that it's the cheapest station in the area.

I say stocks because the gas in your tank is the value you bought it for that you could've gotten cheaply. I mean if you know the gas price is going to increase for sure (such as during a long weekend), then yes I will fill my tank please.

I almost always (like 97 times in 100) just fill up. Too much time wasted to buy gas a few gallons at a time for a small savings, IMO.

(Incidentally, one of our cars is a turbodiesel sedan and gets almost 500 miles to the tank, 600 on a roadtrip. It's amazingly pleasant to only have to fill up about once a month in normal usage.)

On long trips, especially coming home from long trips, I'm usually in at least a little bit of a hurry to get there and be done driving. It's therefore worth $1.50 (assuming I pay $.10 per gallon more than is necessary for my 15 gallon tank) to me to only have to fill up once. In practice, I've found the price aren't usually even that much, so the time saved is that much better of a deal.

Well I agree with the time saved part, but in the long run the $1.50s add up to something of significance.

But, my point is that when gas is expensive and you're in dire need of it, fill it up as much as is necessary until the next favorable stop.

But the time saved adds up at the same pace, as well. If spending $1.50 to save 10 minutes is worth it in each invidual case, in aggregate it's worth it as well.

In my part of around here, stations near the freeway are more expensive than stations farther from, and stations closer to the center of Big Town are more expensive than stations farther from. So you let those two forces push you to cheaper stations.

Almost always, a full tank is a bad bet unless if you know from experience that it's the cheapest station in the area.

A few years ago, I started playing the time-domain analogue of this game (which you sort of allude to later). When gas prices are low, I keep my tank close to full. I'll stop at the gas station (which is not out of the way of my normal travels) to refill even if it's only a couple gallons. When prices are higher, I can use that reserve to ride it out and wait for prices to fall again. I don't use any more or less gas this way, but I do get a marginally better price on it.

Problem rises when gas station operators, assuming they are rational, know that customers are using this strategy. In this case, they will set a price that will be function of how far and frequent other gas stations are located. The favorable outcome for customer is random chance here, somebody with less gas who would have used the same strategy would end up at more expensive station.

Since you're (presumably) driving off the highway into an unknown town, my guess is that the first station or two will have more expensive gas, because these stations know there are people coming the same way from the highway who need gas asap (and don't mind paying a premium price). This is confirmed by the example in your post.

http://gasbuddy.com/

Reply on news.ycombinator.com