Buying train tickets is the definition of a crapshoot. Sometimes you'll get lucky, and when you plug your travel times into the National Rail website you'll be pleasantly surprised to find you can bag yourself a cheap, flexible off-peak ticket. But most of the time, you end up piecing together the cheapest combination of advance tickets that you can find.
The reason train tickets fluctuate all over the place is simple: Train companies want to extract as much cash as possible from travellers, while making sure to fill up trains - so will raise and cut prices based on estimated demand, using some clearly fairly opaque algorithms that we consumers never get to see.
The good news for us then is that if private ticket-seller The Trainline is to be believed, it has figured out a way to fight back: By using its enormous dataset of user journeys and ticket bookings to predict when tickets will hike in price - and therefore predict when is the optimal time for you to buy in order to save the most cash.
The "Price Prediction" feature will draw its data from 125,000 journeys booked every day, as well as past user searches, and will be analysed with some clever data analytics techniques. The upshot of what this could conceivably do is fairly stark when you give it an actual case study.
According to the company, a standard class advance single from London Euston to Manchester Piccadilly is £32 if you book 80 days before you intend to travel - and this rises to £38 with 41 days to go, £42 with 13 days to go, £87 with two days to go, and to an eye-watering £126 if you book on the day you intend to travel. So if the company's prediction algorithm is right, you could save £88 by being amazingly forward-thinking on your travel.
Now all they need to do is figure out how to predict whether you'll be stuck in front of a kid who kicks the back of your seat for the whole journey and life will be perfect.