Keep this life skill for later
Sign in to save
← All life skills

Financial literacy · a life skill

Advanced Expense Tracking

the usual week is not the average week

Advanced tracking looks at the shape of each category, not just its total. A week's spending can't go below zero, but it can run as high as the week demands, so most categories have plenty of ordinary weeks and a few big ones. The ordinary week, the one you remember, sits below the average. Measure each category's shape, forecast the lopsided ones at their average, and name the big week before it arrives.

  • Challenging
  • Life-Changing impact
  • Learn it: 30 days
  • Twelve weeks per category
a lopsided category, a bigger miss
Weekly forecasts against what budgeting app users then spent, for online shopping and groceries Online shopping, the more lopsided category: forecast as usual, 102 pounds against 151 spent, 49 short. After giving a reason the week might differ, 142 against 139, 3 over. Groceries: forecast as usual, 77 against 98, 21 short; after a reason, 81 against 96, still 16 short. NEXT WEEK'S FORECAST AND WHAT WAS SPENT · MEAN £ forecast spent that week ONLINE SHOPPING · SKEWNESS .98 Forecast as usual £48.65 short £102 £151 Gave a reason first £3.10 over £142 £139 GROCERIES · SKEWNESS .55 Forecast as usual £21.44 short £77 £98 Gave a reason first £15.52 short £81 £96 £60 £80 £100 £120 £140 £160

Study 5 of Howard, Hardisty, Sussman & Lukas (2022): 1,738 users of a UK budgeting app each forecast one category for the coming week, and the app tracked what they spent.

The method

Measure the shape, then forecast the average.

  1. Pull twelve weeks for one category

    Groceries, eating out, online shopping: whatever your bank app or tracking sheet already splits out. Twelve weeks, not four, because a lopsided category can look even for a month.

  2. Find the middle week and the average

    Sort the weeks and take the middle one: that's close to the week you think of as usual. Then add them all up and divide by twelve.

  3. Read the shape

    If the average sits well above the middle week, a few big weeks are carrying it. That category is lopsided, and a forecast from memory will come in short.

  4. Forecast lopsided categories at their average, and name the big week

    Use the average, not the usual, for those categories. Then write one reason next week could be a big one: an order you've been putting off, a birthday, a trip.

  5. Re-run it each month

    Drop the oldest weeks, add the newest. Categories change shape with the seasons, a new hobby or a move, and the forecast should follow.

    Twelve weeks, middle and average, shape, forecast, repeat.

Example: Tomás pulls twelve weeks of online shopping from their bank app. Most weeks are between £0 and £35; two are £140 and £210. The middle week is £16.50 and the average is £41.17. Forecasting from the middle week, they'd come in about £25 short in an average week, £1,283 over a year. So they forecast £41, and note that a new pair of running shoes is overdue.

The skew lab

See the shape of each category.

1 · Pick a category

2 · Its last twelve weeks

Oldest first, from your bank app or tracking sheet. Put 0 for a week with nothing in it; leave a week blank if you don't know it.

    3 · Its shape

    Each dot is a week. Red dots: weeks above the average.

    4 · Every category

    CategoryMiddle weekAverageShape

    Your weeks are kept only in this browser, so next month's check can start from this one. Nothing is sent anywhere.

    The research behind it

    Spending has a long right tail.

    Why does a forecast built on the usual week come in short? Ray Howard, David Hardisty, Abigail Sussman and Marcel Lukas point to the shape of spending. A week's spending is fenced in at zero on one side and open on the other. So most weeks bunch together near the low end, and a few big ones stretch far out to the right. Statisticians call that positive skew.

    In a skewed category, the usual week sits at the peak, and even the middle week falls below the average. The average is what adds up over a month, but the usual week is what memory offers. In the authors' first study, members of a Canadian credit union had weekly totals with a skewness of 1.03, high by the usual rule of thumb, and their forecasts fell short week after week.

    An illustration of a right-skewed spending category. A curve rises steeply from zero to a peak, the usual week, then falls away in a long tail of big weeks. The middle week sits to the right of the peak, and the average week further right still. THE SHAPE OF A LOPSIDED CATEGORY · AN ILLUSTRATION the usual week middle week average week the long tail: a few big weeks 0 · nothing spent, the floor bigger weeks →
    Not data: the general shape the authors describe. The usual week is the most common amount; the middle week splits the weeks in half; the average is pulled right by the tail.

    To test the shape on its own, they showed 400 people 52 weekly amounts, one every 1.2 seconds, and asked them to forecast week 53. Both sequences averaged $200 and varied by the same amount. In one, the weeks were spread evenly either side of $200. In the other, the most common week was $190, with fewer, higher weeks balancing it.

    After the lopsided sequence, 31% forecast under $200, against 19% after the even one, and the average forecast fell from $198.00 to $191.41. The share forecasting over $200 barely moved. The authors call this a conservative test: people had just watched every week go by, which real life never shows you.

    Forecasts for week 53. After even weeks: 18.7% under 200 dollars, 64.0% exactly 200, 17.2% over. After lopsided weeks: 31.0% under, 50.8% exactly, 18.3% over. FORECASTS FOR WEEK 53 · SHARE OF PEOPLE under $200 exactly $200 over $200 Weeks spread evenly around $200 mean $198.00 18.7% 64.0% 17.2% Most weeks $190, a few higher mean $191.41 31.0% 50.8% 18.3% 0% 25% 50% 75% 100%
    Study 4 of Howard et al. (2022), 400 online participants, preregistered. Same average, same spread; only the shape differed, and forecasts slid towards the most common week.

    The last study used real money. A UK budgeting app emailed about 100,000 of its users, and 1,738 forecast either their online shopping or their groceries for the coming week, while the app tracked what they spent. Over the previous twenty weeks, online shopping was the more lopsided category, and forecasts for it fell further short: 32% under, against 22% for groceries.

    Half were first asked for at least one reason next week might differ from a typical week. For online shopping, that lifted forecasts by 39% and they landed within £3.10 of what was spent. For groceries it barely moved them, and the gap stayed.

    Two categories are only two, and they differ in more than shape: online shopping also ran higher and varied more. The researchers controlled for both and the result held. The skew lab, the middle-week comparison and the twelve-week window are our own steps built on these findings, not something the studies tested.

    Online shopping had a skewness of 0.98 and usual forecasts fell 32% short. Groceries had a skewness of 0.55 and usual forecasts fell 22% short. TWO CATEGORIES, SIDE BY SIDE how lopsided (skewness) usual forecast fell short by Online shopping 0.98 32% Groceries 0.55 22% 0 0.5 1.0 0% 20% 40%
    Study 5 of Howard et al. (2022). Skewness over each person's previous twenty weeks in the app; shortfall as a share of what was spent, for those who forecast as usual.

    The idea

    Why it works.

    A forecast is really a guess at the average week, because the average is what adds up over a month or a year. Memory hands you the usual week instead. In an even category the two are close, and the usual week will do. In a lopsided one, the gap between them is the money that keeps surprising you.

    Measuring the shape tells you which categories to distrust. Forecasting those at their average puts the big weeks back into the number, and naming a reason tells you which week the big one is likely to land in.

    When it goes wrong

    Habits that help.

    • Split before you measure. “Shopping” that mixes groceries with one-off purchases hides both shapes. Measure them apart.
    • Keep the zero weeks. A week with no takeaway is part of the shape. Leaving it out pushes the middle week up and hides how lopsided the category is.
    • Don't trim the big weeks. They feel like exceptions, but in a lopsided category the exceptions are the pattern. Every few weeks, one comes.
    • Average the forecast, name the reason. The average spreads a big week across all of them. A reason says when it's due, so the week it lands isn't a shock.

    Practise this in HourLife.

    In Finances: Expense log. Log your money once a month, set a monthly budget and track spending by category, and keep every subscription and its charges in one list.