Balloon Pump

What you see

The balloon sits in the middle of the board with the current pot written inside it and its type (A, B, C or D) written underneath. Under the balloon, a row of chips shows the earlier balloons of the same type: each chip is the number of pumps that balloon took, plain if you banked it and red if it burst. Below the board are two buttons, Pump and Bank; you can also press P to pump and B to bank. Two pills above the board show the round (the balloon you are on, for example Round 3 / 10) and your banked points.

Each pump adds a fixed number of points to the pot and inflates the balloon. If you bank, the pot moves to your points and the next balloon is tied on. If the balloon bursts, it shows Burst, the pot is lost, and after a short pause the next balloon appears. You must pump at least once before you can bank. A line under the buttons says what happened to the last balloon, for example Balloon 3 banked: 7 pumps, 35 points. When every balloon is done, the board lists all your balloons by type.

One run of the game is one trial: a set of balloons, from 6 at level 1 up to 28 at level 10. There is no time limit, and speed has no effect on the score.

How the balloons behave

• Each balloon type has a limit. A type-A balloon at level 1 can take at most 16 pumps; other types go up to 8, 24, 48, 64, 96 or 128, depending on the level.
• Burst points are spread evenly across each type's range. Within a type, the game cuts the range up to the limit into equal slices, one per balloon, and puts a burst point in the middle of each slice, rounded to the nearest pump and never below 2. Then it shuffles the order. So each type has some early bursters, some late ones and some in the middle, and the same set of burst points comes back every time you meet that type at that level; only the order changes.
• The first pump is always safe. No balloon bursts on pump one.
• Balloons come in blocks by type. You meet all the type-A balloons first, then all of type B, and so on, starting with the smaller-limit types.
• The balloon's size tracks its type's range. Every type reaches the same full size at its limit, so a sturdy type grows in smaller steps per pump. The size never tells you where this particular balloon will burst.

What changes with level

What it measures

Balloon Pump feeds the Risk calibration facet and carries 32 per cent of your Decision Making score.

It descends from the Balloon Analogue Risk Task (BART), published by Carl Lejuez and colleagues in 2002. In the standard version each pump earned 5 cents in a temporary bank, and a balloon could burst anywhere from the first pump to the 128th, with an average burst point of 64; the original study also used more fragile balloon types with lower average burst points. Participants could collect at any time; a burst lost that balloon's money. The usual score is adjusted average pumps: the mean number of pumps on balloons that did not burst. Burst balloons are left out because a burst cuts off how far you would have gone.

In the first study of 86 adults, riskier pumping correlated with sensation seeking and impulsivity and with self-reported addictive, health and safety risk behaviours, and explained variance in those behaviours beyond questionnaires. Retested on separate days about two weeks apart, adjusted pumps correlated at about 0.77. Over longer gaps behavioural risk tasks are less stable: in a study of 39 risk measures, their average six-month reliability was about 0.46, so treat any single run as one reading, not a verdict.

• Expected return
• Chance of losing it all

Classic BART treats more pumping as more risk-seeking. Our version adds what the facet name promises: calibration. Several balloon types with very different limits appear in one run, as in the original's multiple balloon colours, so a fixed habit cannot work. You have to learn each type's range and adjust.

How it is scored here

For each balloon type the game works out the best-paying fixed pump count: the count that banks the most on that type's set of burst points, found by trying every count from 0 to the limit (on a tie it takes the lower count). Because the set is the same every run at a level, that count is the same too:

Four of these tie with a higher count that banks the same: 8 pumps for the 16-limit type at levels 1 and 4, 6 for the 10-limit type at level 7, and 4 for the 8-limit type at level 10. The benchmark for your run is what the best counts bank on your balloons, which is the most any single count per type can bank. Your score is what you banked divided by that benchmark, capped at 100 per cent (varying your count from balloon to balloon can beat it by luck).

• 98 per cent or more counts as correct and moves you up a level for the next run; banking nothing at all counts as incorrect and moves you down; anything else holds your level. New sessions start at level 3.
• Speed is not scored.
• With so few balloons per type, a rule of thumb is not enough for full marks. Stopping at exactly half of every type's limit scores between 70 and 100 per cent depending on the level (100 only at level 1); three-quarters of the limit scores 44 to 76 per cent; pumping to the limit scores nothing. No single share of the limit, rounded up, down or to the nearest pump, reaches 98 per cent at every level: the best manages 87 per cent at its weakest level. These figures are exact, since every run at a level has the same burst points.

Your session report shows your score as Value captured and adds three descriptive measures that never change the score. Pumps per balloon against the best count is your average pumps minus the best counts, in pumps: more than half a pump above means you pushed past the point where the odds turn, more than half a pump below means you banked early. Change in pumps after a burst compares each balloon that follows a burst with the balloon that burst, and gives the average difference in pumps. Balloons burst is the count.

Across sessions your Balloon Pump score sits on our 0–100 scale, where 50 means what the difficulty you work at predicts. Runs with more balloon types count as harder. A score appears after about four runs, small samples are pulled towards 50, and recent sessions count for more.

Where you will meet it

The balloon task is the most widely reported game in commercial batteries.

Business Insider's description of the pymetrics balloon game matches the classic design: pump for 5 cents, collect at any time, lose the pot if it bursts, as many balloons as you can in three minutes. The magazine reported that a cautious style is neither better nor worse than an adventurous one, and quoted pymetrics' chief executive: there is no right or wrong in the spectrum. Prep sites add that the pymetrics balloons come in colours with different burst behaviour, and that the Arctic Shores battery also includes a Balloons game; those details are candidate-reported.

The method

1. Count pumps, not points

   The pot is pumps × points per pump. The first pump shows you the points per pump, since the pot then holds exactly one pump's worth, and the table above gives it for every level. At 5 per pump, a pot of 40 means 8 pumps. Think in pumps; it makes each type's range easy to compare.

2. Read a new type from its growth

   When a new type letter appears, watch how much the balloon grows with each pump. Every type reaches the same full size at its limit, so small steps mean a sturdy type with a big range, and large steps mean a fragile one.

3. Aim just under a burst point near the middle

   Start near half the type's range: the balloon is then roughly three-quarters of its full width. The best count is always one pump short of one of the type's burst points, between 37 and 58 per cent of the limit, and the table above gives it for every level.

4. Update from every burst

   A type's burst points are spread evenly, so their average sits near half the range, and the same set comes back every run at a level. Each burst shows you one of them exactly. Note them per type and settle on stopping one pump below the one that pays best.

5. Hold the rule within a type; change it between types

   Balloons of the same type deserve the same stopping point. When the type letter changes, reset: the next type almost certainly has a different range.

Worked example 1: one type, six balloons

Level 1: one type with a limit of 16, six balloons, 5 points a pump. The six burst points are always 2, 4, 7, 9, 12 and 15, in a shuffled order. A balloon banks your k pumps only if k is below its burst point.

How much would you bank by always stopping at 3, 8, 12 or 15 pumps?

Show the answer

Stop at	Balloons that survive	Banked
3	the five bursting at 4 or later	5 × 3 × 5 = 75
8	9, 12, 15	3 × 8 × 5 = 120
12	15 only (12 bursts on pump 12)	1 × 12 × 5 = 60
15	none	0

The benchmark is the best count, 6 pumps, which banks 120 (four survivors × 6 pumps × 5). Stopping at 8 ties it at 120, but 7 banks only 105 (three survivors × 7 × 5). So 6 and 8 score 100 per cent; 7 scores 88 per cent; 3 scores 63 per cent; 12 scores 50 per cent; 15 scores nothing. Notice that the balloon bursting at 12 pops on the 12th pump: to bank it you must stop at 11 or fewer.

Worked example 2: the sturdier type arrives

Level 3. You have just finished five type-A balloons (limit 16) and settled on banking at 7 pumps, that type's best count. The type under the balloon switches to B, whose limit is 48, and its balloons grow in noticeably smaller steps. If you keep stopping at 7, what happens?

Show the answer

With five balloons of a 48-limit type, the burst points are 5, 14, 24, 34 and 43. Stopping at 7 banks four of them: 4 × 7 × 5 = 140. The best count for type B is 23 pumps, which banks the three balloons bursting at 24, 34 and 43: 3 × 23 × 5 = 345. Exactly half the limit, 24 pumps, does worse: the balloon bursting at 24 pops, leaving 2 × 24 × 5 = 240.

Carrying type A's habit into type B scores 140 ÷ 345, about 41 per cent on those balloons. The fix is the step-two cue: small growth steps mean a big range, so the balloon at 8 pumps still looks small. Keep pumping until it is about three-quarters of its full size.

Worked example 3: the early burst

Level 5, type B (limit 32). The very first type-B balloon bursts on pump 3. Rattled, you bank the remaining four at 2 pumps each. How did that compare with sticking to the type's best count of 15?

Show the answer

The five burst points for this 32-limit type are 3, 10, 16, 22 and 29. You met the 3 first. Banking the other four at 2 pumps earns 4 × 2 × 5 = 40. Stopping at 15 banks the balloons bursting at 16, 22 and 29: 3 × 15 × 5 = 225. The panic cost more than four-fifths of the value.

Because burst points are spread evenly within a type, an early burst means the low end of the range has been used up, not that the type is fragile. One burst should nudge your estimate, not replace it. Your report's Change after a burst shows whether you tend to overreact like this.

Tips and traps

Levelling up

Higher levels add types (up to four), widen the spread of limits (8 to 128 at levels 9 and 10), and give you only five to seven balloons of each type to learn from. Points per pump rise from 5 to 8, which makes every burst hurt more in absolute terms but does not move the ideal stopping point.

• Learn faster. With five balloons per type, you cannot afford two or three to calibrate. Rely on the growth cue from the first pump of each new type.
• Watch the extremes. At level 9 an 8-limit type is best at 4 pumps and a 128-limit type at 74; at level 10 they are best at 3 and 63. Both appear in the same run.
• Expect to hold levels. Moving up needs 98 per cent of the benchmark, so a single misjudged type can hold you. Aim for clean, type-by-type calibration rather than heroics.

1. Week 1
   • Three runs; before each type, say its stopping point aloud
   • After each run, check Risk appetite: aim for between −0.5 and +0.5 pumps
   • Note how the balloon looks at half its range
2. Week 2
   • Three runs; judge each new type from growth per pump within its first three pumps
   • Keep one stopping point per type unless two bursts in a row say otherwise
   • Check Change after a burst: aim for small adjustments
3. Week 3
   • Two runs at your current level; target 98 per cent or more
   • Rework worked example 2 with the limits you met
   • Compare your banked total with the benchmark shown in the review

Check yourself

5 practice questions

1. For a balloon type whose burst point is equally likely anywhere up to a limit of 40 pumps, where is the value-maximising stopping point?
2. A balloon's burst point is 12. What is the most pumps you can make and still bank it?
3. How is a run of Balloon Pump scored?

and 2 more

Create a free account to answerSign in

Sources

1. Lejuez, C. W., Read, J. P., Kahler, C. W., Richards, J. B., Ramsey, S. E., Stuart, G. L., Strong, D. R., & Brown, R. A. (2002). Evaluation of a behavioral measure of risk taking: The Balloon Analogue Risk Task (BART). Journal of Experimental Psychology: Applied, 8(2), 75–84. doi.org/10.1037/1076-898X.8.2.75
2. White, T. L., Lejuez, C. W., & de Wit, H. (2008). Test–retest characteristics of the Balloon Analogue Risk Task (BART). Experimental and Clinical Psychopharmacology, 16(6), 565–570. doi.org/10.1037/a0014083
3. PsyToolkit. Balloon Analogue Risk Task (BART): experiment library entry. psytoolkit.org
4. Frey, R., Pedroni, A., Mata, R., Rieskamp, J., & Hertwig, R. (2017). Risk preference shares the psychometric structure of major psychological traits. Science Advances, 3(10), e1701381. doi.org/10.1126/sciadv.1701381
5. Feloni, R. (2017, June 28). Consumer-goods giant Unilever has been hiring employees using brain games and artificial intelligence. Business Insider. businessinsider.com
6. JobTestPrep. Free pymetrics games online practice and guide (candidate-preparation site; descriptions are unofficial). jobtestprep.com
7. JobTestPrep UK. Arctic Shores assessment (candidate-preparation site; descriptions are unofficial). jobtestprep.co.uk

Put it into practice: Balloon PumpSign in to try it