# Risk Tolerance Questionnaire – Scoring and Blending

Today’s post is about building a composite Risk Tolerance Questionnaire from the three publicly available ones that I have found.

Contents

###### Risk Tolerance Questionnaire

A couple of weeks ago I looked at the various aspects of Risk Tolerance:

- risk aversion
- risk capacity
- risk tolerance
- risk perception
- risk composure

So we know broadly how to use Risk Tolerance to inform our asset allocation, but how do we measure our tolerance?

- I offered a simple approach last week (max equity percentage = max drawdown you can tolerate * 2.5), but we’d really like a questionnaire.

There are many services that will test your risk tolerance for a fee.

- I have no doubt that the market leaders are satisfactory, but since I haven’t paid to complete any of them myself, I won’t advertise their names here.
- If anyone has used one of the leading services, please let us know what you thought in the comments section below.

Publicly available questionnaires are thin on the ground.

- I’ve managed to find three, and today I’ll focus on blending them together into a single questionnaire.

###### Blending challenges

I won’t name the three source questionnaires here, but with a few minutes googling, you should be able to find them.

- Let’s take a look at each one in turn.

Risk Tolerance Questionnaire number 1 (RTQ1) has 13 multiple choice questions

- there are 8 questions with 4 possible answers, 3 questions with 3 possible answers, and 2 questions with 2 possible answers
- the questionnaire comes with a scoring grid, which is almost simple
- questions with 4 answers are marked from 1 to 4, those with 3 answers from 1 to 3, and those with 2 answers are marked either 1 or 3
- usually the scores are in ascending order, but one question is in reverse order
- so the maximum score is 47 (8*4 + 5*3), and the minimum score is 13
- there are five results bands: 18 or below, 19-22, 23-28, 29-32, 33 or above

Risk Tolerance Questionnaire number 2 (RTQ2) has 8 questions with multiple choice answers, and an additional number input (maximum drawdown tolerated).

- 4 questions have 5 possible answers, 3 questions have 4 possible answers and 1 question has two possible answers (but could be modified to have more)
- the scoring grid uses multiples of either 3 or 4 (depending on whether there are 4 or 3 answers) to reach 12 for the riskiest answer on each question)
- so the maximum score is 84 (12 *7, since question 1 is used as a filter rather than being scored)
- there are 6 results bands: 0, 1-11, 12-32, 33-53, 54-74, 75+
- both 0 and 1-11 result in an all income (no risk) portfolio, so I will combine these two bands so that I have five in total
- each of the results bands converts to a simple asset allocation (growth / income split, equivalent to the traditional bond / equity split)
- note that two of the questions present the same information, once in graph form and once in text form – I will probably combine these two questions, to give a total of 7 from RTQ2

Risk Tolerance Questionnaire number 3 (RTQ3) has 7 questions with multiple choice answers

- all 7 questions have five possible answers
- the scoring grid uses scores from 1-5 for each question
- so the maximum score is 35 (and the minimum is 7)
- there are five results bands: 7-10, 11-17, 18-24, 25-31, 32-35
- I also have a completed version of RTQ3, with choices that score to 28, placing the investor in the 25-31 band
- this is labelled “Aggressive”, and a target asset allocation is provided
- I don’t have asset allocations for the other bands, but this one band can be used to benchmark RTQ3’s allocations against RTQ2’s allocations

###### Blending approach

So assuming there are no duplicates, we have 13 + 7 + 7 = 27 questions in total.

At first sight, blending the questionnaires together should not be difficult:

- all the questions can be rebased to scores out of twelve
- all RTQs have five scoring bands – I will map RTQ1 and RTQ3’s scoring bands to RTQ2’s scoring bands, since RTQ2 comes with target asset allocations that I can use as a basis for my own

I have a few other rules that I will try to implement:

- I plan to simplify the questions where possible
- round numbers will be used where appropriate
- possible answers will always run in ascending order of riskiness
- reference to dollars will be converted to pounds (since I am targeting UK investors)

My suggested asset allocations will be to (a) return assets (equity) and (b) uncorrelated / stability assets

- the latter is non-equity, which could include property, commodities, cash, bonds, DB pensions and other asset classes

Since I also believe in investors owning their own home, I will also provide alternative asset allocations for 25% property ownership (ex-London) and 40% property ownership (to represent London).

###### Tool to build questionnaire

I haven’t assembled a questionnaire since my undergraduate days, and we used to just type out the questions and photocopy them in those days.

Nowadays there are a lot of online tools.

- I will be using Google Forms, as it’s a tool I’ve wanted to learn more about for some time
- it should also make sharing the questionnaire (and collecting any results) quite easy

Let’s see.

###### Scoring the questions

Here is how I will convert the existing questions’ scores to the 1-12 mechanism.

First, RTQ1:

The multiplier is the amount by which we’ve increased the average score on a question.

- We might need this when we look at blending the scoring charts together.

Next, RTQ2:

This ought be simple, but we are merging questions 2 and 8, and question 1 is normally used as a filter and not scored.

- As luck would have it, these two effects cancel each other out, and the net multiplier for the 7 questions we are using remains 1.

Finally, RTQ3:

This is simple, since all 7 questions are scored in the same way.

###### Scoring bands

I looked into using the average multiplier to scale up the ranges, but unfortunately the multiplier isn’t constant across the range of answers.

- For example, on RTQ3, the multiplier from a score of 1 becoming 0 is zero, but the multiplier on a score of 5 becoming 12 is 2.4.

So instead I decided to use the share of the total scoring range for the questionnaire that is occupied by each band.

- This is not a statistically perfect approach but it should be good enough for a first pass.
- If anyone knows of a better (and relatively easy to implement) solution, please let me know in the comments.

Here’s how this works for RTQ1:

Here’s the RTQ2 table (the simplest, since the new bands are the old bands):

And here is the RTQ3 table:

Finally, here are the combined ranges for the five bands, out of a grand total of 324 points (12 * 27):

All of these charts will be modified if I find that some of the questions are duplicates.

- The current weightings only (approximately) work if I end up using 13, 7 and 7 questions from RTQ1, RTQ2 and RTQ3 respectively.

That’s it for today.

- Next week I will actually combine the questions and present the questionnaire and the scoring sheet.

Until next time.

*Share this with , Twitter, Google+, Pinterest, LinkedIn, Tumblr, Reddit and StumbleUpon.*