Risk Score Calculation
TIFIN Wealth calculates the estimated forward Risk Score of individual securities and portfolios using a parametric Value at Risk (VaR) calculation at the 99% level for a one-year period. TIFIN Wealth estimates the forward expected return and volatility of the security or portfolio using Capital Market Assumptions found in the Investment Forecasts settings page.
Let's take a look at an example portfolio with a risk score of a 33 that falls within the client's risk band of a 26 - 40.
Portfolio Risk Score: there is a 1% chance that the portfolio will lose 30% or more of its value over the next 12 months.
Risk Preference Score of 26 means the client is willing to lose up to 18% in the chance of a market downturn.
Risk Capacity Score of 40 means the client has the assets, income, job stability, and overall ability to recover from a 40% drawdown in the portfolios returns.
How are the Risk Scores calculated?
To calculate a Portfolio’s Risk Score, TIFIN Wealth uses forward looking returns, volatility, and correlation between the assets (using CMAs) to determine the expected return and volatility of the portfolio. Then, the portfolio's expected return & standard deviation is used to calculate the 1% VaR (Value at Risk) of the overall portfolio.
Methodology: Forward Looking Assumptions
Large fluctuations in interest rates and loose monetary policy can create material differences between realized (past) and expected (future) returns across various asset classes including bonds and equities. Several sophisticated investment managers have adjusted their models and believe that future returns will be lower than the returns we have witnessed over the past few decades.
TIFIN’s risk methodology utilizes forward returns for all securities based on 3 factors:
Expected Asset Class Returns & Volatility – TIFIN averages Capital Market Assumptions (CMAs) reported across the largest global asset managers (i.e., JP Morgan, BlackRock, etc.) to generate a “wisdom of the experts” estimate. These fund managers have significant teams researching and developing capital market assumptions that are foundational for all investors. Assumptions are updated bi-annually when available.
Exposure to Primary Asset Class – determine the variability of returns through a rolling 5- year regression between the security and its primary asset class.
Value at Risk (VaR) – the portfolio is then given a score based on Value at Risk (VaR). TIFIN calculates the estimated forward Risk Score of individual securities and portfolios using a parametric VaR calculation at the 99% confidence level for a 12 month period.
In this example, the recommended portfolio has a Risk Score of 22. Net of this advisor's fees, this is showing:
Possible losses of -23.04%; meaning the recommended portfolio value could lose up to $15k (VaR) or more in a given year (Portfolio Value x Risk Score %)
Possible gains represent a potential 33.24% in upside portfolio value gains within a given year using the same distribution at a 1% chance.
Capital Market Assumptions
The chart below shows the 10-year Capital Market Assumptions (CMAs) reported across several asset manager partners. We source CMAs from over 10 of the largest global financial institutions such as JP Morgan, Morgan Stanley, BNY Mellon, Wells Fargo, BlackRock, & others.
Calculating Expected Returns & Volatility
TIFIN uses these aggregated Capital Market Assumptions (CMAs) to forecast security and portfolio-specific returns and volatility. Using Geometric Brownian Motion, TIFIN Wealth projects the Current and Recommended portfolios’ balances for 10 years.
Geometric Brownian motion (GBM) is used to describe random movements of financial assets over time where the size of the variations is proportional to the current value of the asset. GBM is the industry standard model used in finance to forecast security returns. The benefits of GBM over Monte Carlo (if you make standard industry assumptions such as normality and statis volatility) is speed (runs much faster) and will return the same results every time due to forecasting every point on the distribution versus making a random distribution. Potential drawbacks of industry standard GBM (and Monte Carlo) is the assumption of constant volatility and no drastic stock price jumps; however, over long enough time periods these potential drawbacks become mitigated.
Here is a breakdown of how we calculated expected forward returns & volatility:
Assign each security within the portfolio to 1 of the 17 asset classes we have CMAs for based upon the security's Morningstar category.
Calculate the security's exposure to the asset class using a 5-year regression of monthly data.
Use the security's exposure to the asset class and the CMAs of the asset class to calculate the security's expected return and volatility.
After calculating the forward expected return and volatility for every security in the portfolio, take the historical correlation between the securities in the portfolio over the last 5 years to project out a forward-looking covariance matrix (measure of how two assets move together and the magnitude in which they move together).
History: proxy data is used to backfill history if the security does not have 5 years’ worth of data.
Calculating Expected Forward Looking Returns
Multiply the CMA Expected Return * Beta = Forward Looking Return
Map the security to the appropriate asset class > look at the forward looking returns per the asset class
Run a 5 year regression to identify the beta (how variable are the returns compared compared to US Large Cap)
Example
US Large Cap CMA Expected Return = 6.94%
The exposure of fictitious security ABC to the Large Cap asset class is 1.65 (65% more volatile than the Large Cap asset class)
Now take 1.65 * 6.94% to determine ABC's forward looking return of 11.5%
Calculating the Expected Forward Looking Volatility
Take the daily volatility of ABC / daily volatility (from CMAs) of its respective asset class
Then multiply the above ratio by the expected forward looking volatility (from CMAs) of it's respective asset class
1.85 * 16.37%= 30%
Now, we use the formula to calculate the 1% VaR, or Risk score for ABC Stock
11.5% - (2.317*30) = 58
Complete the above steps for each individual security in the portfolio
In short, to calculate the Portfolio’s Risk Score, TIFIN uses forward looking returns, volatility, and correlation between the assets (using CMAs) to determine the expected return and volatility of the portfolio.
Calculating Correlations of Portfolio Holding
Once we've calculated each individual security's expected return and volatility, we also calculate the forward pairwise covariance matrix (another way to think about this is the correlation) between each security in the portfolio. We then use each security's weight, each security's expected volatility, and the pairwise covariance (correlations) between each holding in the portfolio to calculate the portfolio's standard deviation.
For reference here is a screenshot of the formula to calculate portfolio standard deviation, which includes the covariance mentioned above:
This is for a 2 asset portfolio:
For a three asset portfolio:
We calculate the portfolio's expected return which is a simple weighted average and then use the portfolio's expected return and the portfolio standard deviation to calculate the 1% VaR of the portfolio.