Sir, you just saved my day. I had been confused about how to deal with the monthly figures of market risk. I am doing my assignment and since 2 days I was stuck here. Thank you so muchhhh, you are awesome.
Dear Professor, I am a student from physics background. Great to see your step by step tutorial on this subject. Two question: 1.Why you use AVERAGE function ( arithmetic average of return) but not GEOMEAN function ([geometric average of the (1+return) ]-1) 2.Why this E(Rtsla) from CAPM is preferred but not the E(rtsla) from compounding the daily return? Obviously Tesla (or any arbitrary portfolio) is not necessary to be (or even close to )linear relationship with the benchmark S&P500, in other words, the correlation between Tesla (or any arbitrary portfolio) and the benchmark S&P500 is not necessary to be (or even close to ) +(or -) 1. Then CAPM is just something that never makes sense. Simply compounding the daily return makes a much better sense.
Hi Mark, 1) I found the arithmetic average of the daily continuously-compounded return. I then compounded that average 365 times to make convert it to yearly (A couple options here, I could have continuously compounded it (e^(0.031%*365) - 1 = 11.98% yearly) and because there are ~252 per year I probably should have used (e^(0.031%*252) - 1 = 8.13% yearly or (1+0.00031)^252 -1 = 8.12% yearly, which is very close to (ln(3115.34/2109.99)) -1)/5 = 7.79%/year)). 2) I wouldn't say it's preferred, but if you are using the CAPM and its Beta that's the way to do it. IMHO you should always take Beta and CAPM with a grain of salt ... it's based on the past, and the R-squared on the correlation coefficient is generally low. Being you have a Physics background, it's not as precise as calculations you are used to.
@@DavidJohnk just one question. While calculating risk free rate, I understand it is already annual data but I want to take 3 months T bill rate as risk free, So, do I have to convert this average figure into 3 months by dividing it by 12 and then multiplying by 3 ?
@@gowthamdhanasekaran3908 I don't remember where I knew that from, I learned it quite a while ago. Sorry. If you look at the history you can see sometimes it's quite high. In present times it's low so hard to tell. It's also in % already.
That's a great video!... Thanks for your explanation... I would like to learn how to calculate the cost of the debt with your same case (Tesla) and to use with the CAPM to calculate the WACC
In addition to my earlier comments, John you mentioned using log returns or simple returns won't make much difference. But it's actually making in TSLA stock. Using log you are getting annualized return of 47% Vs 81% when using simple returns. Big difference. What should be the preferred way of returns in Stock when large time periods are used? Will really appreciate your comments.
The CAPM is an equilibrium model used to explain the excess return of risky securities in the capital markets. According to the model, the excess return of a risky security (the difference between the return of the security and the return of the risk-free rate) can be explained by its sensitivity to the excess return of the overall market (the difference between the return of the market and the return of the risk-free rate). This can be described by the equation 𝑬(𝒓𝒊) − 𝒓𝒇 = 𝜷(𝑬(𝒓𝒎) − 𝒓𝒇) Where 𝑬(𝒓𝒊) represents the expected return of the risky security, 𝒓𝒇 is the return of the risk-free asset. Hence, the left hand side represents the expected excess return of the risky security. 𝜷 measures the sensitivity of the excess return of the risky security to the excess return of the market overall and 𝑬(𝒓𝒎) represents the expected return of the market. If we rename the variables as 𝒆𝒓𝒊𝒔𝒌𝒚 = 𝑬(𝒓𝒊) − 𝒓𝒇 to represent the expected excess return of the risky asset and 𝒆𝒎𝒂𝒓𝒌𝒆𝒕 = 𝑬(𝒓𝒎) − 𝒓𝒇 to represent the expected excess return of the market we can write the following linear model Page 3 of 7 𝒆𝒓𝒊𝒔𝒌𝒚 = 𝜶 + 𝜷 × 𝒆𝒎𝒂𝒓𝒌𝒆𝒕 + 𝜺 If the CAPM is correct then 𝜶 = 𝟎. Find the correlation between the returns of each risky asset and the returns of the market. The FTSE100 can be used as a proxy for the market. Use the returns of each share in order run a regression for the CAPM. The risky free rate is the SONIA rate for the period. Comment on the 𝑹 𝟐 of the regression. Create a confidence interval for the parameters of the model 𝜶 and 𝜷 and comment on their statistical significance. does this video answer this question ? NEED HELP
The continuous return has properties which periodic return doesn't, but it's works fine to use the periodic return formula: (new-old)/old or the short cut periodic return formula: new/old - 1
hey, was a great video...but i was just wondering if the slope formula y/x is correct in Excel aren't the known y the dependent variables and the knowN x the independent variables?, so it should be =SLOPE(C3:C1260,F3:F1260)?
Great video David! Quick question, when calculating the expected return on a stock using Markowitz's efficient frontier portfolio theory why don't we use the expected return on a stock as calculated via CAPM as opposed to calculating its simple average- which is what is usually taught?
Good question. I'm sorry I can't really answer fully without some more thought on it. My first guess is that there's no reason you couldn't substitute the CAPM expected return. I would like to say that I don't think the CAPM is that great of a model because of the generally low r-squared values it has.
Must the timeframe of the market return be the same as the one of the Beta? Or could I use the past 5 years for calculating my Beta and the last 10 years for my market return?
Hi there, thanks for the super helpful video. Question about the rf rate. Since you converted the expected market return from daily to yearly, shouldnt those daily risk free rates be convrted to yearly as well? or am i missing something? thanks again! Also if I am doing this on a new company with only 2 years of data, is a 2 year period sufficient to get an accurate beta?
Hello sir, excellent clear video thank you. What if the expected return of a market is negative for a specific year? If we assume positive beta and a risk free rate lower than the return of the market , then would we have a negative CAPM? And what is the interpretation of a negative CAPM value? Thank you.
It's possible for the CAPM to calculate a negative expected return. Remember, the CAPM is just a model which is attempting to predict the future on past historic performance.
Have a good day Sir, thank you so much for the video. Have you ever computed the model international CAPM written by Adler and Dumas? I have read about it, but No one made an application of this model into detail. Thank you so much in advance and look forward to your answer.
Thank you for the video, so clear and helpful. I am struggling with downloading the s&p500 excel though, what can I do? It doesn’t give me the option to download. My professor wants us to find the risk free rate for 5 companies, does that mean I have to find the beta for each company first so I can check what days to use for the treasury website? ( my class is in Spanish for study abroad which is why I’m struggling a bit more with what I’m supposed to do)
Hi Dafne, You can try downloading the S&P500 here: finance.jasonstrimpel.com/bulk-stock-download/ use ^GSPC as the ticker. Yahoo finance doesn't have that option on their website any longer. If you're using daily data just use the latest 90 day treasury rate for Rf from here: www.marketwatch.com/investing/bond/tmubmusd03m?countrycode=bx for instance today it's 4.684% so the daily Rf is (1+.04684)^(1/252)-1 = 0.0182%/trading day. Hope that helps! Dr. Johnk
Great Video. Quick question, why are we using 365 days when annualising the returns. I thought it was 252 due to the fact that the stock axchange doesn't trade everyday (weekends, holidays etc) and also the closing prices are only available for those days.
When converting E(rm) daily into E(rm) yearly, shouldn't we use the number of trading days (252) as opposed to the number of calendar days (365) in a year?
Yes, I agree, that would be more correct in this case. With monthly or weekly conversion I would stick to 12 and 52. Another thing that I don't do is subtract the risk-free rate from the market and stock returns ... it doesn't affect the results much so I skipped that step.
@@DavidJohnk Thanks! Great video. I am attempting to calculate the Sharpe ratios of various weighted portfolios to find the ideal weighting of individual stocks within a portfolio. Not sure if I can combine this analysis with CAPM in a meaningful way, but interesting nonetheless.
Hi thank you for this video. One question: why would you use the average risk free rate to calculate the cost of equity and not the current risk free rate. From what i understand using a historical average risk free rate is problematic since it doesn’t take into account current market conditions.
Thanks David, just have a question about the Beta. In the video, you get to Beta = 1.197 by regressing R(S&P) with R(Tesla); but the CAPM model' beta, according to the equation, shouldn't it be the slope of the two excess returns, [R(Tesla) - Rf] / [R(S&P) - Rf]? I'm not sure the two Betas are the same? If I regress the two excess returns, will I get a different beta than 1.197? Thanks
Hi, you're technically correct, although when I recorded the video rf was so low, the beta was not efflected much at all. I can see with higher risk-free rates that you should do it that way.
Hello Professor! Thank you for this video, I just have a couple questions that I was not able to find answers to online to help with my research. 1. What time period should I use when doing these models? 2. Do different time periods tell me different things? 3. Should I only be using the past year of data if I want to analyze a current portfolio? Or will going as far back as 2015 help? Or maybe that far back is not as significant for us nowadays. Sorry to ask so many questions, if instead of answering these you could let me know what book/article would be good for the technical side of the model please let me know! Thank you!
1. There is no set rule but I would set the time length accordingly to how often you plan to trade, longer periods for longer times between trades. If your studying 10 years of data I would go weekly instead of daily. 2. and 3. I think answered those in 1. a year would be good Good luck
For the Returns calculation, whats the difference between using the method of: (New Value/Original Value) - 1, vs using (New Value-Original Value) / Original Value?
hello professor, if we were applying this model for a multiple assets porfolio for 3 months only on a daily-data, should we convert the risk free rate or not ???
Thanks a lot professor, i'am actually doing different stock portfolio analysis i was wondering if i should use the optimal porfolio for this model or just the one i chosed first
Thank you so much! I had been multiplying daily returns by 52. For converting weekly return standard deviation into annualized return standard deviation, can you just do sqrt(52)*weekly return standard deviation?
Hi! Great video ! I have a question, they gave us tbill uk tender middle rate (UKTBTND) and I need to run regressions every month to find a and beta of stocks for every month that i have. Rm= ftse 100 , do I need to calculate the annual tbill uk? And how?
Very nicely explained. Thanks so much. Question: In case I am using 5 year data for a stock and market, can I just take an average of the 5 year return, or is this not the right way? Will really appreciate your response.
I have a question related to market return as you said in the video we should convert to annual, so if my data are daily for 2020 and 2021 only, It is right to make an average for only 2 years ( annual ) and ignore the rest of years? thank you very much
Hi David, Great video. I want to ask one question did you mistakenly took 26 weeks for 90days T-billswhen you delete the headings and count to 7. I think it should be count 6 for 13-weeks and not count 7. please clarify.
No, it was the coupon equivalent 13 weeks on the 7th column. The first column is the date. I don't think it would be a huge error to use the 26 week anyway :)
If weekly return are used and then annualised, should Beta with COVARIANCE.P(C3:C1260,F3;F1260)/VAR.P(C3:C1260) be multiplied by root(52) to annualise Beta
Would it not be more accurate to convert daily into yearly prices based on the number of observations per year- that being raising to the power of 252 vs 365? Or am I missing something here? Thanks
Great video but one question: if the treasury rate yields are 90-days and are the daily yields, how can we say that they are already yearly? Also, if I used the 6-month Yield curve rates, are those also annualized and will I have to convert them into semi annual rates to compare them with a 6 month market return?
@@DavidJohnk Thank you for your response, Professor. So the 6-month rates are not of T-Bills which mature in 6 months but are the annualized rates? So I must divide the 6-month T-Bills by 2 to get the annualized rate?
@@sudroy You can either annualize your six-month return to r(annual) =(1+r(six-month))^2 - 1 or make the t-bill annual yield a r(six-month) = (1+r(t-bill))^(1/2) - 1 ... don't just divide by two.
How does one interpret the result given from CAPM? Do we compare E(Rtsla) to E(Rtsla) yearly and what would this mean for the value of the stock? Thanks for the video!
Great question! The CAPM is a backward-looking model that tries to predict the expected return in the future. The key word is "expected". In real life returns are almost always different, they could be lower or higher. I would definitely use more than one model for stock valuation, not just the CAPM. The CFI website has a good summary: corporatefinanceinstitute.com/resources/knowledge/trading-investing/stock-valuation/
@@DavidJohnk Thanks! Based on the result we got from the CAPM here, does it look like investing in Tesla stock is a good or bad idea? Why is the expected return based on average historical data so far from the expected return calculated using CAPM?
@@klaramanning2589 Hi Klara, the difference in calculations is due to the methodology difference. Both estimates are based on past performance and should be taken with a grain of salt. Tesla has done very well the last year, but it is a very volatile stock. If you are planning on holding the stock for a longer time and don't get nervous with large up and down price swings, I think it might be a great idea to hold some of their stock in a portion of the total value of a portfolio. I think Tesla has many synergies and a solid future strategy, including two huge factories being built (Austin, TX and Berlin, Germany) and a factory expansion in China. They also have a supercharger network, state of the art self-driving technology, new models including the Cybertruck, which has over 650,000 preorders and counting. I could keep going but when I invest I look at the future for the company, and I think Tesla is pretty solid in that regard. Of course, if you research it, you will see people with the opposite opinion to mine. I do "put my money where my mouth is" as I presently own Tesla stock. Hope that helps!
Hi, thank you for the great video. But I have a question. When calculating E(Rm) , what power should be used in the yearly equation if the monthly average is on 8 years of monthly data (96 months)? Will it just be 12?
Yes this is a time-series regression, you can take a look at this video also: th-cam.com/video/j65sNTANYVk/w-d-xo.html He uses monthly data, runs a different regression, and subtracts the risk-free rate from the numbers to get Excess returns (technically, a more correct way to do it)
I don't suppose you know how to look up daily treasury bills for Canada? i am at a road block when it comes to calculating CAPM. HELP!! Thanks a million!
Thanks! GM is listed on the S&P 500, but the CAPM theory is based on the entire market, not just the S&P 500. I used the S&P 500 as an estimate of the entire market. Hope that makes sense.
@@DavidJohnk thank you! May I ask why the average rf that you calculated is yearly? I thought you took the daily rates? Shoudn't you need to convert it into the effective yearly rate?
Great question, dividing and subtracting 1 gives the daily rate. The beta calculation will be close either way. Dividing the natural logarithms gives a continuous return. Not a big difference, but continuous returns are additive (return for day 1 plus return for day 2 adds up to the continuous return for both days). Hope that makes sense!
Hello Professor thanks for the clarification! Can i know since my data for risk free rate is daily , how can i make it to monthly risk free rate ? ! Appreciate your help
Assume a 30-day month and your daily rate is .05%/day. It would be: (1+.0005)^30-1 = 1.5109%/month. Just multiplying by 30 would be 1.5000%/month (an error)
You can do a Beta for each, then the portfolio Beta will be a simple weighted avg of the betas. It's one of the great things about Beta! If you invest $1,000 total and $250 in one and $750 in the other your weights would be 25% and 75% respectively. The portfolio beta would be equal to (beta stock one X 25%) + (beta stock two X 75%). You could also calculate Beta on the portfolio return, it would be the same. Great question!
Hi Alberto, I used the 90-day T-bill because that's the one I see most often in my financial research readings. It kind of makes sense to use a shorter maturity because it has a smaller maturity risk premium.
@@DavidJohnk Thanks for the reply this makes perfect sense. If you don't mind I have another question. In the part where you are anualazing the market return, is there a reason why you are using the formula "R_yearly=(1+R_daily)^365 ", from my understanding this formula is used when dealing with simple returns. Since you are compounding the return, shouldn't the multiplication of the average daily market return and 365 days suffice? Given the properties of using natural logarithm. Thanks again for taking your time and answering my question :)
@@albertogonzalez8398 You're correct, the returns were calculated with ln(end/begin) therefore it's continuously compounded. The calculations will most likely be very close as long as you stay consistent between Market and security/portfolio. Times 365 days or 252 trading days will also work. The fed doesn't do t-bill rate continuously either, but that number is very small anyway. like I said, I think if you stay consistent you should be pretty close.
can anyone please try calculating KLSE expected rate of return? What is the reason behind it get a negative value rate of return on KLSE Malaysia index?
@@DavidJohnk I am working on an exam, where I've been provided with the monthly return on 10yr government bonds over a 99 month period. I'm expected to calculate an estimate for the risk-free rate with this information, and I am not quite sure how to do that. Could you explain? Thank you David
One question; why you didn't convert average Rf to yearly... for market return you converted daily to yearly but not for risk free rate though you downloaded daily 90-day treasury bill.
It seems to me that there is something strange about the annualized return formula. If daily return is only 0.03%, how is it possible that annual return be 11.97%.
Excel automically turns the percantages into numbers when calculating. However, when you use calculator, yu have got to remember that 0.03% actually has a value of 0.0003 not 0.03. If you use the actual value (0.0003) and put that into the expected yearly return formula, you should get the same answer as the one David got in excel.
Hello! The dataset we're working with has montly data (for both market return and stock return) of the past 10 years. In that case, how do we calculate the risk free rate of return for CAPM? Could we still use this daily data from the treasury.gov website?
The data from treasury.gov is in annualized returns already expressed in %. You can use the monthly data but annualize it. You should keep units consistant, hope that makes sense.
This is one of those rare gems that one stumbles upon on TH-cam and is eternally grateful for. Thank you very much for the fantastic upload!
I'm glad it was helpful!
Sir, you just saved my day. I had been confused about how to deal with the monthly figures of market risk. I am doing my assignment and since 2 days I was stuck here. Thank you so muchhhh, you are awesome.
Glad it helped!
Helped a lot during my undergraduate project about Samsung Electronics!
Awesome, glad it helped!
me struggling in 2024 being happy after seeing this very useful vid. Thank you very much dear mister
It helped me to do my Finance futures contract assignment perfectly. Love from India ❤️
Great! Glad it helped.
Exactly what I’ve been looking for to complete my assignments. Thank you so much!
Glad it helped!
Great explanation on how CAPM is calculated. Excellent. thanks
You're welcome. Glad it was useful.
Dear Professor,
I am a student from physics background. Great to see your step by step tutorial on this subject.
Two question:
1.Why you use AVERAGE function ( arithmetic average of return) but not GEOMEAN function ([geometric average of the (1+return) ]-1)
2.Why this E(Rtsla) from CAPM is preferred but not the E(rtsla) from compounding the daily return? Obviously Tesla (or any arbitrary portfolio) is not necessary to be (or even close to )linear relationship with the benchmark S&P500, in other words, the correlation between Tesla (or any arbitrary portfolio) and the benchmark S&P500 is not necessary to be (or even close to ) +(or -) 1. Then CAPM is just something that never makes sense. Simply compounding the daily return makes a much better sense.
Hi Mark, 1) I found the arithmetic average of the daily continuously-compounded return. I then compounded that average 365 times to make convert it to yearly (A couple options here, I could have continuously compounded it (e^(0.031%*365) - 1 = 11.98% yearly) and because there are ~252 per year I probably should have used (e^(0.031%*252) - 1 = 8.13% yearly or (1+0.00031)^252 -1 = 8.12% yearly, which is very close to (ln(3115.34/2109.99)) -1)/5 = 7.79%/year)). 2) I wouldn't say it's preferred, but if you are using the CAPM and its Beta that's the way to do it. IMHO you should always take Beta and CAPM with a grain of salt ... it's based on the past, and the R-squared on the correlation coefficient is generally low. Being you have a Physics background, it's not as precise as calculations you are used to.
@@DavidJohnk Understood. Much appreciated with your prompt reply!
By far the most practical way to calculate return thru CAPM. Thanks a lot😘😘😘
Thanks! You're welcome.
Thanks from Malaysia. Fantastic explanation.
You're welcome, I'm glad to hear it helped.
That's very helpful, thank you!
You're welcome, glad it helped.
Thank you for this video. It was of great help.
Glad it helped!
@@DavidJohnk just one question. While calculating risk free rate, I understand it is already annual data but I want to take 3 months T bill rate as risk free, So, do I have to convert this average figure into 3 months by dividing it by 12 and then multiplying by 3 ?
@@netexponent6217Hi, I'm not sure what you're trying to do, but here's how I would convert annual to every quarter: rquarterly = (1 + rannual)^(1/4)-1
Thank you so much. Great work
You're welcome, Thanks for the positive feedback
Mr Johnk, could you please explain why daily risk-free data turns into annual after averaging it? (at 15:15)
Hi, It is already annual and in percent from the fed. All we did was avg. it. Hope that makes sense.
@@DavidJohnk how did u identify that it is already annual?
@@gowthamdhanasekaran3908 I don't remember where I knew that from, I learned it quite a while ago. Sorry. If you look at the history you can see sometimes it's quite high. In present times it's low so hard to tell. It's also in % already.
Thank you very much your lecture helped me a lot
You're welcome, glad it helped!
That's a great video!... Thanks for your explanation... I would like to learn how to calculate the cost of the debt with your same case (Tesla) and to use with the CAPM to calculate the WACC
Hii, thanks for this fanatastic video. Its really easy to understand.I reached late here but still saved alot time.
I'm glad it helped!
Another lifeguard to be grareful at! 👍🏼
Thanks Man it helped me so much. Good Job.
Glad it helped!
Amazing video... Thanks pretty much Sir
You're welcome
awesome thanks for sharing this video, Sir David! :) wishing you a great year 2021 :)
You're welcome glad it was of use to you.
Thanks a lot! Helped me a lot
You're welcome, I'm glad it helped!
Terrific explanation. Thank you
Thanks for the positive feedback!
In addition to my earlier comments, John you mentioned using log returns or simple returns won't make much difference. But it's actually making in TSLA stock. Using log you are getting annualized return of 47% Vs 81% when using simple returns. Big difference. What should be the preferred way of returns in Stock when large time periods are used? Will really appreciate your comments.
The CAPM is an equilibrium model used to explain the excess return of risky securities in the capital markets. According to the model, the
excess return of a risky security (the difference between the return of the security and the return of the risk-free rate) can be explained by its
sensitivity to the excess return of the overall market (the difference between the return of the market and the return of the risk-free rate). This
can be described by the equation
𝑬(𝒓𝒊) − 𝒓𝒇 = 𝜷(𝑬(𝒓𝒎) − 𝒓𝒇)
Where 𝑬(𝒓𝒊) represents the expected return of the risky security, 𝒓𝒇 is the return of the risk-free asset. Hence, the left hand side represents the
expected excess return of the risky security.
𝜷 measures the sensitivity of the excess return of the risky security to the excess return of the market overall and 𝑬(𝒓𝒎) represents the
expected return of the market.
If we rename the variables as 𝒆𝒓𝒊𝒔𝒌𝒚 = 𝑬(𝒓𝒊) − 𝒓𝒇 to represent the expected excess return of the risky asset and 𝒆𝒎𝒂𝒓𝒌𝒆𝒕 = 𝑬(𝒓𝒎) − 𝒓𝒇 to
represent the expected excess return of the market we can write the following linear model
Page 3 of 7
𝒆𝒓𝒊𝒔𝒌𝒚 = 𝜶 + 𝜷 × 𝒆𝒎𝒂𝒓𝒌𝒆𝒕 + 𝜺
If the CAPM is correct then 𝜶 = 𝟎.
Find the correlation between the returns of each risky asset and the returns of the market. The FTSE100 can be used as a proxy for the
market.
Use the returns of each share in order run a regression for the CAPM. The risky free rate is the SONIA rate for the period. Comment on the 𝑹
𝟐
of the regression.
Create a confidence interval for the parameters of the model 𝜶 and 𝜷 and comment on their statistical significance. does this video answer this question ?
NEED HELP
Why do you use continuous return (ln(new/old) when calculating ERm instead of just percent change ((new-old)/old)?
The continuous return has properties which periodic return doesn't, but it's works fine to use the periodic return formula: (new-old)/old or the short cut periodic return formula: new/old - 1
hey, was a great video...but i was just wondering if the slope formula y/x is correct in Excel aren't the known y the dependent variables and the knowN x the independent variables?, so it should be =SLOPE(C3:C1260,F3:F1260)?
Hi Esha, it should be Tesla returns on the y-axis (dependent variable).
You are a life saver. Thank you so much for this video and your help!
You're welcome, glad it was helpful!
Great video David! Quick question, when calculating the expected return on a stock using Markowitz's efficient frontier portfolio theory why don't we use the expected return on a stock as calculated via CAPM as opposed to calculating its simple average- which is what is usually taught?
Good question. I'm sorry I can't really answer fully without some more thought on it. My first guess is that there's no reason you couldn't substitute the CAPM expected return. I would like to say that I don't think the CAPM is that great of a model because of the generally low r-squared values it has.
Great video sir. The beta you came up with ie. Co var / var one is it levered beta or unlevered beta?
It's unlevered. Levered Beta takes in to account capital structure effect.
this helped me a lot!! Thank you
Glad it helped!
Thank you so much!!!
You're welcome, glad it helped.
Thanks for the great upload
You're welcome, glad it was of help.
Must the timeframe of the market return be the same as the one of the Beta? Or could I use the past 5 years for calculating my Beta and the last 10 years for my market return?
I would try to keep with the same timeframe
Hi there, thanks for the super helpful video. Question about the rf rate. Since you converted the expected market return from daily to yearly, shouldnt those daily risk free rates be convrted to yearly as well? or am i missing something? thanks again! Also if I am doing this on a new company with only 2 years of data, is a 2 year period sufficient to get an accurate beta?
The Fed already has the risk-free rates yearly and in percent.
Hello sir, excellent clear video thank you. What if the expected return of a market is negative for a specific year? If we assume positive beta and a risk free rate lower than the return of the market , then would we have a negative CAPM? And what is the interpretation of a negative CAPM value? Thank you.
It's possible for the CAPM to calculate a negative expected return. Remember, the CAPM is just a model which is attempting to predict the future on past historic performance.
Have a good day Sir, thank you so much for the video. Have you ever computed the model international CAPM written by Adler and Dumas? I have read about it, but No one made an application of this model into detail. Thank you so much in advance and look forward to your answer.
No, I haven't. I will look into it.
@@DavidJohnk Thank you so much Sir, in my opinion the ICAPM is a bit complicated. I will look forward to your video about it. Have a nice day Sir.
Thank you so much Professor! You are a life saver
You're welcome, glad it helped!
Can somebody answer what to use as the risk free rate? Do we use the bond rate or the treasury rate for our risk free rate?
I see the 10 year treasury bond rate used occasionally
Thank you for the video, so clear and helpful. I am struggling with downloading the s&p500 excel though, what can I do? It doesn’t give me the option to download.
My professor wants us to find the risk free rate for 5 companies, does that mean I have to find the beta for each company first so I can check what days to use for the treasury website?
( my class is in Spanish for study abroad which is why I’m struggling a bit more with what I’m supposed to do)
Hi Dafne, You can try downloading the S&P500 here: finance.jasonstrimpel.com/bulk-stock-download/ use ^GSPC as the ticker. Yahoo finance doesn't have that option on their website any longer. If you're using daily data just use the latest 90 day treasury rate for Rf from here: www.marketwatch.com/investing/bond/tmubmusd03m?countrycode=bx for instance today it's 4.684% so the daily Rf is (1+.04684)^(1/252)-1 = 0.0182%/trading day. Hope that helps! Dr. Johnk
Great Video. Quick question, why are we using 365 days when annualising the returns. I thought it was 252 due to the fact that the stock axchange doesn't trade everyday (weekends, holidays etc) and also the closing prices are only available for those days.
you're right 252 is better!
@@DavidJohnk thank you
Hi there! Thank you for your assignment saving video. Quick question, why do you use population var and covar instead of sample?Thank again
Sample is better.
When converting E(rm) daily into E(rm) yearly, shouldn't we use the number of trading days (252) as opposed to the number of calendar days (365) in a year?
Yes, I agree, that would be more correct in this case. With monthly or weekly conversion I would stick to 12 and 52. Another thing that I don't do is subtract the risk-free rate from the market and stock returns ... it doesn't affect the results much so I skipped that step.
@@DavidJohnk Thanks! Great video. I am attempting to calculate the Sharpe ratios of various weighted portfolios to find the ideal weighting of individual stocks within a portfolio. Not sure if I can combine this analysis with CAPM in a meaningful way, but interesting nonetheless.
@@matthewrosseland9096 I have a video where I show how to do that. th-cam.com/video/U-QiAGXS3aE/w-d-xo.html
More or less. If you have any questions, feel free to reach out to me. You know where to find me.
hello, thx for the video very useful, but when i calculate the average for the risk free i have the error #DIV/0! how can i fix that? thank you.
Not sure, check the formula for zero's is all I can think of.
Hi thank you for this video. One question: why would you use the average risk free rate to calculate the cost of equity and not the current risk free rate. From what i understand using a historical average risk free rate is problematic since it doesn’t take into account current market conditions.
I've seen the current rate often used, that's a good way and easier to do.
Thanks David, just have a question about the Beta.
In the video, you get to Beta = 1.197 by regressing R(S&P) with R(Tesla); but the CAPM model' beta, according to the equation, shouldn't it be the slope of the two excess returns, [R(Tesla) - Rf] / [R(S&P) - Rf]?
I'm not sure the two Betas are the same? If I regress the two excess returns, will I get a different beta than 1.197? Thanks
Well in fact, the differences are minor. Thanks anyway, the video is very helpful!
Hi, you're technically correct, although when I recorded the video rf was so low, the beta was not efflected much at all. I can see with higher risk-free rates that you should do it that way.
Thank you so Much
You're very welcome.
Hello Professor!
Thank you for this video, I just have a couple questions that I was not able to find answers to online to help with my research.
1. What time period should I use when doing these models?
2. Do different time periods tell me different things?
3. Should I only be using the past year of data if I want to analyze a current portfolio? Or will going as far back as 2015 help? Or maybe that far back is not as significant for us nowadays.
Sorry to ask so many questions, if instead of answering these you could let me know what book/article would be good for the technical side of the model please let me know! Thank you!
1. There is no set rule but I would set the time length accordingly to how often you plan to trade, longer periods for longer times between trades. If your studying 10 years of data I would go weekly instead of daily.
2. and 3. I think answered those in 1. a year would be good Good luck
Dear Sir, thank you for the video. May I ask why the annualised return on market is not used instead (geometric mean)?
I'm not sure what you mean, I did use a geometric calculation when converting the avg return to yearly.
Apologies for not being clear. I would like to ask if we can use the annualized rate of return for the market instead?
For the Returns calculation, whats the difference between using the method of: (New Value/Original Value) - 1, vs using (New Value-Original Value) / Original Value?
The two formulas give exactly the same answer. The first one is considered easier by most.
@@DavidJohnk ok thanks prof!
hello professor, if we were applying this model for a multiple assets porfolio for 3 months only on a daily-data, should we convert the risk free rate or not ???
You could convert the risk free to daily and just leave everything in daily. Just keep units consistent
Thanks a lot professor, i'am actually doing different stock portfolio analysis i was wondering if i should use the optimal porfolio for this model or just the one i chosed first
Hi, What if your rf is suppose to monthly?
Thank you so much! I had been multiplying daily returns by 52. For converting weekly return standard deviation into annualized return standard deviation, can you just do sqrt(52)*weekly return standard deviation?
You're welcome!
Hi! Great video ! I have a question, they gave us tbill uk tender middle rate (UKTBTND) and I need to run regressions every month to find a and beta of stocks for every month that i have. Rm= ftse 100 , do I need to calculate the annual tbill uk? And how?
So grateful found this video.. Btw from the TSLA CAPM expected return, it was considered undervalued?
Here's a video I made on that subject, not sure it will help: th-cam.com/video/-85S9nHUI88/w-d-xo.html
Very nicely explained. Thanks so much. Question: In case I am using 5 year data for a stock and market, can I just take an average of the 5 year return, or is this not the right way? Will really appreciate your response.
I have a question related to market return as you said in the video we should convert to annual, so if my data are daily for 2020 and 2021 only, It is right to make an average for only 2 years ( annual ) and ignore the rest of years?
thank you very much
Yes, I would just work with two years of data to make the time-frames match.
GREAT WORK
Thanks for the positive feedback.
Hi David, Great video.
I want to ask one question did you mistakenly took 26 weeks for 90days T-billswhen you delete the headings and count to 7. I think it should be count 6 for 13-weeks and not count 7. please clarify.
No, it was the coupon equivalent 13 weeks on the 7th column. The first column is the date. I don't think it would be a huge error to use the 26 week anyway :)
@@DavidJohnk Thank you so much. your video helped me a lot in finishing one of the CAPM assignments.
while calculating Rf you have taken bond yield daily one as yearly. could you please check it!
If weekly return are used and then annualised, should Beta with COVARIANCE.P(C3:C1260,F3;F1260)/VAR.P(C3:C1260) be multiplied by root(52) to annualise Beta
Would it not be more accurate to convert daily into yearly prices based on the number of observations per year- that being raising to the power of 252 vs 365? Or am I missing something here? Thanks
You're correct, it would be more accurate.
Great video but one question: if the treasury rate yields are 90-days and are the daily yields, how can we say that they are already yearly?
Also, if I used the 6-month Yield curve rates, are those also annualized and will I have to convert them into semi annual rates to compare them with a 6 month market return?
Great question! They annualize it, similarly to what was done in the video. The six-month is the same.
@@DavidJohnk Thank you for your response, Professor. So the 6-month rates are not of T-Bills which mature in 6 months but are the annualized rates? So I must divide the 6-month T-Bills by 2 to get the annualized rate?
@@sudroy You can either annualize your six-month return to r(annual) =(1+r(six-month))^2 - 1 or make the t-bill annual yield a r(six-month) = (1+r(t-bill))^(1/2) - 1 ... don't just divide by two.
@@DavidJohnk Thank you so much, Professor! What is the formula you used in the (1+(t-bill)^0.5-1?
How does one interpret the result given from CAPM? Do we compare E(Rtsla) to E(Rtsla) yearly and what would this mean for the value of the stock? Thanks for the video!
Great question! The CAPM is a backward-looking model that tries to predict the expected return in the future. The key word is "expected". In real life returns are almost always different, they could be lower or higher. I would definitely use more than one model for stock valuation, not just the CAPM. The CFI website has a good summary: corporatefinanceinstitute.com/resources/knowledge/trading-investing/stock-valuation/
@@DavidJohnk Thanks! Based on the result we got from the CAPM here, does it look like investing in Tesla stock is a good or bad idea? Why is the expected return based on average historical data so far from the expected return calculated using CAPM?
@@klaramanning2589 Hi Klara, the difference in calculations is due to the methodology difference. Both estimates are based on past performance and should be taken with a grain of salt. Tesla has done very well the last year, but it is a very volatile stock. If you are planning on holding the stock for a longer time and don't get nervous with large up and down price swings, I think it might be a great idea to hold some of their stock in a portion of the total value of a portfolio. I think Tesla has many synergies and a solid future strategy, including two huge factories being built (Austin, TX and Berlin, Germany) and a factory expansion in China. They also have a supercharger network, state of the art self-driving technology, new models including the Cybertruck, which has over 650,000 preorders and counting. I could keep going but when I invest I look at the future for the company, and I think Tesla is pretty solid in that regard. Of course, if you research it, you will see people with the opposite opinion to mine. I do "put my money where my mouth is" as I presently own Tesla stock. Hope that helps!
@@DavidJohnk That’s very helpful, thank you!
Another helpful video. Thanks David! I just subscribed to your channel.
You're welcome, thanks for subscribing!
hey prof will you paste the file over here, the link abv is not working
Hi, thank you for the great video. But I have a question.
When calculating E(Rm) , what power should be used in the yearly equation if the monthly average is on 8 years of monthly data (96 months)? Will it just be 12?
Yes, if it's 12 observations per year use 12. Hope that helps!
@@DavidJohnk Thank you very much. This helps a lot.
Hi, i have a question. Is this a time-series regression. Also is the expected returm the same as the excess return?
Yes this is a time-series regression, you can take a look at this video also: th-cam.com/video/j65sNTANYVk/w-d-xo.html
He uses monthly data, runs a different regression, and subtracts the risk-free rate from the numbers to get Excess returns (technically, a more correct way to do it)
I don't suppose you know how to look up daily treasury bills for Canada? i am at a road block when it comes to calculating CAPM. HELP!! Thanks a million!
I'm not sure, in many cases, you can use the US risk-free rate.
Great video, can I ask why do you use S&p 500 as the market. If using for example GM, do I still use the same market?
Thanks! GM is listed on the S&P 500, but the CAPM theory is based on the entire market, not just the S&P 500. I used the S&P 500 as an estimate of the entire market. Hope that makes sense.
@@DavidJohnk Makes sense, thanks a bunch!
Hello!! Thank you for the video! May i check with you if we can use 26 weeks or 52 weeks treasury yields instead of just 90 days?
That will work.
@@DavidJohnk thank you! May I ask why the average rf that you calculated is yearly? I thought you took the daily rates? Shoudn't you need to convert it into the effective yearly rate?
@@desmondyang8529 Treasury already reports it as yearly and in % no conversion necessary
Is there a specific reason for using LN?
Why not divide the new by the old price then subtract one?
Great question, dividing and subtracting 1 gives the daily rate. The beta calculation will be close either way. Dividing the natural logarithms gives a continuous return. Not a big difference, but continuous returns are additive (return for day 1 plus return for day 2 adds up to the continuous return for both days). Hope that makes sense!
what's the difference between the expected return of tesla that we acquired from yahoo finance and the CAPM one?
One was simply looking at the historical average returns, the other CAPM expected return.
Hello Professor thanks for the clarification! Can i know since my data for risk free rate is daily , how can i make it to monthly risk free rate ? ! Appreciate your help
Assume a 30-day month and your daily rate is .05%/day. It would be: (1+.0005)^30-1 = 1.5109%/month. Just multiplying by 30 would be 1.5000%/month (an error)
Posting the risk-free rate on daily basis doesn't mean it's a daily rate (it still might be a yearly rate, changing each day).
I need to calculate this on 8 years of monthly historical data? What would be the power used for calculating yearly E(Rm)?
Sir, when you calculated Risk free rate why its yearly ?
What if you are comparing more than two stocks? Average them, or do a beta for each and every single combination?
You can do a Beta for each, then the portfolio Beta will be a simple weighted avg of the betas. It's one of the great things about Beta! If you invest $1,000 total and $250 in one and $750 in the other your weights would be 25% and 75% respectively. The portfolio beta would be equal to (beta stock one X 25%) + (beta stock two X 75%). You could also calculate Beta on the portfolio return, it would be the same. Great question!
while transform the data, my data have some null data, how to clear it
Thank you so much sir...🙏👍
You're welcome, glad it helped!
Where can you find coupon rates for times dating back to 1980's. The U.S Dept of Treasury only goes back to 2002
I don't know off the top of my head. Sorry couldn't help more!
How do we calculate the covariance matrix belong to yearly?
Thank you SO MUCH!
You're welcome!
Great video!
Thanks! I'm glad it helped.
another great video
Glad you enjoyed it
Why we take 90day?
You don't have to use the 90 day, it's very typical the 10 year bond is often used as well in financial research.
Thank you for this great video! I'm a bit curious as to why did you choose the 90 Day T-bill vs 182 or 364 Days T-bill?
Hi Alberto, I used the 90-day T-bill because that's the one I see most often in my financial research readings. It kind of makes sense to use a shorter maturity because it has a smaller maturity risk premium.
@@DavidJohnk Thanks for the reply this makes perfect sense. If you don't mind I have another question. In the part where you are anualazing the market return, is there a reason why you are using the formula "R_yearly=(1+R_daily)^365 ", from my understanding this formula is used when dealing with simple returns. Since you are compounding the return, shouldn't the multiplication of the average daily market return and 365 days suffice? Given the properties of using natural logarithm. Thanks again for taking your time and answering my question :)
@@albertogonzalez8398 You're correct, the returns were calculated with ln(end/begin) therefore it's continuously compounded. The calculations will most likely be very close as long as you stay consistent between Market and security/portfolio. Times 365 days or 252 trading days will also work. The fed doesn't do t-bill rate continuously either, but that number is very small anyway. like I said, I think if you stay consistent you should be pretty close.
I don't know whether anyone will reply to me or not, but what can we do if the average of daily rm comes negative?
Why did you convert the market return to yearly but not the risk free return?
Risk free rate is already yearly
sir, Can the calculation of the return be applied to the Sharpe ratio?
yes, it's just (rj - rf)/std dev j ... make sure frequency (daily, monthly, etc.) is the same on all calculations.
@@DavidJohnk thanks sir, your video helps me a lot
Why is the first date left blank when you calculated the daily returns?
You need two days to calculate the return. You lose one day because of that.
should the days line up?
Yes, they should. That's why I try to keep the dates beside each set of data. So I can keep an eye on that.
@@DavidJohnk oh man I am using the SP 500 as my benchmark but when I check the dates it skips some dates.
@@mikeylejan8849 You could take the average of the date on either side to get close. Hope that makes sense!
can anyone please try calculating KLSE expected rate of return? What is the reason behind it get a negative value rate of return on KLSE Malaysia index?
Good day professor, may I ask for your email to seek for our short research consultation related to irregular TBills in the Philippines. Thank you!
Is there a way to calculate the beta of only the past 3 months?
Sure just use daily data for 3 months.
how is the risk free rate already in yearly return?
That's the way the fed provides it.
@@DavidJohnk I am working on an exam, where I've been provided with the monthly return on 10yr government bonds over a 99 month period. I'm expected to calculate an estimate for the risk-free rate with this information, and I am not quite sure how to do that. Could you explain? Thank you David
@@DavidJohnk Please, sir?
hello, why have you chosen 13 weeks for the coupons?
Many finance professionals use the 90-day (13 week) t-bill for the risk-free rate. Sometimes other maturity US treasury bonds are used and acceptable.
Hi David, can you please tell me why did you select that column only( 13 weeks bank discount coupon equivalent) as Rf rate?
I wanted the 90-day t-bill.
@@DavidJohnk okay. Got it. Thanks..
One question; why you didn't convert average Rf to yearly...
for market return you converted daily to yearly but not for risk free rate though you downloaded daily 90-day treasury bill.
It seems to me that there is something strange about the annualized return formula. If daily return is only 0.03%, how is it possible that annual return be 11.97%.
The formula is fine, but you may want to use 252 instead of 365 due to that many observations in a year.
Excel automically turns the percantages into numbers when calculating. However, when you use calculator, yu have got to remember that 0.03% actually has a value of 0.0003 not 0.03. If you use the actual value (0.0003) and put that into the expected yearly return formula, you should get the same answer as the one David got in excel.
Hello! The dataset we're working with has montly data (for both market return and stock return) of the past 10 years. In that case, how do we calculate the risk free rate of return for CAPM? Could we still use this daily data from the treasury.gov website?
The data from treasury.gov is in annualized returns already expressed in %. You can use the monthly data but annualize it. You should keep units consistant, hope that makes sense.
What about Systematic/unsystematic rate?
The systematic rate is the Market risk premium (rm - rf). Unsystematic is everything else.
Shouldn't we use (1+daily chang%)^trading days to get the yearly return?
I've seen it done both ways ... that might be more correct since there are ~252 observations.