Wednesday, 28 October 2015

Update to Real Time Asset Allocation Experiment

I'm going to post a link to my real time asset allocation spreadsheet at the top of the blog, re: the index funds I've been buying.

I know it's not exciting, but I want to be able to be accountable for (hopeful) accumulation (as opposed to destruction) of wealth over time.

Last weekend, I posted a hypothetical study of compound return regarding dollar cost averaging into SPY = the market.  I came to the conclusion that a systematic monthly allocation into x % SPY (or equivalent) and x% bonds for my long term tax deferred accounts using a rule based methodology (i.e., add incremental units only when I am getting a lower price) may result in an expected compound return over time (before dividends and interest) of between 4% and 5%.

I further reasoned out that rather than arbitrarily add incremental units of the market once a month based on lower monthly market closes vs. the previous month, that I should expect compound returns to improve by almost 100 bps over time by adding more incremental capital once cumulative annual returns go negative.

So with all of this mumbo jumbo in mind, here's where I'm at, and here's what I have done.

On August 25th, I allocated $900 into various index funds at my discount brokerage across the accounts I managed.  These were allocated evenly across:
  • TD e-Series CDN Index
  • TD e-Series US  / DJIA CDN $ hedged Index
  • TD e-series Int'l / Euro Index CDN $ hedged
  • TD d-series NA Dividend Index
  • TD e-series CDN bond index

On September 23rd, I allocated $3,300 into various index funds at my discount brokerage across the accounts I managed.  These were allocated evenly across the same funds.

Here is my template for tracking the monthly index unit allocations.  For what it's worth, October will be a non-equities allocation month as I have not been able to get a lower price for any of the individual equity index funds I already own (or for SPY for that matter).  I have allocated some additional capital into the bond index this month. The bond index I'm tracking below is based on the prices of the ishares Canadian government bond index fund.

DatePrice SPY (input)Units IndexPrice Bonds (input)Units BondsCumulative unitsM2MP/L Per unitCumulative costValue Index (input)Value Bonds (input)P/L unitsCpd. annualized return
8/25/2015187.234.8122.240.004.81900.000.0090090000.00%0
9/23/2015193.614.4622.0422.6941.964,230.6230.624,2002,8005000.73%4.45%


A few comments:

  • Although the study uses SPY and ishares XGB, the actual holding period returns will differ between what I've bought each month vs. what I'm using to above to track real time compound cumulative annualized return.  The actual cost basis of the "Value Index" and "Value Bonds" cells in the table above is based on what I've actually allocated in the accounts.  For the most part, I expect the correlation of returns between the index funds I've bought and SPY to be fairly close over time.
  • The compound cumulative annualized return is nonsensical for the first year.  I'm doing a straight annualization of the monthly P/L, which holds little significance over too short a time period.  The compound cumulative annualized return really only starts making sense after the first year (or three)
  • While the historical study I originally did looked at month end closing prices of historical SPY for hypothetical allocations, I over-rode the system rules (which didn't actually exist two months ago) by allocating capital on days in the market where I perceived there to be an opportunity to allocate capital.  In retrospect, these two dates, August 25th and September 23rd (my birthday) turned out to be very good days to have bought the index.  I didn't know this at the time, and I had I known then what I know now, I should have allocated much more capital at the time!  But, I'm really just a schmuck writing a blog hoping to make my way towards a long term compound return of between 4% and 6% on a portion of my retirement assets, and I don't expect to be able to make great timing choices ad infinitum.  If I perceive really grizzly days in the market intra-month, I may override the time constraint rules if SPY is lower at that point than the previous month's close, but for the most part, I will stick to the rules.


Update to New 52 Week Lows Screen

As I've continued tracking 52 week lows in order to generate ideas for further research, I've discovered the following:

1) The process of tracking 52 week lows is more important than finding ideas:

I started this process in July 2015 and I've managed to refine my own internal process of screening along the way.  For example, when I first started this exercise, my weekly list was close to 200 entries: I basically took every single new low off the new lows listing ad hoc.

As I've progressed through my daily review of new lows, I have a number of filters going on inside my head before ideas make my tracking spreadsheet:


  • Reasonable to (preferably) no leverage, 
  • Preferable small to micro-caps, 
  • Ignore P/E within reason, as it's symptomatic of what has happened, not what may happen going forward.  Also, negative eps in any one year is not a reason for exclusion as loss years may be symptomatic of transition.
  • Want to see high enough gross and operating margins over at least 5 years (possibly indicative of pricing power / competitive advantage)


2) Pretax cap rate (EBIT / EV) may produce a number of interesting ideas for further research, but FCF Cap Rate is a better metric for identifying candidates:

I've developed my tracking list with a view to researching ideas where the pretax cap rate, EBIT / EV, is sufficiently high (in excess of 15%).  I am not seeing a ton of ideas (especially lately).

But what I'm really interested in is valuation.  Since I've been ranting all along on the merits of free cash flow, I've decided to add a TTM FCF column to my list, and I've done a very quick and dirty "perpetual" FCF calculation using FCF / .12 , +/- net cash (the difference between EV and mkt cap) in order to take the ratio of 1 - ((FCF / .12) +/- net cash ) / Market cap.

This is my own quick and dirty calculation for margin of safety at 12% Re.

This way, I have both my preliminary pretax cap rate + a simple analytic for FCF perpetual no growth valuation lined up next to each other on the screen.  If I get hits with a high enough pretax cap rate in excess of 15% and a possible capitalized FCF discount to current market cap, I may have the beginnings of an idea for further research

3) All of the above is very boring:

It is a monotonous process.  It means sitting down at my screen nightly and logging results in. Hopefully though, the process is akin to "painting a fence".  Here's a quote from Joel Greenblatt's Little Book on just this:


"I LOVE MOVIES, and The Karate Kid is one of my favorites. Of course, I would like any art form where eating popcorn and candy are part of the deal. But there is one scene in this particular movie that holds special meaning for me. In it, the old karate master, Mr. Miyagi, is supposed to be teaching his teenage apprentice, Daniel, how to fight. The boy is new at school and being bullied by a group of karate-trained toughs. Daniel hopes learning karate will help him stand up to his tormentors and win the girl of his dreams. But instead of teaching him karate, Mr. Miyagi puts Daniel to work—waxing cars, painting fences, and sanding floors.

So after a whirlwind of scenes showing poor Daniel working his fingers to the bone—waxing, painting, and sanding—the youth has finally had enough. He confronts Mr. Miyagi and essentially says, “Why am I wasting my time doing these simple and menial tasks when I should be learning karate?” Mr. Miyagi has Daniel stand up from his sanding duties and starts throwing jabs at the young boy while yelling “Wax on! Wax off!” Daniel deflects each jab with the swirling motions he learned from so many hours waxing cars. Next, Mr. Miyagi throws a punch while yelling “Paint the fence.” Once again, Daniel deflects the punch, this time using the up-and-down action of painting a fence. Similarly, Mr. Miyagi’s karate kick is then stopped by Daniel’s expert floor-sanding ability. 

In effect, by learning these few simple techniques, Daniel has unwittingly become a karate master. Now in good movies, the viewer participates in something called the willing suspension of disbelief. In other words, we kind of know that Ralph Macchio, the actor who plays Daniel in the movie, couldn’t really use that waxing thing to defend himself in a dark alley. In the real world, before he could finish his first coat, Mr. Macchio would probably get smacked in the head and drop like a sack of potatoes. But while caught up in a movie, we’re ready and more than willing to believe that Mr. Miyagi’s simple methods can truly work wonders."




Monday, 26 October 2015

Home Capital Group December 10, 2018 Bonds

Sorry equities, another rant on fixed income.

It's funny, I ran a screen on TD's fixed income database mid-week last week and I found the following result, and I wanted to do a quick post on my thought process last week but I didn't have time:



















My first thought on finding these bonds was something along the lines of:

"Ok, these yield 3.35% to maturity (on the offer), but I have to pay par + accrued interest.  What's my spread over Rf?"

So I searched the 3 year GIC's on offer, and here's what I found (I should probably compare to 3 year government yields for true Rf):
















My second thought was something along the lines of:

"Ok, these yield about 150 bps over 3 year GICs (I'll call these Rf), but I have to pay par.  Am I getting compensated enough for risk paying par?"

FYI, these bonds are all gone from my broker's database, so I'm assuming that investors must have bought all 732,000 on offer.

The globe ran an (overly simplistic) article on these bonds last week , link here:

(In case the link doesn't work, here is the article)

"Wednesday, October 7, 2015
ROB CARRICK
Inside the Market
Wild times in the markets are creating opportunities in bonds for yield-starved investors.
You probably know all about how yields on dividend stocks have been rising with the recent market pullback. Less understood is the fact that there's now a decent selection of bonds issued by companies with middling credit ratings that yield in excess of 3 per cent.
Some examples: Home Trust Co.'s 3.4-per-cent bonds maturing Dec. 10, 2018, had a yield of 3.3 per cent at the end of September. Cominar REIT 3.6 per cent bonds coming due June 21, 2019, had a yield of 3.5 per cent. Ford Credit Canada 2.9-per-cent bonds maturing Sept. 16, 2020, had a yield of 3 per cent. To put these numbers in context, five-year Government of Canada bonds were yielding about 0.8 per cent, five-year bank GICs were yielding about 1.5 per cent in late September and alternative banks were offering 2.5 per cent at best.
There's more risk in corporate bonds than GICs backed by Canada Deposit Insurance Corp.
We're talking here about the risk of a default on interest or capital repayment at maturity, and of price volatility ahead of maturity. You can see this risk level reflected in the fact that the aforementioned bonds, and others like them, are rated BBB (high) and BBB (low). That puts them on the lower rung of what qualifies as an investment grade bond, which means suitable for use by pension funds.
Ways to minimize risk with bonds such as these include keeping your term as short as possible and familiarizing yourself with the challenges faced by the companies that issued them.
Diversification is also important - consider adding several different names to a portfolio that also contains GICs or government and blue chip corporate bonds."

I think the key takeaway here is, why were Home Trust Co.'s bonds rated BBB by Fitch, and BBBh per TD's database (not sure who they use for ratings), when bonds of equal duration (but with a higher credit score) yielded about the equivalent to 3 year GIC's?

A bit of background.  The company has been a subject of hotly contested debate over the last year over loan practices.  The globe profiled short seller Mark Cohodes during July 2015 on his Home Capital short call, link here.

If anyone has the chance to read up on Cohodes, he's brilliant (he's also pretty funny on twitter), but strong caveat, he doesn't sugar coat anything.  I have the utmost respect for participants like Cohodes who basically peel back the layers on situations that mainstream money managers don't. Many a mainstream manager has stayed long Home Capital throughout the ongoing debacle. Who's going to end up being right when all is said and done is anyone's guess.

Cohodes' thesis is that Home Capital, Canada's largest alt/subprime lender, is basically a house of cards. His thesis seems to have been helped along by lower loan originations earlier this year coupled with Home Capital severing ties over fraudulent external broker mortgage originations shortly thereafter. Personally, I'm not smart enough to pass judgement.  I don't know enough about the business, and there seem to be plenty of exceptionally smart professional money managers who think very highly of management.

What I can do though, in my capacity as small fish who does not need to buy anything, is to attempt to reframe and stress test the company's balance sheet in the event that Cohodes is right.  It's not that I think he's right.  But the question of "what if" he's right is a very poignant question.


First, here are two excerpts from the Q2 2015 F/S:

Loans:

















and, liquidity:





















I recall reading that in analyzing a bank, a bank's liabilities are it's assets and it's assets are it's liabilities.  Makes intuitive sense in the context of HCG.  The company takes deposits in via Home Trust and makes loans.  As at June 30, 2015, they had deposits on hand of $14.97B (+ debt not shown, $300M of which are our subject bonds above), and they loaned $23B.

What's interesting about the loan composition is that more than 50% of loans are non-securitized single family residential loans.  Here's a graphic on composition from the report:



















And here's further disclosure from the MD&A regarding insured vs. uninsured mortgages:



















The difference between insured and uninsured?  CMHC.  From the annual report:


"The traditional single-family residential portfolio is the Company's "Classic" mortgage portfolio which consists of mortgages with loan-to-value ratios of 80% or less, serving selected segments of the Canadian financial services marketplace that are not the focus of the major financial institutions.  These mortgages are funded by the Company's deposit products."


I did the following quick and dirty analysis when I initially looked at these bonds, and I'm probably oversimplifying here, but my take at the time was, should there ever be a run on Home Trust bank, where is my margin of safety if I were to pay par and chase yield today?

PS, I know, this is most likely a huge "tail" event in most investors' minds (a lot of whom are much smarter than me), but it doesn't mean that it cannot happen, it just means that investors expect that it's "unlikely" to happen.

First, here's the balance sheet on the presumption that the sky is not falling, i.e., 100% of uninsured alt mortgages are collectible.


AssetsAs reportedAdjustmentRestated
Cash + securities for sale1,365100%1,365
Loans for sale21100%21
Loans:
Securitized2,814100%2,814
Non-securitized15,147100%15,147
Allowance-35100%-35
Total loans17,92617,926
Other:
Restricted assets733100%733
Derivatives other3510%0Derivatives, other
G/W1200%0
Total LT4710
Total Assets19,78319,311
Liabilities
Deposits on demand1,436100%1,436
Fixed date deposits13,531100%13,531
Senior debt152100%152
MBS366100%366
Canada Mtg Bond3,145100%3,145
Other liab's3510%0Derivatives, other
Total liab's18,98018,629
Liquidation value tomorrow, assuming all loans are 100% collectible1.04


Next, here's the balance sheet on the presumption that 75% of uninsured alt mortgages are collectible.

AssetsAs reportedAdjustmentRestated
Cash + securities for sale1,365100%1,365
Loans for sale21100%21
Loans:
Securitized2,814100%2,814
Non-securitized15,14775%11,360
Allowance-35100%-35
Total loans17,92614,139
Other:
Restricted assets733100%733
Derivatives other3510%0Derivatives, other
G/W1200%0
Total LT4710
Total Assets19,78315,525
Liabilities
Deposits on demand1,436100%1,436
Fixed date deposits13,531100%13,531
Senior debt152100%152
MBS366100%366
Canada Mtg Bond3,145100%3,145
Other liab's3510%0Derivatives, other
Total liab's18,98018,629
Liquidation value tomorrow, assuming 75% of uninsured loans are collectible0.83


Finally, here's the balance sheet on the presumption that 50% of uninsured alt mortgages are collectible:

AssetsAs reportedAdjustmentRestated
Cash + securities for sale1,365100%1,365
Loans for sale21100%21
Loans:
Securitized2,814100%2,814
Non-securitized15,14750%7,573
Allowance-35100%-35
Total loans17,92610,352
Other:
Restricted assets733100%733
Derivatives other3510%0Derivatives, other
G/W1200%0
Total LT4710
Total Assets19,78311,738
Liabilities
Deposits on demand1,436100%1,436
Fixed date deposits13,531100%13,531
Senior debt152100%152
MBS366100%366
Canada Mtg Bond3,145100%3,145
Other liab's3510%0Derivatives, other
Total liab's18,98018,629
Liquidation value tomorrow, assuming 50% of uninsured loans are collectible0.63


Let's assume for a moment that I look at whether I should pay par for the bonds as a probability distribution with three outcomes based on the uninsured mortgages, 1) the bonds are covered, 2) the bonds are .83 x covered, and 3) the bonds are .63 x covered.

If I assign tail risk of 5%, I get .95 x 1.04 + .83 x .025 +.63 x .025 = 1.02, pay par.

But what if tail risk is greater than 5%?  Certainly, Cohodes thinks it must be.

If I assign tail risk of 10%, I get .9 x 1.04 +.83 x.05 + .63 x .05 = $1.  I still pay par.

At what level of tail risk do I not want to pay par?  Around 16%.  This gives me 1.04 x .839 + .83 x .0806 + .63 x .0806 = .99

What does the options market think (and is this at all relevant)?

Here's HCG (I looked at the April 32 straddle):





















The straddle is pricing in a 26% move (either direction).  Pretty hefty premium and reflective of uncertainty.


For comparison's sake, here's XFN (S&P financials index) March $30 ATM straddle:























The straddle is pricing in an 8% move.  Obviously HCG individually is riskier than all the constituents comprising the financials index, but 3 x as risky?

This is mostly nonsensical (and inconclusive), but I do think that the most remote sounding probabilities need to be considered here as tail risk always seem to be one of those obscure concepts that only gets discussed after the unthinkable happens.

Closing comments:

I didn't buy the bonds.