Investing

Trading idea- Based on Graham (TSX)

  • Last Updated:
  • Mar 26th, 2024 10:50 pm
Tags:
None
Deal Fanatic
User avatar
Jul 19, 2003
8137 posts
770 upvotes
rodbarc wrote: Do you mean outside of this model? If so, yes, and I'll have an update about it on an upcoming post, besides a separate post for my DGI annual review.


Rod
great! looking forward to it
hi!
Deal Fanatic
User avatar
Jun 19, 2009
6135 posts
1981 upvotes
Scarborough
Sucks for SIS, thought that one would come back.

I'll have to do some DD on CFX, but interesting to see a new stock being bought instead of one that was purchased already
Deal Fanatic
User avatar
Jun 19, 2009
6135 posts
1981 upvotes
Scarborough
Luckily(?) not much action recently. Thanks for the updates. I always find myself looking for updates Sunday afternoon if there are any changes to be made for Monday
Member
Jul 4, 2012
238 posts
51 upvotes
I'm trying to make sense of the numbers. The model has been running for slightly over 3 years, has an annual turnover of 104.72% and total trading costs of 2634.77$.

1- Are trading costs the sum of 4.95$ trades? If so, why does 2634.77$ / 4.95$ = 532.27 total transactions or roughly 179 transactions a year?
2- If the models holds up to 10 positions and has an annual turnover of 104.72%, it means it generated roughly 10 buys and 10 sells per year: (10*4.95$)+(10*4.95)= 99$/year?

I'm sure I am seeing it wrong, someone please help me with my math :)

I'm trading on Questrade. I'm trying to figure out if I should pay 4.95$/transaction or subscribe to a different pricing plan (buy and sell for 1 cent per share for 89.95$+tx/month). I could bill the 89.95$+tx/month to my margin account enabling me to pay less per trade in my RRSP/TFSA account.
Deal Fanatic
User avatar
Dec 14, 2010
7113 posts
9300 upvotes
Salomon1260 wrote: I'm trying to make sense of the numbers. The model has been running for slightly over 3 years, has an annual turnover of 104.72% and total trading costs of 2634.77$.

1- Are trading costs the sum of 4.95$ trades? If so, why does 2634.77$ / 4.95$ = 532.27 total transactions or roughly 179 transactions a year?
2- If the models holds up to 10 positions and has an annual turnover of 104.72%, it means it generated roughly 10 buys and 10 sells per year: (10*4.95$)+(10*4.95)= 99$/year?

I'm sure I am seeing it wrong, someone please help me with my math :)

I'm trading on Questrade. I'm trying to figure out if I should pay 4.95$/transaction or subscribe to a different pricing plan (buy and sell for 1 cent per share for 89.95$+tx/month). I could bill the 89.95$+tx/month to my margin account enabling me to pay less per trade in my RRSP/TFSA account.
The commission cost is $4.95 / transaction. The model includes slippage as part of the transaction cost, so the commissions look a lot higher than it was actually is. The model buys a stock at ask price and sells at bid price. The slippage (difference between bid and ask) is added as part of the transaction cost, to simulate the worst case scenario when buying stocks that are not very liquid. The end-result is that real-life portfolio for those following these models should reflect the same or better performance than the model, but not worse.


Rod
Build a comprehensive portfolio based on Investing and Trading strategies. Check out these threads and join the discussion:
Investing strategy based on dividend growth

Trading strategy based on Graham principles.
Sr. Member
Jun 3, 2012
645 posts
260 upvotes
Scarborough
Do you primarily do this for fun or to try and beat the market? If the latter goal is your main objective, wouldn't it be easier for you to just track the S&P 500 for instance? Since you started this initiative, its grown by nearly 40% (including dividend reinvestments) compared with your 28% (plus the amount of time it takes you to research and trade). Nonetheless, I appreciate following your website which I find to be very informative and helpful.
Deal Fanatic
User avatar
Dec 14, 2010
7113 posts
9300 upvotes
airmax95 wrote: Do you primarily do this for fun or to try and beat the market? If the latter goal is your main objective, wouldn't it be easier for you to just track the S&P 500 for instance? Since you started this initiative, its grown by nearly 40% (including dividend reinvestments) compared with your 28% (plus the amount of time it takes you to research and trade). Nonetheless, I appreciate following your website which I find to be very informative and helpful.
Both. The point of this thread is to show that quality and valuation drives superior return overtime. To beat the market, you need to address the pitfalls of indexing. That includes buying at a fair valuation (instead of paying market price for everything), buying companies that will keep growing earnings (instead of buying in the mix companies with signs of distress) and reducing or eliminating MER. The universe used for this model is TSX, so it's only fair to compare with XIC.TO, which is the ETF for TSX index. The Canadian market is very different than the US market. The model continues to outperform the Canadian index (apples to apples) because the index is poorly diversified and it market capped weight. Other models on the website offer different drivers for both US and Canadian markets, and the US models are a better comparison to SP500. The US models have outperformed SP500 as well, performance is updated weekly on the website.

Having said, a comparison to index benchmark is invalid in my opinion, because you aren't considering risk-adjusted returns. The model is attempting to manage drawdowns whereas the index fund doesn't. Also, not every model shares the same goals. A model focused on growth of income with an universe of stable, low beta company, will many times underperform the market by design, because the drivers of performance are very different. However, the rules regarding risk-adjusted return should help to outperform the index in the long run, simply because the index doesn't attempt to minimize drawdown. The index is meant for investing for long term, while these models are meant to invest for the short term / trading, so again, not a very valid comparison. Most people simply compare the performance of buying an ETF like SPY because it's easy and simple to buy it, and they're simply looking at performance - not at risk, volatility, or drivers behind that performance.

The advantage of holding individual stocks is that the better performing stocks can be harvested (sell high) rather than having to sell a piece of everything (selling low) - besides buying them with a higher margin of safety. Following an algo trading model is straightforard, researching takes no time, since the computer does the hard work, I just had to spend the time once coding it. The ideas are decades old, being based on Graham principles, as a simple example. This is simply the automation of one of many sound financial ideas. And trading needs 2 extra luxuries when compared to investing, which is why it's so hard and so many people fail at it: reducing drawdown and locking profits for short term.

The financial theory is just part of the puzzle. Discipline to stick with the plan and separating emotions is the other part that is hard to implement, and it's more of a soft skill (temperament) than hard skill.

There are ALWAYS going to be market inefficiencies. The Efficient Market Hypothesis disregards all the investors who are making short-term and long-term trades for reasons that have little to do with company performance. Thousands of traders are buying and selling based on minute movements in prices. Some investors are wildly overreacting to market news while others have decided to keep buying stock in a company no matter what happens. In other words, there’s nothing efficient about the market. To keep one step ahead, to find the inefficiencies and take advantage of them, takes some work. But it's worth it. Using a platform and computing power that can process more information than most investors can gives you an edge, even though we have access to the same public information. That and the ability to keep discipline, which is a lot easier following a mechanical model that has no emotions when reporting what to buy or sell.


Rod
Build a comprehensive portfolio based on Investing and Trading strategies. Check out these threads and join the discussion:
Investing strategy based on dividend growth

Trading strategy based on Graham principles.
Member
Jul 4, 2012
238 posts
51 upvotes
rodbarc wrote: The commission cost is $4.95 / transaction. The model includes slippage as part of the transaction cost, so the commissions look a lot higher than it was actually is. The model buys a stock at ask price and sells at bid price. The slippage (difference between bid and ask) is added as part of the transaction cost, to simulate the worst case scenario when buying stocks that are not very liquid. The end-result is that real-life portfolio for those following these models should reflect the same or better performance than the model, but not worse.


Rod
Thanks for the answer. I'm guessing the annual turnover percentage in your premium models don't include weight distribution changes which would add to the total costs without changing the turnover % (further explaining the gap between costs and turnover).

Following a weekly rebalance and assuming position changes, when does the model buy/sell? Next business day? At what price? What if the price gaps up/down?

What is the best strategy following a rebalance to make the purchases? Would consistence be key as it would even out over time?
Deal Fanatic
User avatar
Dec 14, 2010
7113 posts
9300 upvotes
Salomon1260 wrote: Thanks for the answer. I'm guessing the annual turnover percentage in your premium models don't include weight distribution changes which would add to the total costs without changing the turnover % (further explaining the gap between costs and turnover).

Following a weekly rebalance and assuming position changes, when does the model buy/sell? Next business day? At what price? What if the price gaps up/down?

What is the best strategy following a rebalance to make the purchases? Would consistence be key as it would even out over time?
Rebalancing (keeping weights accordingly) and portfolio reconstitution (revisiting buy / sell rules) is done weekly. Rebalancing on the model is only registered when it's worth above $1,000, then it sells / buys to adjust the weight accordingly, otherwise the model doesn't rebalance (it waits the weight to be disconnected by $1,000 to take action).

The signals are placed on the weekend, but sometimes Compustat (or CapitalIQ) is late with their feeding, so their commitment is to try to get the data by Monday before the market opens. The buy / sell instructions are considering Friday's close price, but later the model adjusts to consider the average price of the following business day (typically Monday), to be as realistic as possible regarding someone buying anytime during the day. So by Tuesday, the model updates the graph / cost with the average of the day on Monday. What matters is that when there's a signal to sell on Monday, then 100% will be sold at Monday's price, and the proceeds will be used to buy what is in the buy rule, at Monday's price, so the fact that signals are sent earlier doesn't change the methodology. Prices and performance is adjusted later accordingly.

This model uses an equal weight balance distribution, so I typically wait until I'm off by $1,000 to distribute. If your portfolio is small, I'd say to rebalance between 5% and 10%. Consistence is important, as it prevents rebalancing too often (hurting on commissions) or rebalancing too little (which has the risk to allow a single position to have too much weight).

Please let me know if you have any questions.


Rod
Build a comprehensive portfolio based on Investing and Trading strategies. Check out these threads and join the discussion:
Investing strategy based on dividend growth

Trading strategy based on Graham principles.
Member
Jul 4, 2012
238 posts
51 upvotes
Thanks Rod. I love the idea of taking human emotions out of the equation. What material about algorithmic trading would you recommend reading. Any blog/book in particular? I am currently reading F. Piars's stuff.
Deal Fanatic
User avatar
Dec 14, 2010
7113 posts
9300 upvotes
Salomon1260 wrote: Thanks Rod. I love the idea of taking human emotions out of the equation. What material about algorithmic trading would you recommend reading. Any blog/book in particular? I am currently reading F. Piars's stuff.
Fred Piard is great, and he's also a member of portfolio123, which is the platform that I use to write my algorithms.

My blog has info about other models and the resources section lists the books and platforms that I suggest reading to learn more about it. You might like this post, about my inspiration to algo models. I'm working on a post about avoiding curve fitting when developing a model, I'll post it soon.


Rod
Build a comprehensive portfolio based on Investing and Trading strategies. Check out these threads and join the discussion:
Investing strategy based on dividend growth

Trading strategy based on Graham principles.

Top

Thread Information

There is currently 1 user viewing this thread. (0 members and 1 guest)