My (Automated) Investment Journey

[KnowThePast]

Hi everyone,


Time for a new investment thread!


Those who followed my previous investment journey know that I always had a lean towards automated strategies. That interest eventually took me to start a new technology business, which I will now be using to invest my super portfolio.


I will start off with $50,000 and 4% position size. I have chosen a strategy based on Piotroski F score. This is a modified version of it, however, the underlying economic rationale remains the same.

The score is named after Stanford accounting professor, Joseph Piotroski. He showed that by analysing neglected stocks, as defined by having very low price to book ratios, and using a nine-point accounting-based ‘health’ score, you could significantly outperform the market.

Where my strategy differs from his is that I have included yearly price change as a ranked factor. This means that I’ll be investing in companies which have had the lowest price to book and highest yearly price change. In other words, I’m looking to invest in undervalued healthy companies which are trending up and have potentially turned the corner over the preceding year.

Companies I’ll be investing in must be classified as strong (i.e. 8 and above) which means I know that they are profitable, have increasing margins, they can pass an accounting trick test, and have an improving balance sheet. If a company I have invested in falls in to the ‘weak’ category, I’ll sell it.


Here are some screenshots of the setup:


I’ll invest in Australia only. I’ve excluded Foreign and OTC stocks, as well as Metals and Mining industry. I’ve set minimum Market Cap to $50m and minimum share price to $0.30. I’ve sorted companies using a combined score of Price to Book ratio and Yearly Price change.


Portfolio options - $50,000 start, 4% position size, $20 brokerage. I am not factoring in interest on undeployed cash at this point and I am assuming slippage costs of 0.5%. I will set Maximum Trades Per Step to 3, so that capital is deployed over time.



Very simple buy and sell criteria – buy when the score is over 8 and sell when it falls under 5.



I will sell losers after a year. I will also sell 80% of a holding if it has gone up in price 400%.

 

[KnowThePast]

An important factor in deciding on a strategy is how well it performed before. While past performance is no indication of future results, looking at backtest results of many strategies can help decide which one is more likely to outperform going forward. I am not married to any one strategy – I’ve decided to run with this one now, I will almost certainly switch to others in the future.

Here are the results, had someone used this strategy to invest over the last 14 years, compared to the index:



Risk/Return metrics:




Monthly performance (this strategy was less volatile than most):



In a few days, I’ll start investing into this strategy on a weekly basis. As always, I appreciate any feedback/challenges/questions/criticism to keep me sharp and honest.

 

[KnowThePast]

First trades in.

I am limiting investment to 3 trades per week to enter the market gradually. The following 3 stocks have been selected by the strategy, which I have now bought:

 

[kenny]

Thanks KTP for prompting me to revisit the value in incorporating Piotroski's F-score as a screener.
I found this article quite useful for those like me looking for both validation and refresher.

https://www.researchgate.net/publication/301600517_The_Piotroski_F_-score_evidence_from_Australia

The article notes the value in using the score in a Long/Short strategy (F-Score >8/<2) to improve results and how the strategy seems to perform better in small caps and companies not covered by analysts. Worth a read.

KTP, I'd appreciate your thoughts on;

• what do you see as limitations or weaknesses of the F-Score?
• did you consider adding another screen to cover the weaknesses of the F-Score?
• Have you noticed a lack of liquidity and/or unwanted volatility in the flagged shares due to limited free float or large holdings?

 

[Newt]

Interesting premise and thanks for sharing KnowthePast.
Can I ask what software and data feed you are using?

 

[jjbinks]

Interesting premise and thanks for sharing KnowthePast.
Can I ask what software and data feed you are using?

If i remember correctly KTP developed his own software and I think he also collated his own data but I may be wrong on this.

 

[KnowThePast]

Thanks KTP for prompting me to revisit the value in incorporating Piotroski's F-score as a screener.
I found this article quite useful for those like me looking for both validation and refresher.

https://www.researchgate.net/publication/301600517_The_Piotroski_F_-score_evidence_from_Australia

The article notes the value in using the score in a Long/Short strategy (F-Score >8/<2) to improve results and how the strategy seems to perform better in small caps and companies not covered by analysts. Worth a read.

Thanks for that - definitely a good read.

KTP, I'd appreciate your thoughts on;

• what do you see as limitations or weaknesses of the F-Score?
• did you consider adding another screen to cover the weaknesses of the F-Score?
• Have you noticed a lack of liquidity and/or unwanted volatility in the flagged shares due to limited free float or large holdings?

I don't think there will ever be a perfect measure of a company and if there was, it will eventually be exploited until there's no alpha left. I've added Price to Book and Yearly Price change sorters to it, idea is to use the combined score of those to sort the companies in a "better" way. Interestingly, this picks up mostly larger stocks, so smaller stock and liquidity do not seem like they will be ab issue going forward, for this strategy any way.

I don't think it's a matter of finding the perfect score or strategy that will always perform - it is about having tools and data, to be able to find what is most likely to work now to continue tuning what is likely to work now.

 

[KnowThePast]

If i remember correctly KTP developed his own software and I think he also collated his own data but I may be wrong on this.

Good memory, jjbinks!

I've now commercialised that software and it's available to use on a subscription basis - www.forwardcaster.com

Data is from CapitalIQ, we currently support strategy building and backtesting on Australia, New Zealand and NASDAQ exchanges, but will open up all other markets soon.

 

[exchanger]

Good memory, jjbinks!

I've now commercialised that software and it's available to use on a subscription basis - www.forwardcaster.com

Data is from CapitalIQ, we currently support strategy building and backtesting on Australia, New Zealand and NASDAQ exchanges, but will open up all other markets soon.

are you managing the money? how can we use this product?

 

[KnowThePast]

Hi exchanger,

We do not manage money. We provide the data and the tools.

You can build a strategy, like the one I'm using and set it to send you a list of recommended trades at regular intervals.

You can also just use it to backtest any ideas you have.

You can register and play around with it for a limited time for free, feel free to pm me if you have any questions.

Cheers

KTP

 

[KnowThePast]

The joys of starting an automated portfolio in the middle of a reporting season - a week after buying 3 stocks, 2 of them are on the way out.

An interesting observation around automated strategies is that their performance tends to worsen when someone tries to improve it by overriding some decisions. I will stay true to the full automation course and sell AGL (21.37) and SMR (1.155) for a profit of $93.20.

For the buys this week, I've added ANN, PAC and FID to the portfolio. Here's a screenshot of the current portfolio:

 

[BlindSquirrel]

... excuse my ignorance; did those two lines' Piotroski scores drop below 5?

 

[KnowThePast]

... excuse my ignorance; did those two lines' Piotroski scores drop below 5?

Yep.

 

[BlindSquirrel]

ta kindly.

 

[KnowThePast]

This week, added SFC and BXB to the portfolio.

 

[KnowThePast]

Another exciting week!

We've now integrated with Sharesight, those of you who use it for your portfolios can now link them directly to Forwardcaster. The integration is now in beta, will be officially launched in a few weeks.

As for this portfolio, I've added BRG, CUV and CLV to the portfolio, starting to fill up.

 

[KnowThePast]

I've been travelling for 2 weeks and haven't had a chance to update the portfolio.

All stocks still in, no sell recommendations from the strategy. Bought AGL, BLD and AUB. Current holdings below:

 

[brty]

Thanks for that - definitely a good read.



I don't think there will ever be a perfect measure of a company and if there was, it will eventually be exploited until there's no alpha left. I've added Price to Book and Yearly Price change sorters to it, idea is to use the combined score of those to sort the companies in a "better" way. Interestingly, this picks up mostly larger stocks, so smaller stock and liquidity do not seem like they will be ab issue going forward, for this strategy any way.

I don't think it's a matter of finding the perfect score or strategy that will always perform - it is about having tools and data, to be able to find what is most likely to work now to continue tuning what is likely to work now.

I've added Price to Book

Interesting how you mentioned price to book in your first post but only added it as a filter later. It is my understanding that the price/book ratio is an important aspect of Piotroski F-Score and only those in the lowest 10% decile should be included.

Of course with US stocks there are so many more, so the lowest 10% on a P/B ratio offers a lot more alternatives than we get on our market.
I first mentioned Piotroski on this forum over a decade ago, yet no-one seemed interested. Fascinating how over time people become more interested, despite being known about in the late 90's.
I use Piotroski over longer term holds, plus time my entry into them instead of buying straight away on what could prove to be a short term spike in the SP. 8s and 9s usually spike on the day of announcement.

 

[KnowThePast]

Hi brty,

I've used the P/B from the beginning, so far I've made no changes to this strategy. However, I don't use it as a filter, I order the stocks by P/B and Yearly Price Chance. This won't exclude high P/B stocks, but will buy the cheaper one first. The end result will be mostly low P/B stocks.

The P/B ordering is combined with Yearly Price change order - so the strategy will invest in stocks that have the best possible score out of those two, assuming this first past "Piotroski>=8" criteria.

You are quite right about many people not being aware of it, but it is certainly gaining in popularity. And the more people know and use it, the less useful it becomes.

On its own, it no longer shows the out performance that it used to - but still works great when refined with other filters. I'm set to milk it while it works, knowing that it's only a matter of time that it doesn't.

 

[KnowThePast]

Add SYD to the portfolio, 268 @ $7.31