I was hoping some P123 members (and staff) could tell some stories about how they’ve changed their investing style in reaction to new information.
I’m writing a blog post about updating your views effectively (the sixth in my series on How to Be a Great Investor, based on Michael Mauboussin’s white paper “Reflections on the Ten Attributes of Great Investors”).
Mauboussin writes, “great investors do two things that most of us do not. They seek information or views that are different than their own and they update their beliefs when the evidence suggests they should. Neither task is easy. . . . The best investors among us recognize that the world changes constantly and that all of the views that we hold are tenuous. They actively seek varied points of view and update their beliefs as new information dictates. The consequence of updated views can be action: changing a portfolio stance or weightings within a portfolio.”
I’d love to be able to share some concrete examples on our blog and on Seeking Alpha, if you can share them without spilling your secret sauce. I’ll keep all stories anonymous. If you want to share a story with me without posting it on this forum, please e-mail it to me at yuval[at]portfolio123.com.
I will clearly be using methods beyond sims and rank performance after getting a clear idea of how the Designer Models are doing.
To the extent that I can, I should say.
Georg, I am sure, is familiar with ARIMA, for example. But I will use this or some of the other methods Georg uses to look at a model that changes the allocation for Sector ETFs. The difference between ARIMA and the methods I am sure Georg uses is not large enough to go into here. It is something that can be done with the P123 downloads.
Georg may have a model based on this idea but I always like to do it on my own. Georg can speak to this better than I can at this point.
Whatever you believe about RenTec’s influence on the market it may not be as great with ETFs, maybe.
Jim, I’m looking for stories of how users have changed their investing styles and how that has worked out for them. I know that you’ve been working on advanced statistical testing and machine-learning techniques for a while. Have you actually changed your portfolio because of this or is it something you’re still planning to do? If you’ve put your new techniques into practice, how has it worked out for you? Or is it too soon to tell? Thanks!
I have made changes with real money. I think a bunch of talk on my part (or anyones) about paper trading can get quickly in to BS as Marc might rightly call it. Let me stick to real money, if I may.
The one thing I have a clear idea about now is the Designer Models. So, yea I have changed. I have reduced the amount allocated to my ports as I do not think I can do much better than the Designers on average.
Specifically, much of my portfolio is now invested in the Minimum Volatility Portfolio (by MPT) for sector ETFs (SPDR). I mentioned above that I am looking for other ways to move my allocation for these ETFs but I am using the Minimum Volatility Portfolio for now. I have posted the full code for this in Python for anyone interested (different thread).
It is working well in the present market—with less volatility. This has a low correlation to my ports so quite a bit less volatility. Also it MAY do well if there is a recession.
May not be the most interesting post that you will get in response. Nor do I recommend that anyone else do this without consulting with their financial advisor. But that is what I am doing.
A backtest of the minimum volatility portfolio using books does better than SPY with less volatility and a better drawdown in 2008 but let me be the first to say cherry-picked backtests mean nothing.
I definitely have changed my investment style to a emphasize lower turnover strategies over time. I think like many P123 users I was wowed by the alpha of these amazing high turnover simulations, and thought I could minimize slippage enough to emulate them and retain enough alpha cushion to still easily beat my benchmarks. Now I’m more inclined to believe any strategy that depends on a) You having the informational advantage to get the buy/sell signal before everyone else and b) getting in and out of the position before the rest of the crowd is a bit of a fool’s errand. Plus, with experience, I’ve just come to realize I’m not a very good trader by temperament. Even using some limit orders and IB algos, I get impatient and I chase price movements. Plus, I can’t speak for anyone else, but I get “trader fatigue” when I’m just following ticker data all day. I don’t know how day traders do it. I’d rather work a regular job, to be honest.
So now I’ve minimized, if not outright eliminated, volume and price momentum out of my ranks. I still for the most part have weekly rebalances to catch anything like secondary offerings, but they might as well be quarterly because my rankings stay fairly static until the next earnings reports come out. One of the very first things I check on any simulation is the Avg Days Held and Avg Return of any holding.
stopped subscribing to designer models. Very hard to figure out the factors, buy and sell rules and if the designer was using timing and if the timing will work in the future.
Moved to lower turnover and longer holding periods. (Probably gave us some return and trading costs :))
I’ve been a member for a long time and in the early days of price/sales heavy alpha models I went very heavy on my trades. I was always trading with some level of leverage. Up until 2008 it actually worked out very well and I made a lot of money. During the 2008 crash I pulled everything out around a 40% loss and stayed out. I let emotions get to me and missed the resurgence of the market. I kept some of my models going in auto just to see what would could have been and sure enough I may have left millions on the table. However that’s life and it was a choice I made and don’t regret. Those same models would have performed poorly today and I would have given it all back in a slow bleed.
I have learned to be more robust with backtests, rolling, time periods…
I have learned to treat tickers for the companies they represent and not just as a more is better statistical tool. P123 is designed to be comparative and that is no different than me looking for the best TV at the best price. I compare TV’s and prices, I don’t compare TV’s to nail clippers.
I have learned to collect a lot of data before deciding what to use and what to throw away. Better backtests don’t necessarily make better systems.
I have learned robustness starts far before the buy and sell sims. Ranking design can and IMO should include time frame changes as well as performance evaluation. Histograms are great for giving you a mean result but they don’t convey how the mean was established.
I have learned that timing with P123 resolution is ok at preservation of capital but dangerous when trying to use it to create alpha. Others are likely much better at timing than I am and maybe they have more faith in it.
I have learned that we all have an idea of what P123 means to us and I love the diversity of ideas that float the forums.
P123 has come a long way since I joined and I hope it gains greater traction in the coming years. More money for P123 should mean more resources to help create tools that I can use in my arsenal against Mr Market. In the early days I didn’t tell anyone about P123, I thought keeping it a secret helped my edge. Now I realize that was a mistake and I tell anyone that is interested in investing about this site because I want to see it grow and evolve.
As far as trading goes. I do trade using this platform for US stocks but I really wish there was a way to use it for Canadian trades. I don’t see that changing anytime soon.
Barn thanks for the confirmation that the old ways don’t work as well and we will constantly need new tools or techniques to beat the market. I have been hear a long time and started with stocks paused and started to research what works. You can beat 95% of managed money and stock systems by indexing. Starting with SPY or QQQ you have a system that beats managed money and I would say most designer models. Starting with that Hypothesis gives you an advantage that you cannot ignore unless you don’t like the 95% odds. The next thing I did was look for tools that help me manage the volatility to increase the sharp ratio and hopefully switch me into bonds and gold when the next crash happens. I have found a few tools and the next couple of years will tell me if they work or not. So far they are working but it’s very short OS. My goal is to beat the market by a few % and that will put me in the top 99% of investors over 10 years. Stay tuned I will let you know how it goes.
Based on my Hypothesis P123 should focus more effort on ETF’s. How do you optimize an ETF strategy to produce maximum return with minimum risk. Not easy but doable. Most pension funds do this with very simple Stock, bond models and are charging for it I would rather just do it myself and not pay someone.
Replace a couple of words in the Mauboussin quote and the result means even more for me. Change “investors” to “thinkers” or “observers” and change “portfolio” to “life” or something similar. I strive to maintain that flexibility in everything, but we are wired to ignore evidence that goes against our current understandings and prejudices. End of philosophizing.
As an individual retail investor, how have I changed my investing style in reaction to new information? I would include new tools as a very important asset that can allow better use of existing data. Skipping over the massive changes that simply subscribing to Portfolio123 and using its tools and data made in my style and focusing on recent new information and tools within Portfolio123:
The change to commission-free online orders allows me to speed up reconstitution and rebalancing. I was focused on lower turnover models but am now allowing more. This recent change could have a dramatic impact on my style going forward.
“Formula Weight” position sizing has allowed emphasis by stock or ETF more easily and I believe it can improve my results. I usually, but not always, find benefit in using rank, rank position, or short interest in sizing the amount to invest in each holding.
Rolling tests help check a model for timed luck sensitivity and I consider them important.
New information includes the OOS results of portfolios, and here is where I apply frequent effort to improve my models. Since I started using Portfolio123 in 2013, my personal port performances have been decent but not spectacular. How much of my results are due to my choices is difficult to determine, especially since the overall market has been up most of the time. But I do frequently retest ranking, universe, and port rules using recent data and periodically make adjustments to my ports. Since I have limits on the resources I can use due to my membership level (which have improved for my old level due to the membership structure changes, thank you very much), I cannot measure the OOS impact of my propensity to make changes over the life of my existing stock ports.
Concerned with the longevity of this bull market and a paucity of proof that stock ports have a consistent edge, I am focusing more on the use of ETF models and trying to reduce drawdowns while getting a decent return. This includes some use of market timing in either ranking (to change the impact of factors based on market movements) or buy/sell rules (to change the group from which to choose) or both. I focus mostly on environmental factors rather than technical market movements for market timing, and my focus with timing is to minimize drawdowns rather than improve alpha but hope for both.
Specific information whose benefit has changed for me in the last year or two: I use a little industrial momentum in rankings where I used to not use any technical measures. I use short interest both as a ranking factor and for position weighting. I no longer use book value or price ratios in any way. I have not yet found value in removing industries or sectors from consideration in the ports I use.
I consider the give and take of this forum and the linked information and analyses provided by both staff and members to be of very significant value and thank you all. Aside from the entertainment value, there is frequently juicy meat! Allowing for these exchanges and participating goes to the heart of Mauboussin’s point. Thank you for allowing me to participate.
I may be totally off base here, but i think the key to successful investing in the near future will be:
Understanding what changes and what remains relatively constant. After all if the value of a stock is not the discounted cash flow of all future income streams, what is it? Does the rise of algorithmic investing negate the behavioral biases that could be exploited? I feel anything that can be arbitraged away using computers is suspect.
having a strategy different than the market is pursuing. If the market is moving towards shorter hold times, maybe we need to extend our hold times
Interpret financial statements differently in light of current asset light business models, stock buybacks, etc.
Learn to love lower returns?
In short, we have to be both different, and right. Tall order
Previously In my ranking systems I gave stocks that have lower market capitalizations a higher ranking. Lately I have reduced the market cap factor to increase the chance of buying mid-cap and larger cap stocks. I believe with the increased number of Investors purchasing Large cap Index ETF’s that the small cap premium may continue to somewhat underperform even though we are seeing somewhat of an improvement in small cap and value stock p/e ratio expansion. I believe that value and quality factors that are used in ranking systems will never go out of style. Stocks that are growing earnings and producing high ROI and trading at reasonable valuations with manageable debt levels will outperform in the long run. To be prudent I now try to limit the amount of micro and small cap stocks to a maximum of 20-30 % of my total portfolio value. Anyways I am sure some will disagree but that’s my 2 cents.
I’ve tried to do the following as I’ve learned the hard way. Fortunately I haven’t lost money and still made some, but I’ve lost opportunities. These are what I’ve learned:
Minimize Turnover (as slippage can kill you).
Use a broker with good price execution as poor execution will erode market gains even if it means not getting free trades.
Don’t rely on one strategy or factor. If you like a factor combine it with others which diversify. A factor or strategy can stop working for a period of time (or maybe it really never worked).
Strategies need to be causal not correlative. Also be careful of overfitting. I’m my mind a strategy that shows weakness at times is probably more realistic than one that always outperforms. Look at the designer models. Most of them start underperforming as soon as they launch…
Trade stocks with sufficient liquidity. Micro caps look great in simulations but can you really get in and out of trades to make those strategies work? My experience is one of frustration where you slip multiple percent to get in and out.
I read Marc’s writings on valuation principles. It is very good and eye opening. It really got me to be much more critical of backtesting. I think backtesting can still identify good strategies, but again overfitting can kill future returns.
I used to do limit orders during the day and tried to get good entry and exit points. It seemed like more often than not I would either overpay or undersell or end up adjusting limit orders to chase the price to update the position. Now I just do market on close orders. I get the end of day price which matches my sims and eliminates having to time during the day. Quite frankly I got trader fatigue from having to do limit trades. How do day traders do it…how stressful and frustrating? I still trade weekly but I set up trades on weekend and they just go through on Monday at close.
