The violent market moves last Thursday produced a dramatic change in the Geometric Balancing portfolio, which is why I rebalanced mid-week. Rebalancing mid-week is the one part of my strategy which isn’t 100% system based, although that will change going forward. I’ll explain why in this post, and also discuss why I rebalance when I do.
Mid-Week Rebalance
I decided last August to rebalance the strategy mid-week if one of the assets moves more than 10%.
I’ve been toying with the idea of creating a hybrid of the weekly strategy and the full strategy, one that would rebalance mid week if the portfolio moves enough to justify a change. The drawdown in the weekly strategy is higher than the full strategy because the environment can change dramatically over 5 days, leaving the portfolio out of proper balance for a short time. I decided in an effort to combat this lag, I will rebalance the weekly strategy mid-week if a component’s weight changes by more than 10%.
Is this a hard and fast rule? Not really. I’ve done it a few times, and I’ve always rebalanced when the strategy has moved way beyond 10%. I know from both theory and back tests, that more frequent rebalancing can help returns. But this has to be weighed against the increased trading costs, and requires more time to rebalance frequently.1 However, none of the back tests I’ve shown (or run) have an “out of balance rebalance” feature. This mid-week rebalance rule is based entirely on feel as of today.
Why I Rebalance On Friday
The strategy running at the top of the blog is constructed for weekly rebalancing. But it doesn’t matter which day you run this cycle over. Any day would work. The idea is pick the day, and then generally try and stick to it.
I chose Friday because:
- I’m usually less busy on Friday afternoons, so it’s easier to fit into my life
- I figure the most likely time for a large down gap is over the weekend or Monday. So I’d rather that gap hit my portfolio when it’s perfectly aligned and not slightly off, as it could be after a few days.
That’s it. Any day theoretically works.
Employ the Live Signal at or Near the Close
All the calculations I’ve built, and the back tests I’ve shown, are for closing data. Prices from the morning are incomplete. They are partial days, and the formulas are meant for full days. So I place trades near the close to be true to the formulas,.
Now would trading in the morning make that much of a difference? I don’t have a drop of evidence to support this view, but I don’t think so. I think most of the time when the market is calm it would make very, very little difference. In volatile markets it’s more likely to matter, but it may just be a crap shoot either way as to whether it’s good or bad. I certainly wish I had rebalanced in the morning last Thursday. Friday would have been the opposite.
But ultimately that’s not how I designed the system, so I’d prefer to stay away from morning trading if my schedule allows.
What About Trading the Next Day With Yesterday’s Signal?
This actually doesn’t degrade the returns as much as you would think. I’ve run back tests that took a signal and instead waited and “traded” it on the close of the following day. This drops the return by about a quarter of a percent per year, increasing volatility by about a tenth of a percent per year, and increasing maximum draw down by about 2%.
So not that bad. I believe the drawdown is most at risk with the extra day more than anything else.
Furthermore, I struggle to see why taking yesterday’s signal and rebalancing it in the morning instead of waiting until the end of the day would be a bad thing. Especially if you were going to use that signal at the end of the day anyway. Maybe I’m being silly with my “no morning rebalance” rule.
Improving Rebalance Frequency
One of my next research projects will explore how to reduce the number of rebalances. It’s not the actual rebalance that helps. It’s moving the portfolio back to the proper ratios that provides the benefit. If the portfolio is only slightly off, the rebalance doesn’t help much. If it’s way off, it helps more. Theoretically, the rebalancing should not be scheduled at all. It should occur based on crossing a threshold of being too far away from the optimum portfolio.
Condition Monitoring
In equipment maintenance this is called condition monitoring and predicative maintenance. In condition monitoring, you don’t perform maintenance on a pump because it’s the end of the year. You perform maintenance when the bearings get too hot, or the pump vibrates too much. You don’t replace a filter because one month has past, you replace it because the pressure drop across the filter indicates it’s now loaded with particles.
Under this philosophy upkeep on the equipment isn’t schedule based, it’s condition based. When its operating condition has crossed a threshold of being out of balance–out of quality–you act. But if it’s within tolerance you save the time and money, leave it alone, and continue to monitor its operation.2
Geometric Balancing will move towards a conditions based rebalanacing–quality-based rebalancing– not scheduled time-based rebalancing. I’m not there yet, but the ability for midweek rebalancing is my first step.
I would endorse anyone else to explore this concept on their own. Once I’ve fully fleshed out my own system and rules, I will of course discuss it further on the blog.
1-I keep planning on building an automatic trader for geometric balancing and yet I’ve done nothing other than start to plan it out. I’ve found I actually look forward to looking at the strategy’s numbers every Friday to learn why it’s moved. Some day I’ll automate it all, but for now it’s not too much of a hassle.
2-There usually is a top-end timeframe the maintenance still must adhere to. For example, you will monitor the filter every month for a year. If for some reason you don’t change it over that year you will go ahead and change it anyway at a year. I would do the same with my portfolio. I’d probably rebalance at a minimum every month even if the strategy stayed within tolerance levels the entire time.
Thank you for explaining your thoughts on rebalancing.
I like your analogy with condition monitoring but I think there a key difference from portfolio management. When monitoring equipment, you expect it to continuously degrade, i.e. a filter gets saturated, a bearing reaches end of life. These things are not self-correcting and you must take action to avoid a breakdown.
But portfolio allocations can, and often do self-correct as prices change. It’s a difficult thing to guess the ‘best’ time to override the system based on an instantaneous signal. Rebalancing could be event based, like the volatility spike last Thursday, but often that has us selling an asset when we’d be better off buying it.
You’ve previously discussed the advantages of having a cash component in the portfolio. Have you considered a rebalancing strategy where you incrementally deploy cash during a drawdown or volatility spike as a way to rebalance? In individual stocks this can be like catching falling knives, but unless you have some indication of a trend change in an index, bond fund, or gold it might indicate a buying opportunity.
Yes, portfolio’s can revert back. But in the case of Thursday, when the proper portfolio changed so much, we don’t want the market to cut our equity position in half for us. Either way though, the overall trend for a portfolio is toward concentration in a single asset. This is true in pure randomness, and true when one asset returns more than another.
My strategy will do a bit of what you said with Cash, but will do it with bonds and gold instead of cash. If one asset falls, and the portfolio still recommends the same strategy, then it will sell the other two and buy the one that fell. But, sometimes the strategy calls for less of the asset which fell, therefore, it wont purchase it when it falls in price.
I really like your blog – great insights how analyzing geometrical effects change what are otherwise intuitive (but wrong) decisions and can explain a lot of paradoxons.
One gripe that I have with employing your strategy, though, is that in Germany there are unfortunately no such things as tax sheltered accounts. When rebalancing, you are paying some 25% tax here on any capital gains (regardless of your holding period) – a hefty transaction cost. No matter how I calculate these capital gains tax for any rebalancing intervals less than annually, it always eats away any rebalancing advantages (and even annually it almost brings no benefit).
The best way I came up with is to do the rebalancing only via infusion of new capital (or in the retirement phase taking money out) – but obviously this totally screws the high-frequency approach. So for me it comes down to analysing much longer time frames, setting a more permament allocation and then try to balance the portfolio in this direction via new money.
If any reader has a different view on applying rebalancing in jurisdictions without tax sheltered accounts, I would be extremely interested in getting your thoughts…
My next post will be on taxes. How do you account for which shares are sold in Germany and do you have a choice?
As another German I would like to comment on the tax topic.
If you are customer of a german broker, the broker takes care of the tax for you (immediately subtracts the 25% after the transaction).
But I for instance use a foreign Broker. Like one in the Netherlands. Then I have to manually file taxes at the end of the year for capital gains / losses. Benefit: my buying power is not immediately decreased.
Losses are carried over indefinitely to the next tax year.
Hope that helps you BTM. Do you think the rebalancing is still feasible?
I assume you get the losses back when you file?
Your way sounds feasible, but taking out taxes at every profitable rebalance sounds like a challenge.
Hello .
I am a new reader and have never re-balanced before because l used to be an active trader rather than an investor .
Now that l have a ” permanent portfolio ” , l need to setup a re-balancing regime !
Best wishes , Rupert .
Thanks so much for spending the time to put this blog together. It’s fascinating and I can’t say I’ve seen anything else like it. As a fellow mechanical engineer I’ve really enjoyed some of the analogies you’ve used.
I’m fairly new here and haven’t spent much time trying to implement your strategy. One question I had regarding rebalancing though… When you rebalance do you buy fractional ETF shares or are you just buying individual shares and getting as close to the modeled allocations as possible?
I do not buy fractional shares. I just leave the left overs in cash and get close to fully invested.
Why not fractional shares?