The strategy was launched on March 13. It has been tradable as a certificate since June 9. Today, it’s up 18.3 percent, and that number doesn’t tell the whole story.
📊 Where the strategy stands today
Up 18.3 percent since launch, up 7.5 percent over the last three months, and up 1.8 percent over the last month. The largest intraday loss was 14.6 percent. A total of 4,259 euros is invested across 44 watchlist entries. One number is almost never included in such overviews: the strategy’s highest level was 126 points; it currently stands at 118.3. So the strategy is a good six percent away from its own high. The 18.3 percent gain mentioned above isn’t a record high, but rather a recovery. What sounds like a quiet half-year actually had a pretty rough summer.
📉 July
In July, I realized for the first time what the difference is between a backtest and real money. The second rebalancing took place on July 1. After that, the portfolio went down. I track how the purchased portfolio performs against all 843 stocks in my universe, equally weighted. After four weeks, the portfolio was down 6.81 percent. The benchmark stood at plus 2.13. A lag of nearly nine points in four weeks.
That was the momentum crash. What had been performing well before suffered the sharpest decline. Momentum is one of my six factors. A portfolio selected based on this factor is bound to underperform an equally weighted benchmark during such a phase. In backtesting, I’ve seen such stretches multiple times. Still, reading about them is quite different from watching them unfold every morning for four weeks.
I did nothing. Not because I’m disciplined, but because the rule doesn’t call for any intervention. Nothing happens between two scheduled reviews, no matter what’s in the portfolio. After eight weeks, the shortfall had shrunk to 5.5 points. The recovery came when the market turned. If I had sold at the end of July, the loss would have been locked in, and the recovery would have passed me by.
💶 The portfolio after six months
Of the 30 positions, 14 are still from the first day in March. Seven were added in April, and nine in July. Two rebalancings, and nearly half remains unchanged. By weight, about 41 percent is in financials, just under 30 percent in semiconductors and memory, and just under 13 percent in mining stocks. I didn’t set any of these limits myself. The model doesn’t target specific sectors. They’re included because they currently combine favorable valuations with strong momentum. In July, the portfolio rotated out of chip stocks—which had become expensive—into European banks. For the chips, earnings were no longer keeping pace with the stock price; for the banks, it was the other way around. Bankinter $BKT (+1.93%) , CaixaBank $CABK (+0.76%) , BPER $BPE (+1.13%) , Monte dei Paschi $BMPS (+3.32%) and Swedbank $SWED A (+0.61%) are now all side by side in my portfolio. I never had a strong opinion on European banks.
🔁 Let the winners run
Micron $MU (-0.14%) is the perfect example of how this strategy works. I bought it with a 3.3 percent weighting; today it’s at 6.6 percent. I don’t rebalance between trading sessions. Winners stay big; losers get smaller. Compared to a strategy that resets every position to the same weight, this approach yielded 1.86 points last quarter. One quarter isn’t much, but the trend aligns with what the backtest shows.
🔄 On October 1
In the third rebalancing, about half the portfolio changes—15 stocks out and 15 in. Significantly more than last time. The mining stocks are almost entirely out—namely, Antofagasta $ANTO (+2.02%) , Coeur $CDE (+1.14%) , Endeavour $EDV (+0.24%) and Newmont $NEM (+0.6%) . Gold and copper have performed strongly since the spring, and that’s exactly what’s making the stocks in the model expensive. The main additions are refiners and semiconductor companies. I’ll list the exact stocks after the rebalancing. I find it more interesting to see what’s happening with the banks. In July, the model rotated out of expensive chip stocks into European banks. Now it’s rotating back into some of them—Bankinter $BKT (+1.93%) , Swedbank $SWED A (+0.61%) and UniCredit $UCG (+1.8%) are on the sell list. Not because they’ve underperformed—all three are in the black. They’ve risen and thus become more expensive, and that’s enough for the model. This is where a system like this seems least intuitive. It doesn’t sell because something is performing poorly. It sells because something is no longer cheap.
💡 What Six Months Don’t Tell You
Two rebalancings are two data points. The third comes on October 1—only then does a trend even begin. The backtest over six and a half years tells one thing; six months of live trading tell another—and I don’t know today just how far apart those two will diverge.
What I’ve learned after half a year is more valuable than any return figure. A rule set in advance takes the decision out of your hands at the very moment when you’d be least capable of making it.
❓ Did you stick with your strategy in July, or did you rebalance? And do you work with fixed rules, or do you decide on a case-by-case basis?
👇 Let us know in the comments!
➡️ Every month, you’ll find the full Wikifolio update here, along with research from the model in between.
🗞️ Newsletter: codeandcapitalquant.beehiiv.com
📈 Wikifolio: https://www.wikifolio.com/de/de/w/wf0gquant6
Not investment advice; our own research. Past performance is not indicative of future results.
