Report Analytics USA #2

My two Wikifolios are now online for somewhat more than 6 months (the “Large” version was incepted on July 29, 2021, the “Small” version on August 23, 2021). I think this is a good time to take a closer look at the performance. The two Wikifolios are primarily a live test of the ideas that I analyzed in my master thesis. They aren’t very sophisticated at this point. But let’s see how the trading idea performed in a real-world environment with transaction costs and different types of fees. Spoiler: not as good as the hypothetical backtest. There are many issues to be improved, but I guess this doesn’t surprise anybody.

In this post, I focus particularly on implementation issues with Wikifolio. So it’s going to be a bit more technical. Feel free to use the following list as a navigator if you don’t have time for the entire post.

  • In the first part, I do a lot of calculations and explain them by means of the “Report Analytics USA Large” Wikifolio. Most importantly, I identify the “true” performance of the stock selection. We will see that fees and indirect trading costs make a substantial difference.
  • Without explaining it all again, I do the same analyses for the “Report Analytics USA Small” Wikifolio in the second part.
  • In the third and final part, I summarize the identified issues and provide an outlook.

Before we go into that, a few remarks upfront. The two portfolios are my first experience with Wikifolio, so I had no idea how good or bad this platform is. And to be honest, I haven’t thought too much about it, because Wikifolio is currently the only feasible way to implement this strategy anyway (I already discussed this here). However, I severely underestimated how strong costs affect the performance.

Please don’t get this too negative. Wikifolio provides a service and there is nothing wrong with charging fees for this. Nonetheless, I think we should estimate the true impact of those fees and check how they affect performance. For this reason, this post maybe interesting for everyone who is active on Wikifolio. Even if you are not necessarily interested in my website and the stuff I am doing here.

Review – Report Analytics USA Large

The cumulative performance of the “Report Analytics USA Large” Wikifolio currently stands at -1.73% since inception in late July 2021. Given the macro developments of the last weeks, I am happy that the Wikifolio hasn’t lost too much money in absolute terms. However, it clearly missed the goal of relative outperformance vs. the S&P 500 Index. To get a truly investable benchmark, I use an ETF from Lyxor as proxy for the S&P 500 Net Total Return Index in EUR. This is the most suitable comparison because it considers reinvested dividends and, like the Wikifolio, is also denominated in EUR without any currency hedges.

The chart below shows the performance of the Wikifolio and this ETF since inception on July 29, 2021. Over the first months, the Wikifolio moved roughly in line with the index (which is not too bad). However, during November 2021 it suddenly started to underperform. I will give you some thoughts on the reasons for this in another review. But for the moment, let’s just look at the facts.

As of March 11, 2022, the Wikifolio is about 6.4 percentage points behind its benchmark. This is not what I aimed for and there are no excuses – the strategy simply sucked during the first months and in particular since mid of November 2021.

Technical Issues and Fees

However, one thing I want to highlight is the period from late December 2021 to mid of January 2022. As you can see, the line for the Wikifolio is flat. This has nothing to do with the stock selection, but is a technical issue. There was one stock for which Wikifolio didn’t receive a price from its data vendors. Wikifolio’s solution for this problem is somewhat unfortunate: they stop calculating the entire portfolio value and the Wikifolio has no price for that time. They also don’t backfill the gaps when the missing price becomes available again. The time series is therefore incomplete.

The other big issue are substantial fees associated with Wikifolio. Specifically, there is a fixed fee of 0.95% per year and a variable performance fee on all returns above the current High Water Mark.[1]This means that investors only pay performance fees when Wikifolios reach a new all-time-high. In my opinion, this is fair as portfolio managers don’t get paid for recovering prior losses. As far as I understand it, the portfolio manager can set this performance fee between 5% and 30%. Both of my Wikifolios charge the lowest possible performance fee of 5%. But compared to the total fee of 0.09% per year for the Lyxor ETF, the combination of 0.95% fixed and 5% performance fee is substantial and obviously reduces performance.[2]I would love to do it cheaper but at the moment I don’t see another sensible way to test the strategies in practice. Check this post for more details on that. Also note that even if other people invest in my Wikifolios, I only receive a fraction of the performance fee. Everything else goes to Wikifolio and its partners.

Indirect Trading Costs

Unfortunately, we are not even finished yet. There is another, somewhat more hidden layer of fees. While trading the stocks on Wikifolio, I occasionally observed bid-ask spreads that seemed a bit wide for my personal impression.[3]The bid-ask spread is the difference between the price at which a security can be bought (ask) and sold (bid). Traders, banks and market makers earn bid-ask-spreads as compensation for providing liquidity. The wider the spread, the more illiquid the security and vice versa. In addition to that, Wikifolio shows prices only in EUR. So I have no idea if they charge a fair exchange rate for converting the official USD stock prices to EUR. Again, I don’t want to be too negative. A bid-ask-spread and somewhat unfavorable exchange rates are pretty normal for most trading platforms and there is almost no chance to trade at the official exchange prices. Nevertheless, I must know the impact of these indirect costs to evaluate the true performance of my stock selection.[4]For those who are interested in more details: Wikifolio works with L&S Exchange to execute transactions. This is not an official exchange like the NYSE but an external market maker. In general, this is neither better nor worse. However, we should only trade on such platforms during the opening hours of the underlying exchange. Outside these opening hours, vendors like L&S can’t hedge themselves on the exchange. To compensate for that, they quote less favorable prices and wider spreads. As soon as I figured this out (fortunately quite early), I made sure to only trade when the US market was open.

To estimate the effect of indirect trading costs, I use intraday data from Financial Modeling Prep (FMP) as a benchmark.[5]This is of course not an audited data vendor like Bloomberg or Refinitiv but their data quality is very good and I randomly validated a few data points. Overall it should give a sensible picture of indirect trading costs relative to official exchange prices. For each transaction, I combine the price reported by Wikifolio with the intraday price and EURUSD exchange rate from FMP. For example: I bought 138 shares of Adobe on July 29, 2021 at 15:49 UTC. The transaction price on Wikifolio was 525.20 EUR. The intraday price at that minute was 622.82 USD and the EURUSD exchange rate was 1.1880. This translates into an “official” price of 524.26 EUR. So in this case, the share price was 0.94 EUR higher on Wikifolio. That’s a difference of about 0.18% or 18 basis points (BPS).[6]Calculation: (525.20 – 524.26) / 525.20 x 10,000 = 17.89 BPS This may sound small, but over a few hundred or even thousands trades such differences add up quickly.

I do this calculation for all my transactions to get an idea about the overall indirect trading costs. The table below shows conservative summary statistics. Why conservative? Because I always use the most unfavorable intraday price for the respective trade – “high” prices for “Buys” and “low” prices for “Sells”. The comparison is therefore something like “Wikifolio vs. worst possible price at the exchange”. I may overestimate indirect costs with these assumptions, but I am happy to do that. I think it’s better to have lower costs than expected instead of the other way around.

The table shows the mean, standard deviation, and several percentiles[7]Percentiles give you the location of certain values in the distribution. For example: the 25%-percentile of indirect costs across all trades (“Total”) is 10 basis points. This means that 25% of the trades exhibit indirect costs of less than 10 basis points. Vice versa, 75% of the trades were more expensive than 10 basis points. of indirect trading costs. An important observation are negative values at the lower end of the distribution. This means that for some trades, trading on Wikifolio was cheaper than on the exchange. This may be due to my conservative assumption or data inconsistencies, but in general, this is absolutely possible and gratifying. For most transactions, however, there are considerable indirect costs. On average, they amount to 42-43 basis points with almost no difference between “Buys” and “Sells”. This number is somewhat skewed by large positive numbers at the tails. The median is with 25-29 basis points lower than the mean but still indicates significant indirect costs.

How much are 30-40 basis points indirect costs per trade? Let’s go through a simple calculation to illustrate this. The initial value of the Wikifolio was 10 million EUR and there are about 140 equal weighted stocks in the portfolio. Each position is therefore worth about 71,500 EUR. With average indirect trading costs of 40 basis points, that’s 285 EUR per trade. So far, there have been 560 trades leading to total indirect costs of about 160,000 EUR. Relative to the initial portfolio value of 10 million EUR, this is a loss of 1.6%. In practice, it is even more because we could have invested these 160,000 EUR into the strategy which (hopefully) produces positive returns over time.

And that’s just half a year. If I continue to rebalance the Wikifolio in the same way, indirect trading costs will double to about 3% per year. With the 0.95% fixed fee of Wikifolio, that’s a loss of 4% per year just because of trading costs and fees. This is not meant to be an excuse, but consistently outperforming the S&P 500 Index by 4 percentage points per year is very difficult. Therefore, let’s have a look how the strategy performed against the benchmark before all those costs.

Performance w/o Fees and Indirect Costs

The following chart shows the cumulative performance of the Wikifolio without the different types of costs. The orange line is the official price from Wikifolio and, of course, exhibits the lowest performance of all. The gray line shows performance before the 0.95% fixed- and the 5% performance fee. The blue line is the hypothetical performance with closing prices and exchange rates from FMP. For both calculations, I assume that saved fees and costs are just held as cash and are not invested in additional shares.[8]When foregone returns are positive (negative), this assumption underestimates (overestimates) the true impact of fees. The blue line also shows the performance for the period without official prices from Wikifolio. Overall, the impact of fees and costs is roughly in line with the 2% I estimated above and occasionally even higher.

In the next chart, I put everything together: it shows the cumulative performance of the official Wikifolio (I call this “Net”), the hypothetical Wikifolio without any costs (which I call “Model”), and the S&P 500 benchmark. Surprisingly, the “Model” portfolio performed worse than the “Net” version over the first months. This is counter-intuitive and the only way I can explain this are data inconsistencies among Wikifolio and FMP.[9]I suspect that Wikifolio uses closing prices from the L&S Exchange to calculate the portfolio value. These prices include some sort of after-hours trading, so they may differ from the official closing prices. But I can’t prove it, so I don’t criticize any of them. I can never guarantee for it, but I checked the code behind this chart several times and couldn’t find an error on my side. So for the moment, I leave it as it is.

As shown in the table, my strategy was somewhat riskier than the S&P 500 by conventional measures. Annualized volatility and the maximum drawdown are both slightly larger for the “Model” Wikifolio. Given that the strategy did not produce better returns than the index, this is again a poor result. Note however, that because of the short history those numbers aren’t very stable yet. Therefore, I frequently update the chart and the table on the Portfolios page.

Now, a lot of talk about fees and costs but what is the point of all this? First of all, there is no excuse: the Wikifolio is designed to outperform the ETF on the S&P 500 and it clearly has not achieved this goal so far. However, from the cumulative underperformance of 6.4%-points, about 2.7%-points are due to fees and indirect costs. This obviously doesn’t help the investor who receives the net return. But it tells me that the stock selection is not the only problem. More efficient implementation is definitely something I need to work on. As we will see in the next part, this is even more important for the less liquid stocks in the “Report Analytics USA Small” Wikifolio.

Review – Report Analytics USA Small

Unfortunately, the “Small” Wikifolio has no official price since more than 3 months. This is because of the technical issue I explained above: Wikifolio can’t get a price for one stock and therefore stops the entire calculation. In this case, it is the DSP Group. The technology company was acquired and delisted its shares on December 2, 2021.[10]The DSP Group reported about this transaction in this report. For some reason, however, the delisting is not yet implemented on Wikifolio. I already approached them to solve this and I hope they will find a solution soon.

To not make this post too long (and repetitively boring), I jump directly to the chart that shows the impact of fees and indirect costs. The “Net” performance (orange line) currently stands at 2.33% since inception in late August 2022. However, this number is more than three months old. So it doesn’t make too much sense to look at the official price. The “Model” performance (blue line), in contrast, stands at 4.41% since inception. Overall, returns are somewhat more volatile than for the “Large” Wikifolio. This is no surprise as this one intentionally tilts towards small caps.

Indirect trading costs are significantly higher than for the “Large” Wikifolio. This is again no surprise. Small- and mid-caps are, at least in most cases, less liquid than large caps. The magnitude, however, is remarkable. On average, indirect trading costs are almost twice as high as for the “Large” Wikifolio (77 vs. 42 BPS). For the median, the difference is even larger (68 vs. 28 BPS).

Remember my previous calculation? For the “Large” Wikifolio, I estimated a performance loss of about 3% per year due to indirect trading costs. Based on the table above, this easily amounts to 4-5% for the “Small” Wikifolio (assuming that I continue to trade as costly as I did so far). Adding the fixed- and performance fee of Wikifolio, and the strategy quickly becomes unviable. Improving implementation and reducing costs is therefore even more important for the “Small” Wikifolio.

Especially because the stock selection actually wasn’t too bad for small- and mid-caps. The following chart shows again the “Net” and “Model” Wikifolio versus the S&P 500 ETF. It has been volatile but in particular over the last month, the “Model” Wikifolio outperformed the index considerably. As of March 11, 2022 it is now about 1.1 percentage points ahead of the index since inception. I will of course also update this chart on the Portfolios page.

Also note the risk measures in the table above. Although the “Model” Wikifolio was more volatile than the index, its ratio of risk and return was slightly more attractive (0.53 vs. 0.49). The lower maximum drawdown is also a nice result. However, I don’t want to oversell these results.

First, these are just slightly more than 6 months and no official prices. This is a very short period of time for a low-frequency strategy like this. Second, some part of the outperformance of the “Small” Wikifolio is certainly due to the somewhat imprecise benchmark. The S&P 500 is a practical passive benchmark for the US stock market, but not necessarily for small- and mid-caps. I discuss this issue in more detail on the Portfolios page. Finally, if you compare the “Large” and “Small” Wikifolio please note the different inception dates. For a proper comparison, we would have to look at the same time period.

Summary and Outlook

This post contains a lot of unsexy calculations and is fairly technical. But (in my opinion) there are some very interesting results. Not just for my particular strategy but for everyone who is active on Wikifolio.

First. Overall and especially after costs, my two Wikifolios weren’t a good alternative to a standard ETF on the S&P 500 index. Of course, I don’t like that but it’s the brutal reality that every portfolio manager must face. It is a well-documented empirical fact that most active strategies underperform their passive benchmark and I can’t escape from that.[11]For example, Morningstar regularly shows that most active managers underperform and that it is very hard to identify skilled managers ex-ante. This is in line with many academic studies. Standard references in this area are Carhart (1997) and Fama and French (2010). To my defense, however, I stressed several times that the two Wikifolios are just a real-world test of my master thesis and I never marketed them as investments.

Second. I still believe that Wikifolio is a great platform to test strategies like mine, but it is not perfect. There are annoying technical issues, pretty high fees, and significant indirect trading costs. Due to the technical issues, the Wikifolios are sometimes not priced for weeks or even months. For my use-case, this is just annoying and I can easily circumvent it with my “Model” portfolios. But for real-money investments in Wikifolios, this represents a substantial liquidity risk.

The fixed- and performance-fee are in line with the mutual fund industry and we can debate if this is too high or not. I won’t go into details here, but in any case, Wikifolio fees are way higher than those of ETFs. Finally, and this is probably the biggest issue for systematic strategies like mine, indirect trading costs hamper performance considerably. Depending on the liquidity of the stock, bid-ask-spreads and/or unfavorable FX rates amount to 40-80 basis points per transaction on average.

Third. When I worked on my master thesis, I was super excited that I found an outperforming strategy. At that time, however, I did not know how difficult it is to implement such a strategy in practice. It sounds so easy. You have your 150 stocks, you buy them equally weighted into your portfolio and rebalance each month. But there are many issues with that. Trading on Wikifolio is very time-intense manual work[12]Unfortunately, there is no option to upload a list of trades. You must input every single transaction manually., too much trading and rebalancing produces a lot of costs, and keeping faith when the strategy is not working is also easier said than done. Despite the poor results so far, it is nice to gain these experiences hands-on. Its a very different and, in my opinion, more enriching experience than just doing the academic backtest.

So what am I going to do with these insights? First and foremost, I will work on reducing indirect trading costs. So far, my trading has been very unsophisticated. I just rebalanced the Wikifolios on a single day with market orders. To improve this, I will experiment with limit orders to get better transaction prices. Maybe I will also stretch the rebalancing over a longer time period. For example, 3 days or even a week.

In general, I resolve to take care of the Wikifolios more actively. Since inception, I pursued a very disciplined buy-and-hold strategy and rebalanced the portfolio only every three months. I did this on purpose to stick as close to the backtest as possible. The only adjustment I have made is to rebalance quarterly instead of monthly. And that is because I recognized the indirect trading costs. Of course, I want to keep this discipline and will never overrule my model. But I recognized some issues where faster reactions would have been beneficial. For example, the immediate reinvestment of dividends, delistings, or merger announcements.

Another thing is to combine the results of this post with the backtest. The estimated transaction costs should allow me to create a historical simulation of the strategy after all costs. This is going to be interesting because it will tell me if it is even worth to continue the strategy on Wikifolio. Regardless of that, I am also working on a review that focuses more on the stock selection rather than implementation (things like sector allocations, styles, stock examples, etc.). So there is a lot more to come.

This content is for educational and informational purposes only and no substitute for professional or financial advice. The use of any information on this website is solely on your own risk and I do not take responsibility or liability for any damages that may occur. The views expressed on this website are solely my own and do not necessarily reflect the views of any organisation I am associated with. Income- or benefit-generating links are marked with a star (*). Please also read the Disclaimer.