pros and cons of a coat-tailing algorithmic process?

[scorpioncapital]

Say you do a database study of global funds/insiders and what they hold each month, then choose 5-5-5-5 from each size category. Totally algorithmic. What are the pitfalls or pros of this method? So far this year such a strategy has outperformed sp500 by a huge margin. Furthermore, the study seems to show large cap and very tiny companies have had the best performance (the middle of the road the worst). Any thoughts? My initial reaction is that it is somewhat backward looking, or disguised beta as alpha. But I can't really criticize the results if they work. Thoughts? 

Edited October 16, 2023 by scorpioncapital
grammar

[ValueArb]

Not sure I understand, are you saying buy top 5 most popular in each size category among all global funds and insiders? And where do you get the data?

 

My only comment so far is to say one year isn’t enough to give a lot of confidence. 

[scorpioncapital]

So you take a database of all funds or insiders (includes pensions, individuals as insiders, etc..) from a public database (reporting requirements differ by country but there is some minimum to report).

 

You find the funds with the top returns above sp500 over 5 years (you can exclude outliers if you want), with some criteria. E.g. 3 out of 5 outperformance years, over 10 securities held)

 

Say this returns 300 funds. You then dig into their holdings and find which securities had the top mentions across all of them and categorize by market cap size.

 

Pick top 5 from each size class and buy them (say 50 stocks if you split it into 5 size categories). Rebalance monthly or quarterly.

 

For me the backward looking aspect is the big issue. But I can't say why or why not it wouldn't work. Since you are always coat-tailing on the ones that come to the top. After all, all index funds operate this way. Sp500 removes those who fall out, and adds those which have risen to the top. That's why Buffett says it is so hard to beat sp500 because it is never too late to add winners and ditch losers.

 

Edited October 17, 2023 by scorpioncapital

[ValueArb]

I think the problem is that you're getting the worst of it due to market price flows. Lets say you find the most popular position X in midcaps based on the latest months database. So you add it to your portfolio. Then eventually one month you find its no longer a top 5 midcap in the database, so you sell it. Problem is you bought it only after a bunch of funds bought it, and only sold it after a bunch of funds sold it. So your purchase basis is likely to average out significantly higher than those you are coat-tailing, and your sales prices are likely to average out significantly lower than theirs. The performance margin for average market beating funds vs. the indexes are likely to be narrow enough that that your worse purchase/sale prices are more than enough to drop you below index returns.

 

And this is assuming you can get timely monthly data. In the US trades are reported on 13Fs anywhere from 45 to 125 days after they happened in previous quarter. So data is likely to be very stale. 

[scorpioncapital]

Ok, I didn't think about that delay if it has huge consequences. Perhaps you can link it with the funds that outperform to see who has the least turnover and assume there is more than average of 90 days continuity between reports? One fund was just a collection of indexes. I guess for 5 years they just shifted around the index distribution and name, currency used and perhaps used leverage (unknown). But if you assume leverage has a maximum you could back engineer the non-margined return. Not sure if funds can get more than 2-3:1 leverage or not. 

 

[ValueArb]

I don't do quantitative investing, so I have no idea what variables you can tweak and what filtering you can apply to come up with a fund following algorithm that has a long history of outperformance. I just see too many complexities and friction from my vantage point. This is why I'm also skeptical on even reading other investors 13Fs, not only the long informational delays but the differing goals/objectives between my portfolios and theirs. For example, I have no interest in what Buffett is buying, he's forced to buy megacaps and Ted/Todd are forced to buy large caps, I can buy anything and feel there is likely far more higher return opportunities in small/micro caps where stocks aren't liquid enough for the vast majority of professional funds.

 

But that said, I would be interested in access to positions in funds that focus on small caps and microcaps, but never for any automated formula. I'd only use them as suggestions to do my own investigation of, just because some of the best funds all own something doesn't mean they are right about it or did their research properly. If its a mistake I want it to be my mistake and learn from it. If I can't understand why they bought it, let them make the profits,  I'll skip it until I can develop a thesis on it that justifies buying it. I get the allure of automation coming up with a "Magic Formula" but I'm skeptical about finding any. Even Joel Greenblatts basically fell apart in the US market as soon as he publicized it. Probably too many people trading the same ideas, or it could have been a lucky pattern in the data that persisted for a long while but eventually reverted to normal. 

[sleepydragon]

This seems to be you are just buying the momentum factor. So you are buying the stock that has gone up the most for the last 6-12 months. first, you picked the top performing funds, which by definition must have been buying the stocks that had gone up. Then you picked their biggest holdings, which usually be stocks has had gone up a lot. You then equal weighted them, which means you will be buying dips of these hot stocks until the next rebalance 

 

buying momentum soemtimes works (ie in a bull market ) but once a while it gaps down huge.