Closed End Funds are now available for screening, backtestings, ranking, etc.
CEFs are in the ETF world, but you must choose either the Universe “All CEFs” or “All ETFs and CEFs”
We also added two new factors that are only available for CEFs in RATIOS & STATS->VALUATION->Net Asset Value
NAV - The market value of the fund’s assets minus liabilities divided by the basic shares outstanding. This is a MONTHLY item. The value recorded is that of the last Friday’s close. This item applies to closed-end funds only.
NAV%Discount - The discount as % from NAV (monthly data) compared to latest close for closed-end funds. Funds trading below their NAV will show +ve discounts
In the coming days we will try to launch some sampleports/screens.
This is great! Thanks a lot! Great job. There’s enough here right now to build some systems…and update what I do manually. So…thanks!
I already:
A) Updated my current CEF holdings to include a slightly more evolved technical system for asset timing. First trial cut my peak DD by more than 30%. Lowered STDEV. And increased AR%.
B) And…Just finished my first ‘ranking system’ test. 20% top bucket (20 buckets) on all CEF’s…on very first try. And first sim showed double digit returns. So…very, very cool.
Any chance we can get SMA and EMA of premium / discount? Need the relative values on these for some all CEF comparisons.
So…I would really like to use these. And market cap? Or ‘leverage rate.’
We now have MktCap for CEFs. Note that NAV is a monthly value, The Discount is calculated using prices and the closest monthly NAV in the past. Knowing this is important to designing proper functions
One of the goals is to find CEFs discounted more relative to their own historical discounts. We could add some factors like Discount4WkAgo, 8wkAgo, etc, to get started.
This is also a good source of ideas for functions:
The results look very promising, but I have a concern. I ran a screen for 1999-2013 with universe = All CEFs and the only rule being ETFAssetClass=EQUITY. The results are that the equity CEFs return 250% while the market (Russell 3000 in this case) returns about 60%. I had a 4 week rebalance, but longer rebalances up to a year gave about the same results. Do CEFs generally beat the market? If not, then my concern is that there could be some data issues or survivor bias.
Also, are the expenses accounted for? Some CEFs have pretty steep expense ratios.
What about a function similar to PErelative, but using NAV discount? Then you could find CEF’s that were relatively historically cheap - and rank them using this.
Not sure about expense ratios. Looks like they average 1.5%. I will ask. We’re adding other NAV discount functions soon to find CEF discounted relative to their norm.
I previously read an article by morningstar, which I cannot find now, stating that the average return of closed end funds over the last 10 years exceed those of open ended mutual funds. In addition on the CEFA.com page they have the 5 year average close ended fund performance. In the equity section this averages between 17 and 29% year depending on the category. This outperformance may be due to leverage, manager skill, and/or a lack of untimely redemptions. This data suggests they will outperform open ended mutual funds however it does not exclude an added component of survivorship bias to your backtested returns as you suggested
Could you guys add something in terms of ‘asset cap weight’ constraints (like Secweight and CapWeight), where I can define the limits for asset class weights within a single sim? I had this as a feature request for ETF’s in the past.
US EquityWt<=30,
Bond Weight<=30.
Etc.
So, I can use one set of rules to buy a ‘balanced’ portfolio. Being able to do this in one sim. is helpful for relative value type systems.
Is the time series on institutional ownership data on CEF’s hard to extract? This is also one of things I have looked at in the past for choosing funds.
Yes, we need to add taxonomy weight constraints. I need to check about ownership data.
Seems like there are opportunities in CEFs. They are a black box and investors rely on a few data points. It’s not clear yet how many data points we will be able to provide as we haven’t looked at all the items we get from Compustat. We’ve simply treated them like ETFs, classified them, and provided a couple of NAV functions, and Yield. But more is needed.
However, even if we don’t go much deeper than NAV, there are things P123 can do that no other site can. For example adding historical functions to compare today’s NAV with the NAV 3,6,12 months ago. Also comparing Discount in a similar fashion. With some new functions we should be able to screen for:
CEFs that have maintened a steady NAV, but whose discount has increased
CEFs that have grown their NAVs as well as their discount
Find CEFs that have grown their NAV more than the SP500 return
Create NAV growth ratios
etc.
In other words, using our ability to look in the past we can create unique, backtestable strategies, that no other site can do even if we have less data points. Management expense are taken from the NAV, but if the NAV is growing well, does it matter if the manager gets 1% or 2% fee?
Marco,
I did a quick test to see what the expense ratio was on the top 5% of equity CEFs ranked on Discount%. The only filter was AvgDailyTot>250k. Then I looked up the 12 CEFs on CEFConnect.com and the average exp ratio was 1.6%. I thought it was going to be at least 2.5% since (I thought) the CEFs with big discounts usually were undesirable for some reason, one of which would be a high exp ratio. Some have expense ratios > 4%. Its good to see that was not the case in my experiment.
If expense ratio is available in the data, it would be nice to have as a filter and for ranking.
[quote]
The results look very promising, but I have a concern. I ran a screen for 1999-2013 with universe = All CEFs and the only rule being ETFAssetClass=EQUITY. The results are that the equity CEFs return 250% while the market (Russell 3000 in this case) returns about 60%. I had a 4 week rebalance, but longer rebalances up to a year gave about the same results. Do CEFs generally beat the market? If not, then my concern is that there could be some data issues or survivor bias.
Also, are the expenses accounted for? Some CEFs have pretty steep expense ratios.
[/quote]Dan I ran a stock screen over the entire period using all stocks equally weighted in the Prussell 3000 index. They returned 267%. In other words your equal weight CEF screen performed about the same as the equal weight benchmark. That tells me that there is nothing magical about CEFs per se. It was the equal weighting that gave the alpha.