TAA in the real world: theory versus slippage

Today I want to talk about my friend Jose. About a year and half ago, after months of Q & A between us, Jose decided to implement Antonacci’s GEM portfolio in his IRA (see this post for a detailed description). This post describes his first 6 months in GEM, November 2015 through April 2016. I think there are some good lessons here for the implementation of any TAA portfolio or any active strategy for that matter. I covered some TAA implementation tips and tools before. See here and here. Let’s jump right in.

Starting out in any active strategy can be quite a stressful experience. Here I’ll take a look at maybe the simplest yet one of the most effective TAA strategies, Antonacci’s GEM strategy. Jose started out with GEM at the beginning of November 2015. Below are the GEM allocations over the first 6 months, along with what trades were required in the portfolio. You can find all the historical allocations for GEM here.

  • Nov 2015     SPY (buy SPY)
  • Dec 2015     SPY
  • Jan 2016     SPY
  • Feb 2016     AGG (sell SPY, buy AGG)
  • Mar 2016    AGG
  • Apr  2016    SPY  (sell AGG, buy SPY)

Unfortunately for Jose, this was a tough time to start out in GEM. Total return for GEM over these first 6 months was -4.27%. But in the real world you need to pay fees, bid/ask spreads and actually buy the ETFs at prices that are probably different than what the model uses. Below I calculated what Jose’s real portfolio return was during this period assuming he bought the ETFs the day following the model update either at the open or at the close of the trading day.

Quite a bit different for even just 2 changes in the model over these 6 months. It’s easy to sit back and tell Jose not to worry about it that all this will be a wash over the long run, that sometimes it will go in his favor, but it’s a lot different to live through it especially when you’re just starting out. Also, most investors pay more attention to trading fees and bid/ask spreads than slippage which is way worse. In the example, above trading fees and bid/ask spreads would have accounted for about 0.05% of the difference between returns for a $100K portfolio. Slippage kills. The worst part of slippage in these types of models is not the impact to returns, it is the tracking error (the difference between the model’s returns and your returns) and the impact of that on investor confidence. And the more complex the model, the more active it is, the worse the impact it has. After these first 6 months, Jose’s confidence in GEM was already shaken.

Fortunately, there is a solution to the slippage problem in TAA portfolios. It requires some attention to detail and a bit of trading knowledge but is pretty easy to execute these days. The solution is to execute the model’s trades as close to the theoretical price as possible. The trickiest part is to anticipate what the model’s signals are before the close of the trading day when the model updates. You can do this yourself. I used to use this Google sheet to do the calculations for me (see the ‘re-work of returns at current market prices’ at the bottom of the sheet). Or you can have it done for you. AllocateSmartly does this for you  and the models are updated in real-time. It will even send you an email about mid-day warning you of any changes to your portfolio. To me this is one of the biggest pluses of the tool and worth the investment in the tool by itself.

Once you know what you need to buy and sell then you log in to your brokerage platform and enter the orders. If your brokerage platform supports it, I recommend using MOC (market on close orders) for all trades. Since I have been using MOC orders for my TAA portfolios I have not had any trade execute at a price different from the closing price of the ETF. That means zero slippage. If you’re brokerage does not support MOC orders there are other more cumbersome ways. The brokerage at least should support conditional orders where you can set market orders to be automatically submitted to the market at an exact time, for example 15 seconds before the close. Of course, you can manually enter market orders before the close as well but it’s not the best way and you will incur some slippage that way. Basically, whatever method you choose with  today’s tools it’s possible to get very close to zero tracking error in TAA portfolios, zero fees, and zero slippage.

Before I close I want to compare Jose’s experience with what it would have been 6 months later. If Jose had entered the GEM strategy in May 2016 he would have bought SPY and still have been holding it today. Zero trades and a nice big gain. He would have suffered maybe similar slippage and tracking error but in my experience these things often go unnoticed when you’re sitting on decent gains. What a difference 6 months can make.

In summary, the implementation of TAA strategies matters. It’s an overlooked but I think critical part of any kind of active investing especially for it’s effect on investor psychology. Today’s tools make it easier than ever for investors to closely match theoretical TAA strategy results.

Full Disclaimer - Nothing on this site should ever be considered advice, research or the invitation to buy or sell securities. These are my personal opinions only.

8 thoughts on “TAA in the real world: theory versus slippage

  1. I spent my first 3 years using TAA from a service which sent rebalance notices out after the close. Trying to execute a rebalance based on closing prices during the open is even harder on the psyche than it is on the slippage to the wallet.

    When I started building my own TAA models, I determined to model and trade on the close. I have now been using Market On Close orders at Interactive Brokers for 2+ years with zero slippage and no damage to the psyche. There are times when an ETF allocation changes by a handful of shares and I just let the ETF position ride. My annual commission costs run 0.03%+- of account size.

    Edited to remove promotional content.

  2. Paul,
    I’ve always enjoy these types of “rubber meets the road” articles, thanks for publishing. I have several momentum strategies I’ve implemented for myself and it can be difficult to invest a large lump sum into a new strategy all at once. The adjustment I’ve made that helps me is to dollar cost average into a full position over several timing periods. I’ve read several articles that say lump sum is actually proven to be better, but psychologically I’m bothered by it so I made a compromise that works for me. Encourage Jose to stick with it, he won’t be sorry in the long run, especially the nasty bear comes out of hibernation again.

  3. Paul,
    I use mutual funds instead of ETFs in my TAA portfolios. My broker doesn’t offer MOC orders and when I was using ETFs I had to trade them during my lunch break which was not always a good time of the day to execute market orders. So I switched to equivalent mutual funds. Does my mutual fund portfolio also suffer from slippage (or other similar performance drag) that is not obvious to me? Thanks for your articles!

    1. Kirk, mutual funds are a good way to go as long as they are no-load, low-fee, and your broker doesn’t charge exorbitant fees to trade them. With mutual funds you always get the closing price. You just need to make sure you enter the trades the same day as the model signals.

      Paul

  4. Paul,

    I should also mention that the mutual funds I use are no load and have expense ratios that are the same or very close to the equivalent ETFs. I don’t get in trouble with high frequency trading as long as I don’t exchange out of a mutual fund withing the first 30 days of holding it. Thanks again.

  5. Additionally to the slippage issue you describe, it is also difficult psychologically to adopt a new TAA strategy and implement entirely for an account on one day. In my case, I am considering a combined TAA strategy modeled from Allocate Smartly, which means selling most of my positions and entering new positions. I am now considering whether to begin this at the end of this month or right now. Any advice?

    1. It’s a tough process to go through no matter what. I tend to make those big changes over time even though theoretically it’s better to do it all at once. For example, I went from a 25% quant allocation to 45% over the last 2 years.

      Paul

Comments are closed.