Foolish Workshop
By
Cost of Screens,
Part 1:
External Cost Factors
EL SEGUNDO, CA (Oct. 12, 1999) -- One of the things many mechanical investing fans forget to consider when choosing a strategy are the costs of the strategy. With commissions as low as $5 a trade, it's easy to forget that costs can be significant, especially for small investors using strategies that call for frequent trading.
When looking at screen returns such as the Workshop Returns we quote in this area or historical backtested returns here or anywhere else, an investor needs to consider what the return would be after accounting for the costs and taxes that would affect him or her personally. In this article, I will go over the techniques I use to get the after-cost returns for screens I am considering. In my next article, I will go through some specific examples.
The return an individual investor would have received from a strategy follows the following formula:
Actual Return = (Return - Costs) * (1 - Tax Rate)
where Tax Rate = Marginal Federal Tax Rate + Marginal State/Local Tax Rate
For a tax-free account such as an IRA, the tax rate is zero and thus you only worry about returns and costs.
The costs that everyone pays but that are not included in published returns are round-trip trading costs and the spread.
The round-trip trading cost is the amount it costs you to buy and sell the stock. For example if your broker's commission is $10 per trade, then the round-trip trading cost is $20 -- $10 when you buy the stock and $10 when you sell it. The commission actually understates the cost slightly because with each sale there is a small SEC fee but since the fee is very small we will ignore it here.
The more frequently ignored factor is the spread. If you go to an online broker to place an order, you will usually see three prices listed. They are:
a) Last Trade Price
b) Ask Price
c) Bid Price
If you put in a market order to buy a stock, your shares will be bought at the current "Ask" price. The Ask price is literally the "asking" price; that is, the lowest specified price currently on the books for which someone is willing to sell that particular stock. The Bid and Ask prices are established by brokers who "make a market" for that specific stock based on supply and demand and the limit orders that have been entered into the system. It can change moment to moment, of course, as buy orders come in. When there are enough buy orders to fill all of the Ask orders, the price will move up to the next lowest specified price which becomes the new Ask price.
If you put in a market order to sell, the opposite happens. Your stock will be sold at the current Bid price, the highest price for which someone has said they will buy the stock. It's just like an auction.
Now, suppose you could trade commission-free. If you placed market orders to buy shares and sell them immediately, you would lose money even if the "price" of the stock didn't change. That's because the Ask price you bought at will always be a bit higher than the Bid price that you can sell for. Think about it -- if there were a bid on the books as high as the lowest asking price, the trade would have already been executed.
When comparing actual returns to reported returns, you need to consider the cost of the spread because the backtested results do not buy at the Ask and sell at the Bid like you would do. Instead we use whatever the closing price was on the day in question. (That's the only option for most backtested strategies and the easiest way to do current returns.)
Here is a quick quiz. Which price is the closing price:
a) The Bid price
b) The Ask price
c) Neither the Bid price not the Ask price
d) Any of the above.
The correct answer is "d." The closing price can be either the Bid price or the Ask price as well as any other price, even one higher than the Ask price or lower than the Bid price, or anywhere in between. The closing price is just the price of the last trade.
Let's look at two examples of the spread from Monday, October 4. The first stock we will look at is a small stock, Numac Energy <% if gsSubBrand = "aolsnapshot" then Response.Write("(AMEX: NMC)") else Response.Write("(AMEX: NMC)") end if %>. Numac as of 10/1/99 was the #1 stock on the PEG13 screen, which has become a favorite monthly trading strategy on the message board. Numac's closing prices were as follows:
Last Trade: 3 15/16
Ask: 4
Bid: 3 13/16
The other example I'll use is Intel <% if gsSubBrand = "aolsnapshot" then Response.Write("(Nasdaq: INTC)") else Response.Write("(Nasdaq: INTC)") end if %> which was the #1 stock on the Spark Screen:
Last Trade: 77
Ask: 77
Bid: 76 15/16
So if October 4 was the day the PEG13 screen was going into the record books, Numac would be priced at 3 15/16 ($3.9375). However, if you had been the one that made that last purchase, you would have paid $4.00 a share. If you had been selling, you would have received only $3.8125 a share.
With Intel you can see the last trade and the Ask price were the same, but if you were selling, you would have gotten 1/16 ($0.0625) less. That's the spread
Also, notice the difference between the spread for a large, frequently traded stock like Intel and a small stock like Numac. For Intel the spread is $0.0625 (or 0.08%, a pittance). However, for Numac the difference between the spread is a whopping 4.7% (3/16 = $0.1875 and $0.1875 / $4.00 = 4.7%). What this means is that if you were to buy Numac today and sell it a month later, even if the Bid and Ask prices were exactly the same, you would have lost 4.7% without even considering the commissions.
So you can see that, depending on the stock, the spread can be fairly insignificant or an extremely big part of the cost.
How should one account for this cost? The problem is that the closing price we use to calculate returns can be completely unrelated to the spread. The law of averages says that the closing price will reflect the spread about half the time. In other words, the close will be the Ask when our model is buying or the Bid when it is selling. But, only about half the time will they more or less match up like that. So it's reasonable to assume that only about half of the cost of the spread will be reflected in our published returns.
What value should be used to correct for this? It depends on the type of stocks the screen invests in. The higher the trading volume of the stock, the lower the spread. And the higher the price of a stock, the lower the percentage that is lost to spread. So, for example, a screen like Keystone that selects large-cap stocks might only lose about 0.2% or less to the spread. The Relative Strength screens select from a more mixed universe of stocks so I'd go with a factor of around 0.25%. On the other hand, a screen like PEG can include many small stocks, or no small stocks at all. (Numac is an extreme case.) A spread factor of 0.5% is probably a realistic average number to use for this type of screen.
Half a percent? Really? Isn't this much ado about nothing? If you're trading annually, probably. BUT if you are trading more frequently, especially if you are trading monthly with a fairly small portfolio, those spreads add up and compound. By the end of the year those costs can have a significant affect on your return. We will look at that on Thursday.
Until next time, Fool On!